#!/usr/bin/env texlua if not modules then modules = { } end modules ['mtxrun'] = { version = 1.001, comment = "runner, lua replacement for texmfstart.rb", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- one can make a stub: -- -- #!/bin/sh -- env LUATEXDIR=/....../texmf/scripts/context/lua luatex --luaonly mtxrun.lua "$@" -- filename : mtxrun.lua -- comment : companion to context.tex -- author : Hans Hagen, PRAGMA-ADE, Hasselt NL -- copyright: PRAGMA ADE / ConTeXt Development Team -- license : see context related readme files -- This script is based on texmfstart.rb but does not use kpsewhich to -- locate files. Although kpse is a library it never came to opening up -- its interface to other programs (esp scripting languages) and so we -- do it ourselves. The lua variant evolved out of an experimental ruby -- one. Interesting is that using a scripting language instead of c does -- not have a speed penalty. Actually the lua variant is more efficient, -- especially when multiple calls to kpsewhich are involved. The lua -- library also gives way more control. -- to be done / considered -- -- support for --exec or make it default -- support for jar files (or maybe not, never used, too messy) -- support for $RUBYINPUTS cum suis (if still needed) -- remember for subruns: _CTX_K_V_#{original}_ -- remember for subruns: _CTX_K_S_#{original}_ -- remember for subruns: TEXMFSTART.#{original} [tex.rb texmfstart.rb] -- begin library merge do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-string'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local sub, gsub, find, match, gmatch, format, char, byte, rep, lower = string.sub, string.gsub, string.find, string.match, string.gmatch, string.format, string.char, string.byte, string.rep, string.lower local lpegmatch = lpeg.match -- some functions may disappear as they are not used anywhere if not string.split then -- this will be overloaded by a faster lpeg variant function string:split(pattern) if #self > 0 then local t = { } for s in gmatch(self..pattern,"(.-)"..pattern) do t[#t+1] = s end return t else return { } end end end local chr_to_esc = { ["%"] = "%%", ["."] = "%.", ["+"] = "%+", ["-"] = "%-", ["*"] = "%*", ["^"] = "%^", ["$"] = "%$", ["["] = "%[", ["]"] = "%]", ["("] = "%(", [")"] = "%)", ["{"] = "%{", ["}"] = "%}" } string.chr_to_esc = chr_to_esc function string:esc() -- variant 2 return (gsub(self,"(.)",chr_to_esc)) end function string:unquote() return (gsub(self,"^([\"\'])(.*)%1$","%2")) end function string:quote() -- we could use format("%q") return format("%q",self) end function string:count(pattern) -- variant 3 local n = 0 for _ in gmatch(self,pattern) do n = n + 1 end return n end function string:limit(n,sentinel) if #self > n then sentinel = sentinel or " ..." return sub(self,1,(n-#sentinel)) .. sentinel else return self end end do -- roberto's variant: local space = lpeg.S(" \t\v\n") local nospace = 1 - space local stripper = space^0 * lpeg.C((space^0 * nospace^1)^0) function string.strip(str) return lpegmatch(stripper,str) or "" end end function string:is_empty() return not find(self,"%S") end function string:enhance(pattern,action) local ok, n = true, 0 while ok do ok = false self = gsub(self,pattern, function(...) ok, n = true, n + 1 return action(...) end) end return self, n end local chr_to_hex, hex_to_chr = { }, { } for i=0,255 do local c, h = char(i), format("%02X",i) chr_to_hex[c], hex_to_chr[h] = h, c end function string:to_hex() return (gsub(self or "","(.)",chr_to_hex)) end function string:from_hex() return (gsub(self or "","(..)",hex_to_chr)) end if not string.characters then local function nextchar(str, index) index = index + 1 return (index <= #str) and index or nil, sub(str,index,index) end function string:characters() return nextchar, self, 0 end local function nextbyte(str, index) index = index + 1 return (index <= #str) and index or nil, byte(sub(str,index,index)) end function string:bytes() return nextbyte, self, 0 end end -- we can use format for this (neg n) function string:rpadd(n,chr) local m = n-#self if m > 0 then return self .. rep(chr or " ",m) else return self end end function string:lpadd(n,chr) local m = n-#self if m > 0 then return rep(chr or " ",m) .. self else return self end end string.padd = string.rpadd function is_number(str) -- tonumber return find(str,"^[%-%+]?[%d]-%.?[%d+]$") == 1 end function string:split_settings() -- no {} handling, see l-aux for lpeg variant if find(self,"=") then local t = { } for k,v in gmatch(self,"(%a+)=([^%,]*)") do t[k] = v end return t else return nil end end local patterns_escapes = { ["-"] = "%-", ["."] = "%.", ["+"] = "%+", ["*"] = "%*", ["%"] = "%%", ["("] = "%)", [")"] = "%)", ["["] = "%[", ["]"] = "%]", } function string:pattesc() return (gsub(self,".",patterns_escapes)) end local simple_escapes = { ["-"] = "%-", ["."] = "%.", ["?"] = ".", ["*"] = ".*", } function string:simpleesc() return (gsub(self,".",simple_escapes)) end function string:tohash() local t = { } for s in gmatch(self,"([^, ]+)") do -- lpeg t[s] = true end return t end local pattern = lpeg.Ct(lpeg.C(1)^0) function string:totable() return lpegmatch(pattern,self) end function string.tabtospace(str,tab) -- we don't handle embedded newlines while true do local s = find(str,"\t") if s then if not tab then tab = 7 end -- only when found local d = tab-(s-1) % tab if d > 0 then str = gsub(str,"\t",rep(" ",d),1) else str = gsub(str,"\t","",1) end else break end end return str end function string:compactlong() -- strips newlines and leading spaces self = gsub(self,"[\n\r]+ *","") self = gsub(self,"^ *","") return self end function string:striplong() -- strips newlines and leading spaces self = gsub(self,"^%s*","") self = gsub(self,"[\n\r]+ *","\n") return self end function string:topattern(lowercase,strict) if lowercase then self = lower(self) end self = gsub(self,".",simple_escapes) if self == "" then self = ".*" elseif strict then self = "^" .. self .. "$" end return self end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-lpeg'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local lpeg = require("lpeg") lpeg.patterns = lpeg.patterns or { } -- so that we can share local patterns = lpeg.patterns local P, R, S, Ct, C, Cs, Cc, V = lpeg.P, lpeg.R, lpeg.S, lpeg.Ct, lpeg.C, lpeg.Cs, lpeg.Cc, lpeg.V local match = lpeg.match local digit, sign = R('09'), S('+-') local cr, lf, crlf = P("\r"), P("\n"), P("\r\n") local utf8byte = R("\128\191") patterns.utf8byte = utf8byte patterns.utf8one = R("\000\127") patterns.utf8two = R("\194\223") * utf8byte patterns.utf8three = R("\224\239") * utf8byte * utf8byte patterns.utf8four = R("\240\244") * utf8byte * utf8byte * utf8byte patterns.digit = digit patterns.sign = sign patterns.cardinal = sign^0 * digit^1 patterns.integer = sign^0 * digit^1 patterns.float = sign^0 * digit^0 * P('.') * digit^1 patterns.number = patterns.float + patterns.integer patterns.oct = P("0") * R("07")^1 patterns.octal = patterns.oct patterns.HEX = P("0x") * R("09","AF")^1 patterns.hex = P("0x") * R("09","af")^1 patterns.hexadecimal = P("0x") * R("09","AF","af")^1 patterns.lowercase = R("az") patterns.uppercase = R("AZ") patterns.letter = patterns.lowercase + patterns.uppercase patterns.space = P(" ") patterns.tab = P("\t") patterns.eol = S("\n\r") patterns.spacer = S(" \t\f\v") -- + string.char(0xc2, 0xa0) if we want utf (cf mail roberto) patterns.newline = crlf + cr + lf patterns.nonspace = 1 - patterns.space patterns.nonspacer = 1 - patterns.spacer patterns.whitespace = patterns.eol + patterns.spacer patterns.nonwhitespace = 1 - patterns.whitespace patterns.utf8 = patterns.utf8one + patterns.utf8two + patterns.utf8three + patterns.utf8four patterns.utfbom = P('\000\000\254\255') + P('\255\254\000\000') + P('\255\254') + P('\254\255') + P('\239\187\191') patterns.validutf8 = patterns.utf8^0 * P(-1) * Cc(true) + Cc(false) patterns.comma = P(",") patterns.commaspacer = P(",") * patterns.spacer^0 patterns.period = P(".") patterns.undouble = P('"')/"" * (1-P('"'))^0 * P('"')/"" patterns.unsingle = P("'")/"" * (1-P("'"))^0 * P("'")/"" patterns.unspacer = ((patterns.spacer^1)/"")^0 function lpeg.anywhere(pattern) --slightly adapted from website return P { P(pattern) + 1 * V(1) } -- why so complex? end function lpeg.splitter(pattern, action) return (((1-P(pattern))^1)/action+1)^0 end local spacing = patterns.spacer^0 * patterns.newline -- sort of strip local empty = spacing * Cc("") local nonempty = Cs((1-spacing)^1) * spacing^-1 local content = (empty + nonempty)^1 local capture = Ct(content^0) function string:splitlines() return match(capture,self) end patterns.textline = content local splitters_s, splitters_m = { }, { } local function splitat(separator,single) local splitter = (single and splitters_s[separator]) or splitters_m[separator] if not splitter then separator = P(separator) if single then local other, any = C((1 - separator)^0), P(1) splitter = other * (separator * C(any^0) + "") -- ? splitters_s[separator] = splitter else local other = C((1 - separator)^0) splitter = other * (separator * other)^0 splitters_m[separator] = splitter end end return splitter end lpeg.splitat = splitat local cache = { } function lpeg.split(separator,str) local c = cache[separator] if not c then c = Ct(splitat(separator)) cache[separator] = c end return match(c,str) end function string:split(separator) local c = cache[separator] if not c then c = Ct(splitat(separator)) cache[separator] = c end return match(c,self) end lpeg.splitters = cache local cache = { } function lpeg.checkedsplit(separator,str) local c = cache[separator] if not c then separator = P(separator) local other = C((1 - separator)^0) c = Ct(separator^0 * other * (separator^1 * other)^0) cache[separator] = c end return match(c,str) end function string:checkedsplit(separator) local c = cache[separator] if not c then separator = P(separator) local other = C((1 - separator)^0) c = Ct(separator^0 * other * (separator^1 * other)^0) cache[separator] = c end return match(c,self) end local f1 = string.byte local function f2(s) local c1, c2 = f1(s,1,2) return c1 * 64 + c2 - 12416 end local function f3(s) local c1, c2, c3 = f1(s,1,3) return (c1 * 64 + c2) * 64 + c3 - 925824 end local function f4(s) local c1, c2, c3, c4 = f1(s,1,4) return ((c1 * 64 + c2) * 64 + c3) * 64 + c4 - 63447168 end patterns.utf8byte = patterns.utf8one/f1 + patterns.utf8two/f2 + patterns.utf8three/f3 + patterns.utf8four/f4 local cache = { } function lpeg.stripper(str) if type(str) == "string" then local s = cache[str] if not s then s = Cs(((S(str)^1)/"" + 1)^0) cache[str] = s end return s else return Cs(((str^1)/"" + 1)^0) end end local cache = { } function lpeg.keeper(str) if type(str) == "string" then local s = cache[str] if not s then s = Cs((((1-S(str))^1)/"" + 1)^0) cache[str] = s end return s else return Cs((((1-str)^1)/"" + 1)^0) end end function lpeg.replacer(t) if #t > 0 then local p for i=1,#t do local ti= t[i] local pp = P(ti[1]) / ti[2] p = (p and p + pp ) or pp end return Cs((p + 1)^0) end end local splitters_f, splitters_s = { }, { } function lpeg.firstofsplit(separator) -- always return value local splitter = splitters_f[separator] if not splitter then separator = P(separator) splitter = C((1 - separator)^0) splitters_f[separator] = splitter end return splitter end function lpeg.secondofsplit(separator) -- nil if not split local splitter = splitters_s[separator] if not splitter then separator = P(separator) splitter = (1 - separator)^0 * separator * C(P(1)^0) splitters_s[separator] = splitter end return splitter end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-table'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } table.join = table.concat local concat, sort, insert, remove = table.concat, table.sort, table.insert, table.remove local format, find, gsub, lower, dump, match = string.format, string.find, string.gsub, string.lower, string.dump, string.match local getmetatable, setmetatable = getmetatable, setmetatable local type, next, tostring, tonumber, ipairs = type, next, tostring, tonumber, ipairs -- Starting with version 5.2 Lua no longer provide ipairs, which makes -- sense. As we already used the for loop and # in most places the -- impact on ConTeXt was not that large; the remaining ipairs already -- have been replaced. In a similar fashio we also hardly used pairs. -- -- Just in case, we provide the fallbacks as discussed in Programming -- in Lua (http://www.lua.org/pil/7.3.html): if not ipairs then -- for k, v in ipairs(t) do ... end -- for k=1,#t do local v = t[k] ... end local function iterate(a,i) i = i + 1 local v = a[i] if v ~= nil then return i, v --, nil end end function ipairs(a) return iterate, a, 0 end end if not pairs then -- for k, v in pairs(t) do ... end -- for k, v in next, t do ... end function pairs(t) return next, t -- , nil end end -- Also, unpack has been moved to the table table, and for compatiility -- reasons we provide both now. if not table.unpack then table.unpack = _G.unpack elseif not unpack then _G.unpack = table.unpack end -- extra functions, some might go (when not used) function table.strip(tab) local lst = { } for i=1,#tab do local s = gsub(tab[i],"^%s*(.-)%s*$","%1") if s == "" then -- skip this one else lst[#lst+1] = s end end return lst end function table.keys(t) local k = { } for key, _ in next, t do k[#k+1] = key end return k end local function compare(a,b) return (tostring(a) < tostring(b)) end local function sortedkeys(tab) local srt, kind = { }, 0 -- 0=unknown 1=string, 2=number 3=mixed for key,_ in next, tab do srt[#srt+1] = key if kind == 3 then -- no further check else local tkey = type(key) if tkey == "string" then -- if kind == 2 then kind = 3 else kind = 1 end kind = (kind == 2 and 3) or 1 elseif tkey == "number" then -- if kind == 1 then kind = 3 else kind = 2 end kind = (kind == 1 and 3) or 2 else kind = 3 end end end if kind == 0 or kind == 3 then sort(srt,compare) else sort(srt) end return srt end local function sortedhashkeys(tab) -- fast one local srt = { } for key,_ in next, tab do srt[#srt+1] = key end sort(srt) return srt end table.sortedkeys = sortedkeys table.sortedhashkeys = sortedhashkeys local function sortedhash(t) local s = sortedhashkeys(t) -- maybe just sortedkeys local n = 0 local function kv(s) n = n + 1 local k = s[n] return k, t[k] end return kv, s end table.sortedhash = sortedhash table.sortedpairs = sortedhash function table.append(t, list) for _,v in next, list do insert(t,v) end end function table.prepend(t, list) for k,v in next, list do insert(t,k,v) end end function table.merge(t, ...) -- first one is target t = t or { } local lst = {...} for i=1,#lst do for k, v in next, lst[i] do t[k] = v end end return t end function table.merged(...) local tmp, lst = { }, {...} for i=1,#lst do for k, v in next, lst[i] do tmp[k] = v end end return tmp end function table.imerge(t, ...) local lst = {...} for i=1,#lst do local nst = lst[i] for j=1,#nst do t[#t+1] = nst[j] end end return t end function table.imerged(...) local tmp, lst = { }, {...} for i=1,#lst do local nst = lst[i] for j=1,#nst do tmp[#tmp+1] = nst[j] end end return tmp end local function fastcopy(old) -- fast one if old then local new = { } for k,v in next, old do if type(v) == "table" then new[k] = fastcopy(v) -- was just table.copy else new[k] = v end end -- optional second arg local mt = getmetatable(old) if mt then setmetatable(new,mt) end return new else return { } end end local function copy(t, tables) -- taken from lua wiki, slightly adapted tables = tables or { } local tcopy = {} if not tables[t] then tables[t] = tcopy end for i,v in next, t do -- brrr, what happens with sparse indexed if type(i) == "table" then if tables[i] then i = tables[i] else i = copy(i, tables) end end if type(v) ~= "table" then tcopy[i] = v elseif tables[v] then tcopy[i] = tables[v] else tcopy[i] = copy(v, tables) end end local mt = getmetatable(t) if mt then setmetatable(tcopy,mt) end return tcopy end table.fastcopy = fastcopy table.copy = copy -- roughly: copy-loop : unpack : sub == 0.9 : 0.4 : 0.45 (so in critical apps, use unpack) function table.sub(t,i,j) return { unpack(t,i,j) } end function table.replace(a,b) for k,v in next, b do a[k] = v end end -- slower than #t on indexed tables (#t only returns the size of the numerically indexed slice) function table.is_empty(t) -- obolete, use inline code instead return not t or not next(t) end function table.one_entry(t) -- obolete, use inline code instead local n = next(t) return n and not next(t,n) end function table.tohash(t,value) local h = { } if t then if value == nil then value = true end for _, v in next, t do -- no ipairs here h[v] = value end end return h end function table.fromhash(t) local h = { } for k, v in next, t do -- no ipairs here if v then h[#h+1] = k end end return h end table.serialize_functions = true table.serialize_compact = true table.serialize_inline = true local noquotes, hexify, handle, reduce, compact, inline, functions local reserved = table.tohash { -- intercept a language flaw, no reserved words as key 'and', 'break', 'do', 'else', 'elseif', 'end', 'false', 'for', 'function', 'if', 'in', 'local', 'nil', 'not', 'or', 'repeat', 'return', 'then', 'true', 'until', 'while', } local function simple_table(t) if #t > 0 then local n = 0 for _,v in next, t do n = n + 1 end if n == #t then local tt = { } for i=1,#t do local v = t[i] local tv = type(v) if tv == "number" then if hexify then tt[#tt+1] = format("0x%04X",v) else tt[#tt+1] = tostring(v) -- tostring not needed end elseif tv == "boolean" then tt[#tt+1] = tostring(v) elseif tv == "string" then tt[#tt+1] = format("%q",v) else tt = nil break end end return tt end end return nil end -- Because this is a core function of mkiv I moved some function calls -- inline. -- -- twice as fast in a test: -- -- local propername = lpeg.P(lpeg.R("AZ","az","__") * lpeg.R("09","AZ","az", "__")^0 * lpeg.P(-1) ) -- problem: there no good number_to_string converter with the best resolution local function do_serialize(root,name,depth,level,indexed) if level > 0 then depth = depth .. " " if indexed then handle(format("%s{",depth)) elseif name then if type(name) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s[0x%04X]={",depth,name)) else handle(format("%s[%s]={",depth,name)) end elseif noquotes and not reserved[name] and find(name,"^%a[%w%_]*$") then handle(format("%s%s={",depth,name)) else handle(format("%s[%q]={",depth,name)) end else handle(format("%s{",depth)) end end -- we could check for k (index) being number (cardinal) if root and next(root) then local first, last = nil, 0 -- #root cannot be trusted here (will be ok in 5.2 when ipairs is gone) if compact then -- NOT: for k=1,#root do (we need to quit at nil) for k,v in ipairs(root) do -- can we use next? if not first then first = k end last = last + 1 end end local sk = sortedkeys(root) for i=1,#sk do local k = sk[i] local v = root[k] -- circular local t = type(v) if compact and first and type(k) == "number" and k >= first and k <= last then if t == "number" then if hexify then handle(format("%s 0x%04X,",depth,v)) else handle(format("%s %s,",depth,v)) -- %.99g end elseif t == "string" then if reduce and tonumber(v) then handle(format("%s %s,",depth,v)) else handle(format("%s %q,",depth,v)) end elseif t == "table" then if not next(v) then handle(format("%s {},",depth)) elseif inline then -- and #t > 0 local st = simple_table(v) if st then handle(format("%s { %s },",depth,concat(st,", "))) else do_serialize(v,k,depth,level+1,true) end else do_serialize(v,k,depth,level+1,true) end elseif t == "boolean" then handle(format("%s %s,",depth,tostring(v))) elseif t == "function" then if functions then handle(format('%s loadstring(%q),',depth,dump(v))) else handle(format('%s "function",',depth)) end else handle(format("%s %q,",depth,tostring(v))) end elseif k == "__p__" then -- parent if false then handle(format("%s __p__=nil,",depth)) end elseif t == "number" then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=0x%04X,",depth,k,v)) else handle(format("%s [%s]=%s,",depth,k,v)) -- %.99g end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then if hexify then handle(format("%s %s=0x%04X,",depth,k,v)) else handle(format("%s %s=%s,",depth,k,v)) -- %.99g end else if hexify then handle(format("%s [%q]=0x%04X,",depth,k,v)) else handle(format("%s [%q]=%s,",depth,k,v)) -- %.99g end end elseif t == "string" then if reduce and tonumber(v) then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=%s,",depth,k,v)) else handle(format("%s [%s]=%s,",depth,k,v)) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s=%s,",depth,k,v)) else handle(format("%s [%q]=%s,",depth,k,v)) end else if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=%q,",depth,k,v)) else handle(format("%s [%s]=%q,",depth,k,v)) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s=%q,",depth,k,v)) else handle(format("%s [%q]=%q,",depth,k,v)) end end elseif t == "table" then if not next(v) then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]={},",depth,k)) else handle(format("%s [%s]={},",depth,k)) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s={},",depth,k)) else handle(format("%s [%q]={},",depth,k)) end elseif inline then local st = simple_table(v) if st then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]={ %s },",depth,k,concat(st,", "))) else handle(format("%s [%s]={ %s },",depth,k,concat(st,", "))) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s={ %s },",depth,k,concat(st,", "))) else handle(format("%s [%q]={ %s },",depth,k,concat(st,", "))) end else do_serialize(v,k,depth,level+1) end else do_serialize(v,k,depth,level+1) end elseif t == "boolean" then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=%s,",depth,k,tostring(v))) else handle(format("%s [%s]=%s,",depth,k,tostring(v))) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s=%s,",depth,k,tostring(v))) else handle(format("%s [%q]=%s,",depth,k,tostring(v))) end elseif t == "function" then if functions then if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=loadstring(%q),",depth,k,dump(v))) else handle(format("%s [%s]=loadstring(%q),",depth,k,dump(v))) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s=loadstring(%q),",depth,k,dump(v))) else handle(format("%s [%q]=loadstring(%q),",depth,k,dump(v))) end end else if type(k) == "number" then -- or find(k,"^%d+$") then if hexify then handle(format("%s [0x%04X]=%q,",depth,k,tostring(v))) else handle(format("%s [%s]=%q,",depth,k,tostring(v))) end elseif noquotes and not reserved[k] and find(k,"^%a[%w%_]*$") then handle(format("%s %s=%q,",depth,k,tostring(v))) else handle(format("%s [%q]=%q,",depth,k,tostring(v))) end end end end if level > 0 then handle(format("%s},",depth)) end end -- replacing handle by a direct t[#t+1] = ... (plus test) is not much -- faster (0.03 on 1.00 for zapfino.tma) local function serialize(root,name,_handle,_reduce,_noquotes,_hexify) noquotes = _noquotes hexify = _hexify handle = _handle or print reduce = _reduce or false compact = table.serialize_compact inline = compact and table.serialize_inline functions = table.serialize_functions local tname = type(name) if tname == "string" then if name == "return" then handle("return {") else handle(name .. "={") end elseif tname == "number" then if hexify then handle(format("[0x%04X]={",name)) else handle("[" .. name .. "]={") end elseif tname == "boolean" then if name then handle("return {") else handle("{") end else handle("t={") end if root and next(root) then do_serialize(root,name,"",0) end handle("}") end function table.serialize(root,name,reduce,noquotes,hexify) local t = { } local function flush(s) t[#t+1] = s end serialize(root,name,flush,reduce,noquotes,hexify) return concat(t,"\n") end function table.tohandle(handle,root,name,reduce,noquotes,hexify) serialize(root,name,handle,reduce,noquotes,hexify) end -- sometimes tables are real use (zapfino extra pro is some 85M) in which -- case a stepwise serialization is nice; actually, we could consider: -- -- for line in table.serializer(root,name,reduce,noquotes) do -- ...(line) -- end -- -- so this is on the todo list table.tofile_maxtab = 2*1024 function table.tofile(filename,root,name,reduce,noquotes,hexify) local f = io.open(filename,'w') if f then local maxtab = table.tofile_maxtab if maxtab > 1 then local t = { } local function flush(s) t[#t+1] = s if #t > maxtab then f:write(concat(t,"\n"),"\n") -- hm, write(sometable) should be nice t = { } end end serialize(root,name,flush,reduce,noquotes,hexify) f:write(concat(t,"\n"),"\n") else local function flush(s) f:write(s,"\n") end serialize(root,name,flush,reduce,noquotes,hexify) end f:close() end end local function flatten(t,f,complete) -- is this used? meybe a variant with next, ... for i=1,#t do local v = t[i] if type(v) == "table" then if complete or type(v[1]) == "table" then flatten(v,f,complete) else f[#f+1] = v end else f[#f+1] = v end end end function table.flatten(t) local f = { } flatten(t,f,true) return f end function table.unnest(t) -- bad name local f = { } flatten(t,f,false) return f end table.flatten_one_level = table.unnest -- a better one: local function flattened(t,f) if not f then f = { } end for k, v in next, t do if type(v) == "table" then flattened(v,f) else f[k] = v end end return f end table.flattened = flattened -- the next three may disappear function table.remove_value(t,value) -- todo: n if value then for i=1,#t do if t[i] == value then remove(t,i) -- remove all, so no: return end end end end function table.insert_before_value(t,value,str) if str then if value then for i=1,#t do if t[i] == value then insert(t,i,str) return end end end insert(t,1,str) elseif value then insert(t,1,value) end end function table.insert_after_value(t,value,str) if str then if value then for i=1,#t do if t[i] == value then insert(t,i+1,str) return end end end t[#t+1] = str elseif value then t[#t+1] = value end end local function are_equal(a,b,n,m) -- indexed if a and b and #a == #b then n = n or 1 m = m or #a for i=n,m do local ai, bi = a[i], b[i] if ai==bi then -- same elseif type(ai)=="table" and type(bi)=="table" then if not are_equal(ai,bi) then return false end else return false end end return true else return false end end local function identical(a,b) -- assumes same structure for ka, va in next, a do local vb = b[k] if va == vb then -- same elseif type(va) == "table" and type(vb) == "table" then if not identical(va,vb) then return false end else return false end end return true end table.are_equal = are_equal table.identical = identical -- maybe also make a combined one function table.compact(t) if t then for k,v in next, t do if not next(v) then t[k] = nil end end end end function table.contains(t, v) if t then for i=1, #t do if t[i] == v then return i end end end return false end function table.count(t) local n, e = 0, next(t) while e do n, e = n + 1, next(t,e) end return n end function table.swapped(t,s) local n = { } if s then for k, v in next, s do n[k] = v end end for k, v in next, t do n[v] = k end return n end function table.clone(t,p) -- t is optional or nil or table if not p then t, p = { }, t or { } elseif not t then t = { } end setmetatable(t, { __index = function(_,key) return p[key] end }) -- why not __index = p ? return t end function table.hexed(t,seperator) local tt = { } for i=1,#t do tt[i] = format("0x%04X",t[i]) end return concat(tt,seperator or " ") end function table.reverse_hash(h) -- needs another name local r = { } for k,v in next, h do r[v] = lower(gsub(k," ","")) end return r end function table.reverse(t) local tt = { } if #t > 0 then for i=#t,1,-1 do tt[#tt+1] = t[i] end end return tt end function table.insert_before_value(t,value,extra) for i=1,#t do if t[i] == extra then remove(t,i) end end for i=1,#t do if t[i] == value then insert(t,i,extra) return end end insert(t,1,extra) end function table.insert_after_value(t,value,extra) for i=1,#t do if t[i] == extra then remove(t,i) end end for i=1,#t do if t[i] == value then insert(t,i+1,extra) return end end insert(t,#t+1,extra) end function table.sequenced(t,sep,simple) -- hash only local s = { } for k, v in sortedhash(t) do if simple then if v == true then s[#s+1] = k elseif v and v~= "" then s[#s+1] = k .. "=" .. tostring(v) end else s[#s+1] = k .. "=" .. tostring(v) end end return concat(s, sep or " | ") end function table.print(...) print(table.serialize(...)) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-io'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local byte, find, gsub = string.byte, string.find, string.gsub if string.find(os.getenv("PATH"),";") then io.fileseparator, io.pathseparator = "\\", ";" else io.fileseparator, io.pathseparator = "/" , ":" end function io.loaddata(filename,textmode) local f = io.open(filename,(textmode and 'r') or 'rb') if f then -- collectgarbage("step") -- sometimes makes a big difference in mem consumption local data = f:read('*all') -- garbagecollector.check(data) f:close() return data else return nil end end function io.savedata(filename,data,joiner) local f = io.open(filename,"wb") if f then if type(data) == "table" then f:write(table.join(data,joiner or "")) elseif type(data) == "function" then data(f) else f:write(data or "") end f:close() return true else return false end end function io.exists(filename) local f = io.open(filename) if f == nil then return false else assert(f:close()) return true end end function io.size(filename) local f = io.open(filename) if f == nil then return 0 else local s = f:seek("end") assert(f:close()) return s end end function io.noflines(f) local n = 0 for _ in f:lines() do n = n + 1 end f:seek('set',0) return n end local nextchar = { [ 4] = function(f) return f:read(1,1,1,1) end, [ 2] = function(f) return f:read(1,1) end, [ 1] = function(f) return f:read(1) end, [-2] = function(f) local a, b = f:read(1,1) return b, a end, [-4] = function(f) local a, b, c, d = f:read(1,1,1,1) return d, c, b, a end } function io.characters(f,n) if f then return nextchar[n or 1], f else return nil, nil end end local nextbyte = { [4] = function(f) local a, b, c, d = f:read(1,1,1,1) if d then return byte(a), byte(b), byte(c), byte(d) else return nil, nil, nil, nil end end, [2] = function(f) local a, b = f:read(1,1) if b then return byte(a), byte(b) else return nil, nil end end, [1] = function (f) local a = f:read(1) if a then return byte(a) else return nil end end, [-2] = function (f) local a, b = f:read(1,1) if b then return byte(b), byte(a) else return nil, nil end end, [-4] = function(f) local a, b, c, d = f:read(1,1,1,1) if d then return byte(d), byte(c), byte(b), byte(a) else return nil, nil, nil, nil end end } function io.bytes(f,n) if f then return nextbyte[n or 1], f else return nil, nil end end function io.ask(question,default,options) while true do io.write(question) if options then io.write(string.format(" [%s]",table.concat(options,"|"))) end if default then io.write(string.format(" [%s]",default)) end io.write(string.format(" ")) local answer = io.read() answer = gsub(answer,"^%s*(.*)%s*$","%1") if answer == "" and default then return default elseif not options then return answer else for k=1,#options do if options[k] == answer then return answer end end local pattern = "^" .. answer for k=1,#options do local v = options[k] if find(v,pattern) then return v end end end end end function io.readnumber(f,n,m) if m then f:seek("set",n) n = m end if n == 1 then return byte(f:read(1)) elseif n == 2 then local a, b = byte(f:read(2),1,2) return 256*a + b elseif n == 4 then local a, b, c, d = byte(f:read(4),1,4) return 256^3 * a + 256^2 * b + 256*c + d elseif n == 8 then local a, b = readnumber(f,4), readnumber(f,4) return 256 * b + c elseif n == 12 then local a, b, c = readnumber(f,4), readnumber(f,4), readnumber(f,4) return 256^2 * a + 256 * b + c else return 0 end end function io.readstring(f,n,m) if m then f:seek("set",n) n = m end local str = gsub(f:read(n),"%z","") return str end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-number'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local tostring = tostring local format, floor, insert, match = string.format, math.floor, table.insert, string.match local lpegmatch = lpeg.match number = number or { } -- a,b,c,d,e,f = number.toset(100101) function number.toset(n) return match(tostring(n),"(.?)(.?)(.?)(.?)(.?)(.?)(.?)(.?)") end function number.toevenhex(n) local s = format("%X",n) if #s % 2 == 0 then return s else return "0" .. s end end -- the lpeg way is slower on 8 digits, but faster on 4 digits, some 7.5% -- on -- -- for i=1,1000000 do -- local a,b,c,d,e,f,g,h = number.toset(12345678) -- local a,b,c,d = number.toset(1234) -- local a,b,c = number.toset(123) -- end -- -- of course dedicated "(.)(.)(.)(.)" matches are even faster local one = lpeg.C(1-lpeg.S(''))^1 function number.toset(n) return lpegmatch(one,tostring(n)) end function number.bits(n,zero) local t, i = { }, (zero and 0) or 1 while n > 0 do local m = n % 2 if m > 0 then insert(t,1,i) end n = floor(n/2) i = i + 1 end return t end function number.bit(p) return 2 ^ (p - 1) -- 1-based indexing end function number.hasbit(x, p) -- typical call: if hasbit(x, bit(3)) then ... return x % (p + p) >= p end function number.setbit(x, p) return hasbit(x, p) and x or x + p end function number.clearbit(x, p) return hasbit(x, p) and x - p or x end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-set'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } set = set or { } local nums = { } local tabs = { } local concat = table.concat local next, type = next, type set.create = table.tohash function set.tonumber(t) if next(t) then local s = "" -- we could save mem by sorting, but it slows down for k, v in next, t do if v then -- why bother about the leading space s = s .. " " .. k end end local n = nums[s] if not n then n = #tabs + 1 tabs[n] = t nums[s] = n end return n else return 0 end end function set.totable(n) if n == 0 then return { } else return tabs[n] or { } end end function set.tolist(n) if n == 0 or not tabs[n] then return "" else local t = { } for k, v in next, tabs[n] do if v then t[#t+1] = k end end return concat(t," ") end end function set.contains(n,s) if type(n) == "table" then return n[s] elseif n == 0 then return false else local t = tabs[n] return t and t[s] end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-os'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- maybe build io.flush in os.execute local find, format, gsub, upper = string.find, string.format, string.gsub, string.upper local random, ceil = math.random, math.ceil local rawget, rawset, type, getmetatable, setmetatable, tonumber = rawget, rawset, type, getmetatable, setmetatable, tonumber local rawget, rawset, type, getmetatable, setmetatable, tonumber = rawget, rawset, type, getmetatable, setmetatable, tonumber -- The following code permits traversing the environment table, at least -- in luatex. Internally all environment names are uppercase. if not os.__getenv__ then os.__getenv__ = os.getenv os.__setenv__ = os.setenv if os.env then local osgetenv = os.getenv local ossetenv = os.setenv local osenv = os.env local _ = osenv.PATH -- initialize the table function os.setenv(k,v) if v == nil then v = "" end local K = upper(k) osenv[K] = v ossetenv(K,v) end function os.getenv(k) local K = upper(k) local v = osenv[K] or osenv[k] or osgetenv(K) or osgetenv(k) if v == "" then return nil else return v end end else local ossetenv = os.setenv local osgetenv = os.getenv local osenv = { } function os.setenv(k,v) if v == nil then v = "" end local K = upper(k) osenv[K] = v end function os.getenv(k) local K = upper(k) local v = osenv[K] or osgetenv(K) or osgetenv(k) if v == "" then return nil else return v end end local function __index(t,k) return os.getenv(k) end local function __newindex(t,k,v) os.setenv(k,v) end os.env = { } setmetatable(os.env, { __index = __index, __newindex = __newindex } ) end end -- end of environment hack local execute, spawn, exec, iopopen, ioflush = os.execute, os.spawn or os.execute, os.exec or os.execute, io.popen, io.flush function os.execute(...) ioflush() return execute(...) end function os.spawn (...) ioflush() return spawn (...) end function os.exec (...) ioflush() return exec (...) end function io.popen (...) ioflush() return iopopen(...) end function os.resultof(command) local handle = io.popen(command,"r") return handle and handle:read("*all") or "" end if not io.fileseparator then if find(os.getenv("PATH"),";") then io.fileseparator, io.pathseparator, os.type = "\\", ";", os.type or "mswin" else io.fileseparator, io.pathseparator, os.type = "/" , ":", os.type or "unix" end end os.type = os.type or (io.pathseparator == ";" and "windows") or "unix" os.name = os.name or (os.type == "windows" and "mswin" ) or "linux" if os.type == "windows" then os.libsuffix, os.binsuffix = 'dll', 'exe' else os.libsuffix, os.binsuffix = 'so', '' end function os.launch(str) if os.type == "windows" then os.execute("start " .. str) -- os.spawn ? else os.execute(str .. " &") -- os.spawn ? end end if not os.times then -- utime = user time -- stime = system time -- cutime = children user time -- cstime = children system time function os.times() return { utime = os.gettimeofday(), -- user stime = 0, -- system cutime = 0, -- children user cstime = 0, -- children system } end end os.gettimeofday = os.gettimeofday or os.clock local startuptime = os.gettimeofday() function os.runtime() return os.gettimeofday() - startuptime end -- no need for function anymore as we have more clever code and helpers now -- this metatable trickery might as well disappear os.resolvers = os.resolvers or { } local resolvers = os.resolvers local osmt = getmetatable(os) or { __index = function(t,k) t[k] = "unset" return "unset" end } -- maybe nil local osix = osmt.__index osmt.__index = function(t,k) return (resolvers[k] or osix)(t,k) end setmetatable(os,osmt) -- we can use HOSTTYPE on some platforms local name, platform = os.name or "linux", os.getenv("MTX_PLATFORM") or "" local function guess() local architecture = os.resultof("uname -m") or "" if architecture ~= "" then return architecture end architecture = os.getenv("HOSTTYPE") or "" if architecture ~= "" then return architecture end return os.resultof("echo $HOSTTYPE") or "" end if platform ~= "" then os.platform = platform elseif os.type == "windows" then -- we could set the variable directly, no function needed here function os.resolvers.platform(t,k) local platform, architecture = "", os.getenv("PROCESSOR_ARCHITECTURE") or "" if find(architecture,"AMD64") then platform = "mswin-64" else platform = "mswin" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end elseif name == "linux" then function os.resolvers.platform(t,k) -- we sometims have HOSTTYPE set so let's check that first local platform, architecture = "", os.getenv("HOSTTYPE") or os.resultof("uname -m") or "" if find(architecture,"x86_64") then platform = "linux-64" elseif find(architecture,"ppc") then platform = "linux-ppc" else platform = "linux" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end elseif name == "macosx" then --[[ Identifying the architecture of OSX is quite a mess and this is the best we can come up with. For some reason $HOSTTYPE is a kind of pseudo environment variable, not known to the current environment. And yes, uname cannot be trusted either, so there is a change that you end up with a 32 bit run on a 64 bit system. Also, some proper 64 bit intel macs are too cheap (low-end) and therefore not permitted to run the 64 bit kernel. ]]-- function os.resolvers.platform(t,k) -- local platform, architecture = "", os.getenv("HOSTTYPE") or "" -- if architecture == "" then -- architecture = os.resultof("echo $HOSTTYPE") or "" -- end local platform, architecture = "", os.resultof("echo $HOSTTYPE") or "" if architecture == "" then -- print("\nI have no clue what kind of OSX you're running so let's assume an 32 bit intel.\n") platform = "osx-intel" elseif find(architecture,"i386") then platform = "osx-intel" elseif find(architecture,"x86_64") then platform = "osx-64" else platform = "osx-ppc" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end elseif name == "sunos" then function os.resolvers.platform(t,k) local platform, architecture = "", os.resultof("uname -m") or "" if find(architecture,"sparc") then platform = "solaris-sparc" else -- if architecture == 'i86pc' platform = "solaris-intel" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end elseif name == "freebsd" then function os.resolvers.platform(t,k) local platform, architecture = "", os.resultof("uname -m") or "" if find(architecture,"amd64") then platform = "freebsd-amd64" else platform = "freebsd" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end elseif name == "kfreebsd" then function os.resolvers.platform(t,k) -- we sometims have HOSTTYPE set so let's check that first local platform, architecture = "", os.getenv("HOSTTYPE") or os.resultof("uname -m") or "" if find(architecture,"x86_64") then platform = "kfreebsd-amd64" else platform = "kfreebsd-i386" end os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end else -- platform = "linux" -- os.setenv("MTX_PLATFORM",platform) -- os.platform = platform function os.resolvers.platform(t,k) local platform = "linux" os.setenv("MTX_PLATFORM",platform) os.platform = platform return platform end end -- beware, we set the randomseed -- from wikipedia: Version 4 UUIDs use a scheme relying only on random numbers. This algorithm sets the -- version number as well as two reserved bits. All other bits are set using a random or pseudorandom -- data source. Version 4 UUIDs have the form xxxxxxxx-xxxx-4xxx-yxxx-xxxxxxxxxxxx with hexadecimal -- digits x and hexadecimal digits 8, 9, A, or B for y. e.g. f47ac10b-58cc-4372-a567-0e02b2c3d479. -- -- as we don't call this function too often there is not so much risk on repetition local t = { 8, 9, "a", "b" } function os.uuid() return format("%04x%04x-4%03x-%s%03x-%04x-%04x%04x%04x", random(0xFFFF),random(0xFFFF), random(0x0FFF), t[ceil(random(4))] or 8,random(0x0FFF), random(0xFFFF), random(0xFFFF),random(0xFFFF),random(0xFFFF) ) end local d function os.timezone(delta) d = d or tonumber(tonumber(os.date("%H")-os.date("!%H"))) if delta then if d > 0 then return format("+%02i:00",d) else return format("-%02i:00",-d) end else return 1 end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-file'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- needs a cleanup file = file or { } local insert, concat = table.insert, table.concat local find, gmatch, match, gsub, sub, char = string.find, string.gmatch, string.match, string.gsub, string.sub, string.char local lpegmatch = lpeg.match local getcurrentdir = lfs.currentdir local function dirname(name,default) return match(name,"^(.+)[/\\].-$") or (default or "") end local function basename(name) return match(name,"^.+[/\\](.-)$") or name end local function nameonly(name) return (gsub(match(name,"^.+[/\\](.-)$") or name,"%..*$","")) end local function extname(name,default) return match(name,"^.+%.([^/\\]-)$") or default or "" end local function splitname(name) local n, s = match(name,"^(.+)%.([^/\\]-)$") return n or name, s or "" end file.basename = basename file.dirname = dirname file.nameonly = nameonly file.extname = extname file.suffix = extname function file.removesuffix(filename) return (gsub(filename,"%.[%a%d]+$","")) end function file.addsuffix(filename, suffix, criterium) if not suffix or suffix == "" then return filename elseif criterium == true then return filename .. "." .. suffix elseif not criterium then local n, s = splitname(filename) if not s or s == "" then return filename .. "." .. suffix else return filename end else local n, s = splitname(filename) if s and s ~= "" then local t = type(criterium) if t == "table" then -- keep if in criterium for i=1,#criterium do if s == criterium[i] then return filename end end elseif t == "string" then -- keep if criterium if s == criterium then return filename end end end return n .. "." .. suffix end end function file.replacesuffix(filename, suffix) return (gsub(filename,"%.[%a%d]+$","")) .. "." .. suffix end local trick_1 = char(1) local trick_2 = "^" .. trick_1 .. "/+" function file.join(...) local lst = { ... } local a, b = lst[1], lst[2] if a == "" then lst[1] = trick_1 elseif b and find(a,"^/+$") and find(b,"^/") then lst[1] = "" lst[2] = gsub(b,"^/+","") end local pth = concat(lst,"/") pth = gsub(pth,"\\","/") local a, b = match(pth,"^(.*://)(.*)$") if a and b then return a .. gsub(b,"//+","/") end a, b = match(pth,"^(//)(.*)$") if a and b then return a .. gsub(b,"//+","/") end pth = gsub(pth,trick_2,"") return (gsub(pth,"//+","/")) end function file.iswritable(name) local a = lfs.attributes(name) or lfs.attributes(dirname(name,".")) return a and sub(a.permissions,2,2) == "w" end function file.isreadable(name) local a = lfs.attributes(name) return a and sub(a.permissions,1,1) == "r" end file.is_readable = file.isreadable file.is_writable = file.iswritable -- todo: lpeg local checkedsplit = string.checkedsplit function file.split_path(str,separator) str = gsub(str,"\\","/") return checkedsplit(str,separator or io.pathseparator) end function file.join_path(tab) return concat(tab,io.pathseparator) -- can have trailing // end -- we can hash them weakly function file.collapse_path(str,anchor) if anchor and not find(str,"^/") and not find(str,"^%a:") then str = getcurrentdir() .. "/" .. str end if str == "" or str =="." then return "." elseif find(str,"^%.%.") then str = gsub(str,"\\","/") return str elseif not find(str,"%.") then str = gsub(str,"\\","/") return str end str = gsub(str,"\\","/") local starter, rest = match(str,"^(%a+:/*)(.-)$") if starter then str = rest end local oldelements = checkedsplit(str,"/") local newelements = { } local i = #oldelements while i > 0 do local element = oldelements[i] if element == '.' then -- do nothing elseif element == '..' then local n = i -1 while n > 0 do local element = oldelements[n] if element ~= '..' and element ~= '.' then oldelements[n] = '.' break else n = n - 1 end end if n < 1 then insert(newelements,1,'..') end elseif element ~= "" then insert(newelements,1,element) end i = i - 1 end if #newelements == 0 then return starter or "." elseif starter then return starter .. concat(newelements, '/') elseif find(str,"^/") then return "/" .. concat(newelements,'/') else return concat(newelements, '/') end end function file.robustname(str) return (gsub(str,"[^%a%d%/%-%.\\]+","-")) end file.readdata = io.loaddata file.savedata = io.savedata function file.copy(oldname,newname) file.savedata(newname,io.loaddata(oldname)) end -- lpeg variants, slightly faster, not always -- also rewrite previous local letter = lpeg.R("az","AZ") + lpeg.S("_-+") local separator = lpeg.P("://") local qualified = lpeg.P(".")^0 * lpeg.P("/") + letter*lpeg.P(":") + letter^1*separator + letter^1 * lpeg.P("/") local rootbased = lpeg.P("/") + letter*lpeg.P(":") -- ./name ../name /name c: :// name/name function file.is_qualified_path(filename) return lpegmatch(qualified,filename) ~= nil end function file.is_rootbased_path(filename) return lpegmatch(rootbased,filename) ~= nil end local slash = lpeg.S("\\/") local period = lpeg.P(".") local drive = lpeg.C(lpeg.R("az","AZ")) * lpeg.P(":") local path = lpeg.C(((1-slash)^0 * slash)^0) local suffix = period * lpeg.C(lpeg.P(1-period)^0 * lpeg.P(-1)) local base = lpeg.C((1-suffix)^0) local pattern = (drive + lpeg.Cc("")) * (path + lpeg.Cc("")) * (base + lpeg.Cc("")) * (suffix + lpeg.Cc("")) function file.splitname(str) -- returns drive, path, base, suffix return lpegmatch(pattern,str) end -- function test(t) for k, v in next, t do print(v, "=>", file.splitname(v)) end end -- -- test { "c:", "c:/aa", "c:/aa/bb", "c:/aa/bb/cc", "c:/aa/bb/cc.dd", "c:/aa/bb/cc.dd.ee" } -- test { "c:", "c:aa", "c:aa/bb", "c:aa/bb/cc", "c:aa/bb/cc.dd", "c:aa/bb/cc.dd.ee" } -- test { "/aa", "/aa/bb", "/aa/bb/cc", "/aa/bb/cc.dd", "/aa/bb/cc.dd.ee" } -- test { "aa", "aa/bb", "aa/bb/cc", "aa/bb/cc.dd", "aa/bb/cc.dd.ee" } end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-md5'] = { version = 1.001, author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- This also provides file checksums and checkers. local gsub, format, byte = string.gsub, string.format, string.byte local function convert(str,fmt) return (gsub(md5.sum(str),".",function(chr) return format(fmt,byte(chr)) end)) end if not md5.HEX then function md5.HEX(str) return convert(str,"%02X") end end if not md5.hex then function md5.hex(str) return convert(str,"%02x") end end if not md5.dec then function md5.dec(str) return convert(str,"%03i") end end file.needs_updating_threshold = 1 function file.needs_updating(oldname,newname) -- size modification access change local oldtime = lfs.attributes(oldname, modification) local newtime = lfs.attributes(newname, modification) if newtime >= oldtime then return false elseif oldtime - newtime < file.needs_updating_threshold then return false else return true end end function file.checksum(name) if md5 then local data = io.loaddata(name) if data then return md5.HEX(data) end end return nil end function file.loadchecksum(name) if md5 then local data = io.loaddata(name .. ".md5") return data and (gsub(data,"%s","")) end return nil end function file.savechecksum(name, checksum) if not checksum then checksum = file.checksum(name) end if checksum then io.savedata(name .. ".md5",checksum) return checksum end return nil end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-url'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local char, gmatch, gsub, format, byte = string.char, string.gmatch, string.gsub, string.format, string.byte local concat = table.concat local tonumber, type = tonumber, type local lpegmatch, lpegP, lpegC, lpegR, lpegS, lpegCs, lpegCc = lpeg.match, lpeg.P, lpeg.C, lpeg.R, lpeg.S, lpeg.Cs, lpeg.Cc -- from the spec (on the web): -- -- foo://example.com:8042/over/there?name=ferret#nose -- \_/ \______________/\_________/ \_________/ \__/ -- | | | | | -- scheme authority path query fragment -- | _____________________|__ -- / \ / \ -- urn:example:animal:ferret:nose url = url or { } local function tochar(s) return char(tonumber(s,16)) end local colon, qmark, hash, slash, percent, endofstring = lpegP(":"), lpegP("?"), lpegP("#"), lpegP("/"), lpegP("%"), lpegP(-1) local hexdigit = lpegR("09","AF","af") local plus = lpegP("+") local nothing = lpegCc("") local escaped = (plus / " ") + (percent * lpegC(hexdigit * hexdigit) / tochar) -- we assume schemes with more than 1 character (in order to avoid problems with windows disks) local scheme = lpegCs((escaped+(1-colon-slash-qmark-hash))^2) * colon + nothing local authority = slash * slash * lpegCs((escaped+(1- slash-qmark-hash))^0) + nothing local path = slash * lpegCs((escaped+(1- qmark-hash))^0) + nothing local query = qmark * lpegCs((escaped+(1- hash))^0) + nothing local fragment = hash * lpegCs((escaped+(1- endofstring))^0) + nothing local parser = lpeg.Ct(scheme * authority * path * query * fragment) lpeg.patterns.urlsplitter = parser local escapes = { } for i=0,255 do escapes[i] = format("%%%02X",i) end local escaper = lpeg.Cs((lpegR("09","AZ","az") + lpegS("-./_") + lpegP(1) / escapes)^0) lpeg.patterns.urlescaper = escaper -- todo: reconsider Ct as we can as well have five return values (saves a table) -- so we can have two parsers, one with and one without function url.split(str) return (type(str) == "string" and lpegmatch(parser,str)) or str end -- todo: cache them function url.hashed(str) local s = url.split(str) local somescheme = s[1] ~= "" if not somescheme then return { scheme = "file", authority = "", path = str, query = "", fragment = "", original = str, noscheme = true, } else return { scheme = s[1], authority = s[2], path = s[3], query = s[4], fragment = s[5], original = str, noscheme = false, } end end function url.hasscheme(str) return url.split(str)[1] ~= "" end function url.addscheme(str,scheme) return (url.hasscheme(str) and str) or ((scheme or "file:///") .. str) end function url.construct(hash) -- dodo: we need to escape ! local fullurl = { } local scheme, authority, path, query, fragment = hash.scheme, hash.authority, hash.path, hash.query, hash.fragment if scheme and scheme ~= "" then fullurl[#fullurl+1] = scheme .. "://" end if authority and authority ~= "" then fullurl[#fullurl+1] = authority end if path and path ~= "" then fullurl[#fullurl+1] = "/" .. path end if query and query ~= "" then fullurl[#fullurl+1] = "?".. query end if fragment and fragment ~= "" then fullurl[#fullurl+1] = "#".. fragment end return lpegmatch(escaper,concat(fullurl)) end function url.filename(filename) local t = url.hashed(filename) return (t.scheme == "file" and (gsub(t.path,"^/([a-zA-Z])([:|])/)","%1:"))) or filename end function url.query(str) if type(str) == "string" then local t = { } for k, v in gmatch(str,"([^&=]*)=([^&=]*)") do t[k] = v end return t else return str end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-dir'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- dir.expand_name will be merged with cleanpath and collapsepath local type = type local find, gmatch, match, gsub = string.find, string.gmatch, string.match, string.gsub local lpegmatch = lpeg.match dir = dir or { } -- handy function dir.current() return (gsub(lfs.currentdir(),"\\","/")) end -- optimizing for no string.find (*) does not save time local attributes = lfs.attributes local walkdir = lfs.dir local function glob_pattern(path,patt,recurse,action) local ok, scanner if path == "/" then ok, scanner = xpcall(function() return walkdir(path..".") end, function() end) -- kepler safe else ok, scanner = xpcall(function() return walkdir(path) end, function() end) -- kepler safe end if ok and type(scanner) == "function" then if not find(path,"/$") then path = path .. '/' end for name in scanner do local full = path .. name local mode = attributes(full,'mode') if mode == 'file' then if find(full,patt) then action(full) end elseif recurse and (mode == "directory") and (name ~= '.') and (name ~= "..") then glob_pattern(full,patt,recurse,action) end end end end dir.glob_pattern = glob_pattern local function collect_pattern(path,patt,recurse,result) local ok, scanner result = result or { } if path == "/" then ok, scanner = xpcall(function() return walkdir(path..".") end, function() end) -- kepler safe else ok, scanner = xpcall(function() return walkdir(path) end, function() end) -- kepler safe end if ok and type(scanner) == "function" then if not find(path,"/$") then path = path .. '/' end for name in scanner do local full = path .. name local attr = attributes(full) local mode = attr.mode if mode == 'file' then if find(full,patt) then result[name] = attr end elseif recurse and (mode == "directory") and (name ~= '.') and (name ~= "..") then attr.list = collect_pattern(full,patt,recurse) result[name] = attr end end end return result end dir.collect_pattern = collect_pattern local P, S, R, C, Cc, Cs, Ct, Cv, V = lpeg.P, lpeg.S, lpeg.R, lpeg.C, lpeg.Cc, lpeg.Cs, lpeg.Ct, lpeg.Cv, lpeg.V local pattern = Ct { [1] = (C(P(".") + P("/")^1) + C(R("az","AZ") * P(":") * P("/")^0) + Cc("./")) * V(2) * V(3), [2] = C(((1-S("*?/"))^0 * P("/"))^0), [3] = C(P(1)^0) } local filter = Cs ( ( P("**") / ".*" + P("*") / "[^/]*" + P("?") / "[^/]" + P(".") / "%%." + P("+") / "%%+" + P("-") / "%%-" + P(1) )^0 ) local function glob(str,t) if type(t) == "function" then if type(str) == "table" then for s=1,#str do glob(str[s],t) end elseif lfs.isfile(str) then t(str) else local split = lpegmatch(pattern,str) if split then local root, path, base = split[1], split[2], split[3] local recurse = find(base,"%*%*") local start = root .. path local result = lpegmatch(filter,start .. base) glob_pattern(start,result,recurse,t) end end else if type(str) == "table" then local t = t or { } for s=1,#str do glob(str[s],t) end return t elseif lfs.isfile(str) then local t = t or { } t[#t+1] = str return t else local split = lpegmatch(pattern,str) if split then local t = t or { } local action = action or function(name) t[#t+1] = name end local root, path, base = split[1], split[2], split[3] local recurse = find(base,"%*%*") local start = root .. path local result = lpegmatch(filter,start .. base) glob_pattern(start,result,recurse,action) return t else return { } end end end end dir.glob = glob local function globfiles(path,recurse,func,files) -- func == pattern or function if type(func) == "string" then local s = func -- alas, we need this indirect way func = function(name) return find(name,s) end end files = files or { } for name in walkdir(path) do if find(name,"^%.") then --- skip else local mode = attributes(name,'mode') if mode == "directory" then if recurse then globfiles(path .. "/" .. name,recurse,func,files) end elseif mode == "file" then if func then if func(name) then files[#files+1] = path .. "/" .. name end else files[#files+1] = path .. "/" .. name end end end end return files end dir.globfiles = globfiles -- t = dir.glob("c:/data/develop/context/sources/**/????-*.tex") -- t = dir.glob("c:/data/develop/tex/texmf/**/*.tex") -- t = dir.glob("c:/data/develop/context/texmf/**/*.tex") -- t = dir.glob("f:/minimal/tex/**/*") -- print(dir.ls("f:/minimal/tex/**/*")) -- print(dir.ls("*.tex")) function dir.ls(pattern) return table.concat(glob(pattern),"\n") end local make_indeed = true -- false if string.find(os.getenv("PATH"),";") then -- os.type == "windows" function dir.mkdirs(...) local str, pth, t = "", "", { ... } for i=1,#t do local s = t[i] if s ~= "" then if str ~= "" then str = str .. "/" .. s else str = s end end end local first, middle, last local drive = false first, middle, last = match(str,"^(//)(//*)(.*)$") if first then -- empty network path == local path else first, last = match(str,"^(//)/*(.-)$") if first then middle, last = match(str,"([^/]+)/+(.-)$") if middle then pth = "//" .. middle else pth = "//" .. last last = "" end else first, middle, last = match(str,"^([a-zA-Z]:)(/*)(.-)$") if first then pth, drive = first .. middle, true else middle, last = match(str,"^(/*)(.-)$") if not middle then last = str end end end end for s in gmatch(last,"[^/]+") do if pth == "" then pth = s elseif drive then pth, drive = pth .. s, false else pth = pth .. "/" .. s end if make_indeed and not lfs.isdir(pth) then lfs.mkdir(pth) end end return pth, (lfs.isdir(pth) == true) end function dir.expand_name(str) -- will be merged with cleanpath and collapsepath local first, nothing, last = match(str,"^(//)(//*)(.*)$") if first then first = dir.current() .. "/" end if not first then first, last = match(str,"^(//)/*(.*)$") end if not first then first, last = match(str,"^([a-zA-Z]:)(.*)$") if first and not find(last,"^/") then local d = lfs.currentdir() if lfs.chdir(first) then first = dir.current() end lfs.chdir(d) end end if not first then first, last = dir.current(), str end last = gsub(last,"//","/") last = gsub(last,"/%./","/") last = gsub(last,"^/*","") first = gsub(first,"/*$","") if last == "" then return first else return first .. "/" .. last end end else function dir.mkdirs(...) local str, pth, t = "", "", { ... } for i=1,#t do local s = t[i] if s and s ~= "" then -- we catch nil and false if str ~= "" then str = str .. "/" .. s else str = s end end end str = gsub(str,"/+","/") if find(str,"^/") then pth = "/" for s in gmatch(str,"[^/]+") do local first = (pth == "/") if first then pth = pth .. s else pth = pth .. "/" .. s end if make_indeed and not first and not lfs.isdir(pth) then lfs.mkdir(pth) end end else pth = "." for s in gmatch(str,"[^/]+") do pth = pth .. "/" .. s if make_indeed and not lfs.isdir(pth) then lfs.mkdir(pth) end end end return pth, (lfs.isdir(pth) == true) end function dir.expand_name(str) -- will be merged with cleanpath and collapsepath if not find(str,"^/") then str = lfs.currentdir() .. "/" .. str end str = gsub(str,"//","/") str = gsub(str,"/%./","/") return str end end dir.makedirs = dir.mkdirs end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-boolean'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } boolean = boolean or { } local type, tonumber = type, tonumber function boolean.tonumber(b) if b then return 1 else return 0 end end function toboolean(str,tolerant) if tolerant then local tstr = type(str) if tstr == "string" then return str == "true" or str == "yes" or str == "on" or str == "1" or str == "t" elseif tstr == "number" then return tonumber(str) ~= 0 elseif tstr == "nil" then return false else return str end elseif str == "true" then return true elseif str == "false" then return false else return str end end function string.is_boolean(str,default) if type(str) == "string" then if str == "true" or str == "yes" or str == "on" or str == "t" then return true elseif str == "false" or str == "no" or str == "off" or str == "f" then return false end end return default end function boolean.alwaystrue() return true end function boolean.falsetrue() return false end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-unicode'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } if not unicode then unicode = { utf8 = { } } local floor, char = math.floor, string.char function unicode.utf8.utfchar(n) if n < 0x80 then return char(n) elseif n < 0x800 then return char(0xC0 + floor(n/0x40)) .. char(0x80 + (n % 0x40)) elseif n < 0x10000 then return char(0xE0 + floor(n/0x1000)) .. char(0x80 + (floor(n/0x40) % 0x40)) .. char(0x80 + (n % 0x40)) elseif n < 0x40000 then return char(0xF0 + floor(n/0x40000)) .. char(0x80 + floor(n/0x1000)) .. char(0x80 + (floor(n/0x40) % 0x40)) .. char(0x80 + (n % 0x40)) else -- wrong: -- return char(0xF1 + floor(n/0x1000000)) .. char(0x80 + floor(n/0x40000)) .. char(0x80 + floor(n/0x1000)) .. char(0x80 + (floor(n/0x40) % 0x40)) .. char(0x80 + (n % 0x40)) return "?" end end end utf = utf or unicode.utf8 local concat, utfchar, utfgsub = table.concat, utf.char, utf.gsub local char, byte, find, bytepairs, utfvalues, format = string.char, string.byte, string.find, string.bytepairs, string.utfvalues, string.format -- 0 EF BB BF UTF-8 -- 1 FF FE UTF-16-little-endian -- 2 FE FF UTF-16-big-endian -- 3 FF FE 00 00 UTF-32-little-endian -- 4 00 00 FE FF UTF-32-big-endian unicode.utfname = { [0] = 'utf-8', [1] = 'utf-16-le', [2] = 'utf-16-be', [3] = 'utf-32-le', [4] = 'utf-32-be' } -- \000 fails in <= 5.0 but is valid in >=5.1 where %z is depricated function unicode.utftype(f) local str = f:read(4) if not str then f:seek('set') return 0 -- elseif find(str,"^%z%z\254\255") then -- depricated -- elseif find(str,"^\000\000\254\255") then -- not permitted and bugged elseif find(str,"\000\000\254\255",1,true) then -- seems to work okay (TH) return 4 -- elseif find(str,"^\255\254%z%z") then -- depricated -- elseif find(str,"^\255\254\000\000") then -- not permitted and bugged elseif find(str,"\255\254\000\000",1,true) then -- seems to work okay (TH) return 3 elseif find(str,"^\254\255") then f:seek('set',2) return 2 elseif find(str,"^\255\254") then f:seek('set',2) return 1 elseif find(str,"^\239\187\191") then f:seek('set',3) return 0 else f:seek('set') return 0 end end function unicode.utf16_to_utf8(str, endian) -- maybe a gsub is faster or an lpeg local result, tmp, n, m, p = { }, { }, 0, 0, 0 -- lf | cr | crlf / (cr:13, lf:10) local function doit() if n == 10 then if p ~= 13 then result[#result+1] = concat(tmp) tmp = { } p = 0 end elseif n == 13 then result[#result+1] = concat(tmp) tmp = { } p = n else tmp[#tmp+1] = utfchar(n) p = 0 end end for l,r in bytepairs(str) do if r then if endian then n = l*256 + r else n = r*256 + l end if m > 0 then n = (m-0xD800)*0x400 + (n-0xDC00) + 0x10000 m = 0 doit() elseif n >= 0xD800 and n <= 0xDBFF then m = n else doit() end end end if #tmp > 0 then result[#result+1] = concat(tmp) end return result end function unicode.utf32_to_utf8(str, endian) local result = { } local tmp, n, m, p = { }, 0, -1, 0 -- lf | cr | crlf / (cr:13, lf:10) local function doit() if n == 10 then if p ~= 13 then result[#result+1] = concat(tmp) tmp = { } p = 0 end elseif n == 13 then result[#result+1] = concat(tmp) tmp = { } p = n else tmp[#tmp+1] = utfchar(n) p = 0 end end for a,b in bytepairs(str) do if a and b then if m < 0 then if endian then m = a*256*256*256 + b*256*256 else m = b*256 + a end else if endian then n = m + a*256 + b else n = m + b*256*256*256 + a*256*256 end m = -1 doit() end else break end end if #tmp > 0 then result[#result+1] = concat(tmp) end return result end local function little(c) local b = byte(c) -- b = c:byte() if b < 0x10000 then return char(b%256,b/256) else b = b - 0x10000 local b1, b2 = b/1024 + 0xD800, b%1024 + 0xDC00 return char(b1%256,b1/256,b2%256,b2/256) end end local function big(c) local b = byte(c) if b < 0x10000 then return char(b/256,b%256) else b = b - 0x10000 local b1, b2 = b/1024 + 0xD800, b%1024 + 0xDC00 return char(b1/256,b1%256,b2/256,b2%256) end end function unicode.utf8_to_utf16(str,littleendian) if littleendian then return char(255,254) .. utfgsub(str,".",little) else return char(254,255) .. utfgsub(str,".",big) end end function unicode.utfcodes(str) local t = { } for k,v in string.utfvalues(str) do t[#t+1] = format("0x%04X",k) end return concat(t,separator or " ") end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-math'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local floor, sin, cos, tan = math.floor, math.sin, math.cos, math.tan if not math.round then function math.round(x) return floor(x + 0.5) end end if not math.div then function math.div(n,m) return floor(n/m) end end if not math.mod then function math.mod(n,m) return n % m end end local pipi = 2*math.pi/360 function math.sind(d) return sin(d*pipi) end function math.cosd(d) return cos(d*pipi) end function math.tand(d) return tan(d*pipi) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-utils'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- hm, quite unreadable local gsub, format = string.gsub, string.format local concat = table.concat local type, next = type, next if not utils then utils = { } end if not utils.merger then utils.merger = { } end if not utils.lua then utils.lua = { } end utils.report = utils.report or print local merger = utils.merger merger.strip_comment = true local m_begin_merge = "begin library merge" local m_end_merge = "end library merge" local m_begin_closure = "do -- create closure to overcome 200 locals limit" local m_end_closure = "end -- of closure" local m_pattern = "%c+" .. "%-%-%s+" .. m_begin_merge .. "%c+(.-)%c+" .. "%-%-%s+" .. m_end_merge .. "%c+" local m_format = "\n\n-- " .. m_begin_merge .. "\n%s\n" .. "-- " .. m_end_merge .. "\n\n" local m_faked = "-- " .. "created merged file" .. "\n\n" .. "-- " .. m_begin_merge .. "\n\n" .. "-- " .. m_end_merge .. "\n\n" local function self_fake() return m_faked end local function self_nothing() return "" end local function self_load(name) local data = io.loaddata(name) or "" if data == "" then utils.report("merge: unknown file %s",name) else utils.report("merge: inserting %s",name) end return data or "" end local function self_save(name, data) if data ~= "" then if merger.strip_comment then -- saves some 20K local n = #data data = gsub(data,"%-%-~[^\n\r]*[\r\n]","") utils.report("merge: %s bytes of comment stripped, %s bytes of code left",n-#data,#data) end io.savedata(name,data) utils.report("merge: saving %s",name) end end local function self_swap(data,code) return data ~= "" and (gsub(data,m_pattern, function() return format(m_format,code) end, 1)) or "" end local function self_libs(libs,list) local result, f, frozen, foundpath = { }, nil, false, nil result[#result+1] = "\n" if type(libs) == 'string' then libs = { libs } end if type(list) == 'string' then list = { list } end for i=1,#libs do local lib = libs[i] for j=1,#list do local pth = gsub(list[j],"\\","/") -- file.clean_path utils.report("merge: checking library path %s",pth) local name = pth .. "/" .. lib if lfs.isfile(name) then foundpath = pth end end if foundpath then break end end if foundpath then utils.report("merge: using library path %s",foundpath) local right, wrong = { }, { } for i=1,#libs do local lib = libs[i] local fullname = foundpath .. "/" .. lib if lfs.isfile(fullname) then utils.report("merge: using library %s",fullname) right[#right+1] = lib result[#result+1] = m_begin_closure result[#result+1] = io.loaddata(fullname,true) result[#result+1] = m_end_closure else utils.report("merge: skipping library %s",fullname) wrong[#wrong+1] = lib end end if #right > 0 then utils.report("merge: used libraries: %s",concat(right," ")) end if #wrong > 0 then utils.report("merge: skipped libraries: %s",concat(wrong," ")) end else utils.report("merge: no valid library path found") end return concat(result, "\n\n") end function merger.selfcreate(libs,list,target) if target then self_save(target,self_swap(self_fake(),self_libs(libs,list))) end end function merger.selfmerge(name,libs,list,target) self_save(target or name,self_swap(self_load(name),self_libs(libs,list))) end function merger.selfclean(name) self_save(name,self_swap(self_load(name),self_nothing())) end function utils.lua.compile(luafile,lucfile,cleanup,strip) -- defaults: cleanup=false strip=true utils.report("lua: compiling %s into %s",luafile,lucfile) os.remove(lucfile) local command = "-o " .. string.quote(lucfile) .. " " .. string.quote(luafile) if strip ~= false then command = "-s " .. command end local done = os.spawn("texluac " .. command) == 0 or os.spawn("luac " .. command) == 0 if done and cleanup == true and lfs.isfile(lucfile) and lfs.isfile(luafile) then utils.report("lua: removing %s",luafile) os.remove(luafile) end return done end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['l-aux'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- for inline, no store split : for s in string.gmatch(str,",* *([^,]+)") do .. end aux = aux or { } local concat, format, gmatch = table.concat, string.format, string.gmatch local tostring, type = tostring, type local lpegmatch = lpeg.match local P, R, V = lpeg.P, lpeg.R, lpeg.V local escape, left, right = P("\\"), P('{'), P('}') lpeg.patterns.balanced = P { [1] = ((escape * (left+right)) + (1 - (left+right)) + V(2))^0, [2] = left * V(1) * right } local space = lpeg.P(' ') local equal = lpeg.P("=") local comma = lpeg.P(",") local lbrace = lpeg.P("{") local rbrace = lpeg.P("}") local nobrace = 1 - (lbrace+rbrace) local nested = lpeg.P { lbrace * (nobrace + lpeg.V(1))^0 * rbrace } local spaces = space^0 local value = lpeg.P(lbrace * lpeg.C((nobrace + nested)^0) * rbrace) + lpeg.C((nested + (1-comma))^0) local key = lpeg.C((1-equal-comma)^1) local pattern_a = (space+comma)^0 * (key * equal * value + key * lpeg.C("")) local pattern_c = (space+comma)^0 * (key * equal * value) local key = lpeg.C((1-space-equal-comma)^1) local pattern_b = spaces * comma^0 * spaces * (key * ((spaces * equal * spaces * value) + lpeg.C(""))) -- "a=1, b=2, c=3, d={a{b,c}d}, e=12345, f=xx{a{b,c}d}xx, g={}" : outer {} removes, leading spaces ignored local hash = { } local function set(key,value) -- using Carg is slower here hash[key] = value end local pattern_a_s = (pattern_a/set)^1 local pattern_b_s = (pattern_b/set)^1 local pattern_c_s = (pattern_c/set)^1 aux.settings_to_hash_pattern_a = pattern_a_s aux.settings_to_hash_pattern_b = pattern_b_s aux.settings_to_hash_pattern_c = pattern_c_s function aux.make_settings_to_hash_pattern(set,how) if how == "strict" then return (pattern_c/set)^1 elseif how == "tolerant" then return (pattern_b/set)^1 else return (pattern_a/set)^1 end end function aux.settings_to_hash(str,existing) if str and str ~= "" then hash = existing or { } lpegmatch(pattern_a_s,str) return hash else return { } end end function aux.settings_to_hash_tolerant(str,existing) if str and str ~= "" then hash = existing or { } lpegmatch(pattern_b_s,str) return hash else return { } end end function aux.settings_to_hash_strict(str,existing) if str and str ~= "" then hash = existing or { } lpegmatch(pattern_c_s,str) return next(hash) and hash else return nil end end local separator = comma * space^0 local value = lpeg.P(lbrace * lpeg.C((nobrace + nested)^0) * rbrace) + lpeg.C((nested + (1-comma))^0) local pattern = lpeg.Ct(value*(separator*value)^0) -- "aap, {noot}, mies" : outer {} removes, leading spaces ignored aux.settings_to_array_pattern = pattern -- we could use a weak table as cache function aux.settings_to_array(str) if not str or str == "" then return { } else return lpegmatch(pattern,str) end end local function set(t,v) t[#t+1] = v end local value = lpeg.P(lpeg.Carg(1)*value) / set local pattern = value*(separator*value)^0 * lpeg.Carg(1) function aux.add_settings_to_array(t,str) return lpegmatch(pattern,str,nil,t) end function aux.hash_to_string(h,separator,yes,no,strict,omit) if h then local t, s = { }, table.sortedkeys(h) omit = omit and table.tohash(omit) for i=1,#s do local key = s[i] if not omit or not omit[key] then local value = h[key] if type(value) == "boolean" then if yes and no then if value then t[#t+1] = key .. '=' .. yes elseif not strict then t[#t+1] = key .. '=' .. no end elseif value or not strict then t[#t+1] = key .. '=' .. tostring(value) end else t[#t+1] = key .. '=' .. value end end end return concat(t,separator or ",") else return "" end end function aux.array_to_string(a,separator) if a then return concat(a,separator or ",") else return "" end end function aux.settings_to_set(str,t) -- tohash? t = t or { } for s in gmatch(str,"%s*([^, ]+)") do -- space added t[s] = true end return t end function aux.simple_hash_to_string(h, separator) local t = { } for k, v in table.sortedhash(h) do if v then t[#t+1] = k end end return concat(t,separator or ",") end local value = lbrace * lpeg.C((nobrace + nested)^0) * rbrace local pattern = lpeg.Ct((space + value)^0) function aux.arguments_to_table(str) return lpegmatch(pattern,str) end -- temporary here function aux.getparameters(self,class,parentclass,settings) local sc = self[class] if not sc then sc = table.clone(self[parent]) self[class] = sc end aux.settings_to_hash(settings,sc) end -- temporary here local digit = lpeg.R("09") local period = lpeg.P(".") local zero = lpeg.P("0") local trailingzeros = zero^0 * -digit -- suggested by Roberto R local case_1 = period * trailingzeros / "" local case_2 = period * (digit - trailingzeros)^1 * (trailingzeros / "") local number = digit^1 * (case_1 + case_2) local stripper = lpeg.Cs((number + 1)^0) lpeg.patterns.strip_zeros = stripper function aux.strip_zeros(str) return lpegmatch(stripper,str) end function aux.definetable(target) -- defines undefined tables local composed, t = nil, { } for name in gmatch(target,"([^%.]+)") do if composed then composed = composed .. "." .. name else composed = name end t[#t+1] = format("%s = %s or { }",composed,composed) end return concat(t,"\n") end function aux.accesstable(target) local t = _G for name in gmatch(target,"([^%.]+)") do t = t[name] end return t end -- as we use this a lot ... function aux.formatcolumns(result,between) if result and #result > 0 then between = between or " " local widths, numbers = { }, { } local first = result[1] local n = #first for i=1,n do widths[i] = 0 end for i=1,#result do local r = result[i] for j=1,n do local rj = r[j] local tj = type(rj) if tj == "number" then numbers[j] = true end if tj ~= "string" then rj = tostring(rj) r[j] = rj end local w = #rj if w > widths[j] then widths[j] = w end end end for i=1,n do local w = widths[i] if numbers[i] then if w > 80 then widths[i] = "%s" .. between else widths[i] = "%0" .. w .. "i" .. between end else if w > 80 then widths[i] = "%s" .. between elseif w > 0 then widths[i] = "%-" .. w .. "s" .. between else widths[i] = "%s" end end end local template = string.strip(concat(widths)) for i=1,#result do local str = format(template,unpack(result[i])) result[i] = string.strip(str) end end return result end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['trac-inf'] = { version = 1.001, comment = "companion to trac-inf.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- As we want to protect the global tables, we no longer store the timing -- in the tables themselves but in a hidden timers table so that we don't -- get warnings about assignments. This is more efficient than using rawset -- and rawget. local format = string.format local clock = os.gettimeofday or os.clock -- should go in environment local statusinfo, n, registered = { }, 0, { } statistics = statistics or { } statistics.enable = true statistics.threshold = 0.05 local timers = { } local function hastiming(instance) return instance and timers[instance] end local function resettiming(instance) timers[instance or "notimer"] = { timing = 0, loadtime = 0 } end local function starttiming(instance) local timer = timers[instance or "notimer"] if not timer then timer = { } timers[instance or "notimer"] = timer end local it = timer.timing if not it then it = 0 end if it == 0 then timer.starttime = clock() if not timer.loadtime then timer.loadtime = 0 end end timer.timing = it + 1 end local function stoptiming(instance, report) local timer = timers[instance or "notimer"] local it = timer.timing if it > 1 then timer.timing = it - 1 else local starttime = timer.starttime if starttime then local stoptime = clock() local loadtime = stoptime - starttime timer.stoptime = stoptime timer.loadtime = timer.loadtime + loadtime if report then statistics.report("load time %0.3f",loadtime) end timer.timing = 0 return loadtime end end return 0 end local function elapsedtime(instance) local timer = timers[instance or "notimer"] return format("%0.3f",timer and timer.loadtime or 0) end local function elapsedindeed(instance) local timer = timers[instance or "notimer"] return (timer and timer.loadtime or 0) > statistics.threshold end local function elapsedseconds(instance,rest) -- returns nil if 0 seconds if elapsedindeed(instance) then return format("%s seconds %s", elapsedtime(instance),rest or "") end end statistics.hastiming = hastiming statistics.resettiming = resettiming statistics.starttiming = starttiming statistics.stoptiming = stoptiming statistics.elapsedtime = elapsedtime statistics.elapsedindeed = elapsedindeed statistics.elapsedseconds = elapsedseconds -- general function function statistics.register(tag,fnc) if statistics.enable and type(fnc) == "function" then local rt = registered[tag] or (#statusinfo + 1) statusinfo[rt] = { tag, fnc } registered[tag] = rt if #tag > n then n = #tag end end end function statistics.show(reporter) if statistics.enable then if not reporter then reporter = function(tag,data,n) texio.write_nl(tag .. " " .. data) end end -- this code will move local register = statistics.register register("luatex banner", function() return string.lower(status.banner) end) register("control sequences", function() return format("%s of %s", status.cs_count, status.hash_size+status.hash_extra) end) register("callbacks", function() local total, indirect = status.callbacks or 0, status.indirect_callbacks or 0 return format("direct: %s, indirect: %s, total: %s", total-indirect, indirect, total) end) register("current memory usage", statistics.memused) register("runtime",statistics.runtime) for i=1,#statusinfo do local s = statusinfo[i] local r = s[2]() if r then reporter(s[1],r,n) end end texio.write_nl("") -- final newline statistics.enable = false end end function statistics.show_job_stat(tag,data,n) if type(data) == "table" then for i=1,#data do statistics.show_job_stat(tag,data[i],n) end else texio.write_nl(format("%-15s: %s - %s","mkiv lua stats",tag:rpadd(n," "),data)) end end function statistics.memused() -- no math.round yet -) local round = math.round or math.floor return format("%s MB (ctx: %s MB)",round(collectgarbage("count")/1000), round(status.luastate_bytes/1000000)) end starttiming(statistics) function statistics.formatruntime(runtime) -- indirect so it can be overloaded and return format("%s seconds", runtime) -- indeed that happens in cure-uti.lua end function statistics.runtime() stoptiming(statistics) return statistics.formatruntime(elapsedtime(statistics)) end function statistics.timed(action,report) report = report or logs.simple starttiming("run") action() stoptiming("run") report("total runtime: %s",elapsedtime("run")) end -- where, not really the best spot for this: commands = commands or { } function commands.resettimer(name) resettiming(name or "whatever") starttiming(name or "whatever") end function commands.elapsedtime(name) stoptiming(name or "whatever") tex.sprint(elapsedtime(name or "whatever")) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['trac-set'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local type, next, tostring = type, next, tostring local concat = table.concat local format, find, lower, gsub, simpleesc = string.format, string.find, string.lower, string.gsub, string.simpleesc local is_boolean = string.is_boolean setters = { } local data = { } -- maybe just local -- We can initialize from the cnf file. This is sort of tricky as -- laster defined setters also need to be initialized then. If set -- this way, we need to ensure that they are not reset later on. local trace_initialize = false local function report(what,filename,name,key,value) texio.write_nl(format("%s setter, filename: %s, name: %s, key: %s, value: %s",what,filename,name,key,value)) end function setters.initialize(filename,name,values) -- filename only for diagnostics local data = data[name] if data then data = data.data if data then for key, value in next, values do key = gsub(key,"_",".") value = is_boolean(value,value) local functions = data[key] if functions then if #functions > 0 and not functions.value then if trace_initialize then report("doing",filename,name,key,value) end for i=1,#functions do functions[i](value) end functions.value = value else if trace_initialize then report("skipping",filename,name,key,value) end end else -- we do a simple preregistration i.e. not in the -- list as it might be an obsolete entry functions = { default = value } data[key] = functions if trace_initialize then report("storing",filename,name,key,value) end end end end end end -- user interface code local function set(t,what,newvalue) local data, done = t.data, t.done if type(what) == "string" then what = aux.settings_to_hash(what) -- inefficient but ok end for w, value in next, what do if value == "" then value = newvalue elseif not value then value = false -- catch nil else value = is_boolean(value,value) end for name, functions in next, data do if done[name] then -- prevent recursion due to wildcards elseif find(name,simpleesc(w)) then done[name] = true for i=1,#functions do functions[i](value) end functions.value = value end end end end local function reset(t) for name, functions in next, t.data do for i=1,#functions do functions[i](false) end functions.value = false end end local function enable(t,what) set(t,what,true) end local function disable(t,what) local data = t.data if not what or what == "" then t.done = { } reset(t) else set(t,what,false) end end function setters.register(t,what,...) local data = t.data what = lower(what) local functions = data[what] if not functions then functions = { } data[what] = functions end local default = functions.default -- can be set from cnf file for _, fnc in next, { ... } do local typ = type(fnc) if typ == "string" then local s = fnc -- else wrong reference fnc = function(value) set(t,s,value) end elseif typ ~= "function" then fnc = nil end if fnc then functions[#functions+1] = fnc if default then fnc(default) functions.value = default end end end end function setters.enable(t,what) local e = t.enable t.enable, t.done = enable, { } enable(t,what) t.enable, t.done = e, { } end function setters.disable(t,what) local e = t.disable t.disable, t.done = disable, { } disable(t,what) t.disable, t.done = e, { } end function setters.reset(t) t.done = { } reset(t) end function setters.list(t) -- pattern local list = table.sortedkeys(t.data) local user, system = { }, { } for l=1,#list do local what = list[l] if find(what,"^%*") then system[#system+1] = what else user[#user+1] = what end end return user, system end function setters.show(t) commands.writestatus("","") local list = setters.list(t) local category = t.name for k=1,#list do local name = list[k] local functions = t.data[name] if functions then local value, default, modules = functions.value, functions.default, #functions value = value == nil and "unset" or tostring(value) default = default == nil and "unset" or tostring(default) commands.writestatus(category,format("%-25s modules: %2i default: %5s value: %5s",name,modules,default,value)) end end commands.writestatus("","") end -- we could have used a bit of oo and the trackers:enable syntax but -- there is already a lot of code around using the singular tracker -- we could make this into a module function setters.new(name) local t t = { data = { }, -- indexed, but also default and value fields name = name, enable = function(...) setters.enable (t,...) end, disable = function(...) setters.disable (t,...) end, register = function(...) setters.register(t,...) end, list = function(...) setters.list (t,...) end, show = function(...) setters.show (t,...) end, } data[name] = t return t end trackers = setters.new("trackers") directives = setters.new("directives") experiments = setters.new("experiments") -- experiment if trackers and environment and environment.engineflags.trackers then trackers.enable(environment.engineflags.trackers) end if directives and environment and environment.engineflags.directives then directives.enable(environment.engineflags.directives) end -- nice trick: we overload two of the directives related functions with variants that -- do tracing (itself using a tracker) .. proof of concept local function report(...) -- messy .. chicken or egg local p = (commands and commands.writestatus) or (logs and logs.report) if p then p(...) end end local trace_directives = false local trace_directives = false trackers.register("system.directives", function(v) trace_directives = v end) local trace_experiments = false local trace_experiments = false trackers.register("system.experiments", function(v) trace_experiments = v end) local enable = directives.enable local disable = directives.disable function directives.enable(...) report("directives","enabling: %s",concat({...}," ")) enable(...) end function directives.disable(...) report("directives","disabling: %s",concat({...}," ")) disable(...) end local enable = experiments.enable local disable = experiments.disable function experiments.enable(...) report("experiments","enabling: %s",concat({...}," ")) enable(...) end function experiments.disable(...) report("experiments","disabling: %s",concat({...}," ")) disable(...) end -- a useful example directives.register("system.nostatistics", function(v) statistics.enable = not v end) -- experiment if trackers and environment and environment.engineflags.trackers then trackers.enable(environment.engineflags.trackers) end if directives and environment and environment.engineflags.directives then directives.enable(environment.engineflags.directives) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['trac-tra'] = { version = 1.001, comment = "companion to trac-tra.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- the tag is kind of generic and used for functions that are not -- bound to a variable, like node.new, node.copy etc (contrary to for instance -- node.has_attribute which is bound to a has_attribute local variable in mkiv) local debug = require "debug" local getinfo = debug.getinfo local type, next = type, next local format, find = string.format, string.find local is_boolean = string.is_boolean debugger = debugger or { } local counters = { } local names = { } -- one local function hook() local f = getinfo(2,"f").func local n = getinfo(2,"Sn") -- if n.what == "C" and n.name then print (n.namewhat .. ': ' .. n.name) end if f then local cf = counters[f] if cf == nil then counters[f] = 1 names[f] = n else counters[f] = cf + 1 end end end local function getname(func) local n = names[func] if n then if n.what == "C" then return n.name or '' else -- source short_src linedefined what name namewhat nups func local name = n.name or n.namewhat or n.what if not name or name == "" then name = "?" end return format("%s : %s : %s", n.short_src or "unknown source", n.linedefined or "--", name) end else return "unknown" end end function debugger.showstats(printer,threshold) printer = printer or texio.write or print threshold = threshold or 0 local total, grandtotal, functions = 0, 0, 0 printer("\n") -- ugly but ok -- table.sort(counters) for func, count in next, counters do if count > threshold then local name = getname(func) if not find(name,"for generator") then printer(format("%8i %s", count, name)) total = total + count end end grandtotal = grandtotal + count functions = functions + 1 end printer(format("functions: %s, total: %s, grand total: %s, threshold: %s\n", functions, total, grandtotal, threshold)) end -- two -- rest function debugger.savestats(filename,threshold) local f = io.open(filename,'w') if f then debugger.showstats(function(str) f:write(str) end,threshold) f:close() end end function debugger.enable() debug.sethook(hook,"c") end function debugger.disable() debug.sethook() end local function trace_calls(n) debugger.enable() luatex.register_stop_actions(function() debugger.disable() debugger.savestats(tex.jobname .. "-luacalls.log",tonumber(n)) end) trace_calls = function() end end if directives then directives.register("system.tracecalls", function(n) trace_calls(n) end) -- indirect is needed for nilling end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['trac-log'] = { version = 1.001, comment = "companion to trac-log.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- xml logging is only usefull in normal runs, not in ini mode -- it looks like some tex logging (like filenames) is broken (no longer -- interceoted at the tex end so the xml variant is not that useable now) local write_nl, write = texio and texio.write_nl or print, texio and texio.write or io.write local format, gmatch = string.format, string.gmatch local texcount = tex and tex.count --[[ldx--

This is a prelude to a more extensive logging module. For the sake of parsing log files, in addition to the standard logging we will provide an structured file. Actually, any logging that is hooked into callbacks will be \XML\ by default.

--ldx]]-- logs = logs or { } --[[ldx--

This looks pretty ugly but we need to speed things up a bit.

--ldx]]-- local moreinfo = [[ More information about ConTeXt and the tools that come with it can be found at: maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net wiki : http://contextgarden.net ]] local functions = { 'report', 'status', 'start', 'stop', 'push', 'pop', 'line', 'direct', 'start_run', 'stop_run', 'start_page_number', 'stop_page_number', 'report_output_pages', 'report_output_log', 'report_tex_stat', 'report_job_stat', 'show_open', 'show_close', 'show_load', 'dummy', } local method = "nop" function logs.set_method(newmethod) method = newmethod -- a direct copy might be faster but let's try this for a while setmetatable(logs, { __index = logs[method] }) end function logs.get_method() return method end -- installer local data = { } function logs.new(category) local logger = data[category] if not logger then logger = function(...) logs.report(category,...) end data[category] = logger end return logger end -- nop logging (maybe use __call instead) local noplog = { } logs.nop = noplog setmetatable(logs, { __index = noplog }) for i=1,#functions do noplog[functions[i]] = function() end end -- tex logging local texlog = { } logs.tex = texlog setmetatable(texlog, { __index = noplog }) function texlog.report(a,b,c,...) if c then write_nl(format("%-16s> %s\n",a,format(b,c,...))) elseif b then write_nl(format("%-16s> %s\n",a,b)) else write_nl(format("%-16s>\n",a)) end end function texlog.status(a,b,c,...) if c then write_nl(format("%-16s: %s\n",a,format(b,c,...))) elseif b then write_nl(format("%-16s: %s\n",a,b)) -- b can have %'s else write_nl(format("%-16s:>\n",a)) end end function texlog.line(fmt,...) -- new if fmt then write_nl(format(fmt,...)) else write_nl("") end end local real, user, sub function texlog.start_page_number() real, user, sub = texcount.realpageno, texcount.userpageno, texcount.subpageno end local report_pages = logs.new("pages") -- not needed but saves checking when we grep for it function texlog.stop_page_number() if real > 0 then if user > 0 then if sub > 0 then report_pages("flushing realpage %s, userpage %s, subpage %s",real,user,sub) else report_pages("flushing realpage %s, userpage %s",real,user) end else report_pages("flushing realpage %s",real) end else report_pages("flushing page") end io.flush() end texlog.report_job_stat = statistics and statistics.show_job_stat -- xml logging local xmllog = { } logs.xml = xmllog setmetatable(xmllog, { __index = noplog }) function xmllog.report(category,fmt,s,...) -- new if s then write_nl(format("%s",category,format(fmt,s,...))) elseif fmt then write_nl(format("%s",category,fmt)) else write_nl(format("",category)) end end function xmllog.status(category,fmt,s,...) if s then write_nl(format("%s",category,format(fmt,s,...))) elseif fmt then write_nl(format("%s",category,fmt)) else write_nl(format("",category)) end end function xmllog.line(fmt,...) -- new if fmt then write_nl(format("%s",format(fmt,...))) else write_nl("") end end function xmllog.start() write_nl("<%s>" ) end function xmllog.stop () write_nl("") end function xmllog.push () write_nl("" ) end function xmllog.start_run() write_nl("") write_nl("") -- xmlns='www.pragma-ade.com/luatex/schemas/context-job.rng' write_nl("") end function xmllog.stop_run() write_nl("") end function xmllog.start_page_number() write_nl(format("

") write_nl("") end function xmllog.report_output_pages(p,b) write_nl(format("", p)) write_nl(format("", b)) write_nl("") end function xmllog.report_output_log() -- nothing end function xmllog.report_tex_stat(k,v) write_nl("log",""..tostring(v).."") end local nesting = 0 function xmllog.show_open(name) nesting = nesting + 1 write_nl(format("",nesting,name)) end function xmllog.show_close(name) write(" ") nesting = nesting - 1 end function xmllog.show_load(name) write_nl(format("",nesting+1,name)) end -- initialization if tex and (tex.jobname or tex.formatname) then -- todo: this can be set in mtxrun ... or maybe we should just forget about this alternative format if (os.getenv("mtx.directives.logmethod") or os.getenv("mtx_directives_logmethod")) == "xml" then logs.set_method('xml') else logs.set_method('tex') end else logs.set_method('nop') end -- logging in runners -> these are actually the nop loggers local name, banner = 'report', 'context' function noplog.report(category,fmt,...) -- todo: fmt,s if fmt then write_nl(format("%s | %s: %s",name,category,format(fmt,...))) elseif category then write_nl(format("%s | %s",name,category)) else write_nl(format("%s |",name)) end end noplog.status = noplog.report -- just to be sure, never used function noplog.simple(fmt,...) -- todo: fmt,s if fmt then write_nl(format("%s | %s",name,format(fmt,...))) else write_nl(format("%s |",name)) end end if utils then utils.report = function(...) logs.simple(...) end end function logs.setprogram(newname,newbanner) name, banner = newname, newbanner end function logs.extendbanner(newbanner) banner = banner .. " | ".. newbanner end function logs.reportlines(str) -- todo: for line in gmatch(str,"(.-)[\n\r]") do logs.report(line) end end function logs.reportline() -- for scripts too logs.report() end function logs.simpleline() logs.report() end function logs.simplelines(str) -- todo: for line in gmatch(str,"(.-)[\n\r]") do logs.simple(line) end end function logs.reportbanner() -- for scripts too logs.report(banner) end function logs.help(message,option) logs.reportbanner() logs.reportline() logs.reportlines(message) if option ~= "nomoreinfo" then logs.reportline() logs.reportlines(moreinfo) end end -- logging to a file function logs.system(whereto,process,jobname,category,...) local message = format("%s %s => %s => %s => %s\r",os.date("%d/%m/%y %H:%m:%S"),process,jobname,category,format(...)) for i=1,10 do local f = io.open(whereto,"a") if f then f:write(message) f:close() break else sleep(0.1) end end end -- bonus function logs.fatal(where,...) logs.report(where,"fatal error: %s, aborting now",format(...)) os.exit() end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['trac-pro'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local getmetatable, setmetatable, rawset, type = getmetatable, setmetatable, rawset, type -- The protection implemented here is probably not that tight but good enough to catch -- problems due to naive usage. -- -- There's a more extensive version (trac-xxx.lua) that supports nesting. -- -- This will change when we have _ENV in lua 5.2+ local trace_namespaces = false trackers.register("system.namespaces", function(v) trace_namespaces = v end) local report_system = logs.new("system") namespaces = { } local registered = { } local function report_index(k,name) if trace_namespaces then report_system("reference to '%s' in protected namespace '%s', %s",k,name,debug.traceback()) else report_system("reference to '%s' in protected namespace '%s'",k,name) end end local function report_newindex(k,name) if trace_namespaces then report_system("assignment to '%s' in protected namespace '%s', %s",k,name,debug.traceback()) else report_system("assignment to '%s' in protected namespace '%s'",k,name) end end local function register(name) local data = name == "global" and _G or _G[name] if not data then return -- error end registered[name] = data local m = getmetatable(data) if not m then m = { } setmetatable(data,m) end local index, newindex = { }, { } m.__saved__index = m.__index m.__no__index = function(t,k) if not index[k] then index[k] = true report_index(k,name) end return nil end m.__saved__newindex = m.__newindex m.__no__newindex = function(t,k,v) if not newindex[k] then newindex[k] = true report_newindex(k,name) end rawset(t,k,v) end m.__protection__depth = 0 end local function private(name) -- maybe save name local data = registered[name] if not data then data = _G[name] if not data then data = { } _G[name] = data end register(name) end return data end local function protect(name) local data = registered[name] if not data then return end local m = getmetatable(data) local pd = m.__protection__depth if pd > 0 then m.__protection__depth = pd + 1 else m.__save_d_index, m.__saved__newindex = m.__index, m.__newindex m.__index, m.__newindex = m.__no__index, m.__no__newindex m.__protection__depth = 1 end end local function unprotect(name) local data = registered[name] if not data then return end local m = getmetatable(data) local pd = m.__protection__depth if pd > 1 then m.__protection__depth = pd - 1 else m.__index, m.__newindex = m.__saved__index, m.__saved__newindex m.__protection__depth = 0 end end local function protectall() for name, _ in next, registered do if name ~= "global" then protect(name) end end end local function unprotectall() for name, _ in next, registered do if name ~= "global" then unprotect(name) end end end namespaces.register = register -- register when defined namespaces.private = private -- allocate and register if needed namespaces.protect = protect namespaces.unprotect = unprotect namespaces.protectall = protectall namespaces.unprotectall = unprotectall namespaces.private("namespaces") registered = { } register("global") -- unreachable directives.register("system.protect", function(v) if v then protectall() else unprotectall() end end) directives.register("system.checkglobals", function(v) if v then report_system("enabling global namespace guard") protect("global") else report_system("disabling global namespace guard") unprotect("global") end end) -- dummy section (will go to luat-dum.lua) end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['luat-env'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- A former version provided functionality for non embeded core -- scripts i.e. runtime library loading. Given the amount of -- Lua code we use now, this no longer makes sense. Much of this -- evolved before bytecode arrays were available and so a lot of -- code has disappeared already. local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") local format, sub, match, gsub, find = string.format, string.sub, string.match, string.gsub, string.find local unquote, quote = string.unquote, string.quote -- precautions os.setlocale(nil,nil) -- useless feature and even dangerous in luatex function os.setlocale() -- no way you can mess with it end -- dirty tricks if arg and (arg[0] == 'luatex' or arg[0] == 'luatex.exe') and arg[1] == "--luaonly" then arg[-1] = arg[0] arg[ 0] = arg[2] for k=3,#arg do arg[k-2] = arg[k] end arg[#arg] = nil -- last arg[#arg] = nil -- pre-last end -- environment environment = environment or { } environment.arguments = { } environment.files = { } environment.sortedflags = nil local mt = { __index = function(_,k) if k == "version" then local version = tex.toks and tex.toks.contextversiontoks if version and version ~= "" then rawset(environment,"version",version) return version else return "unknown" end elseif k == "jobname" or k == "formatname" then local name = tex and tex[k] if name or name== "" then rawset(environment,k,name) return name else return "unknown" end elseif k == "outputfilename" then local name = environment.jobname rawset(environment,k,name) return name end end } setmetatable(environment,mt) function environment.initialize_arguments(arg) local arguments, files = { }, { } environment.arguments, environment.files, environment.sortedflags = arguments, files, nil for index=1,#arg do local argument = arg[index] if index > 0 then local flag, value = match(argument,"^%-+(.-)=(.-)$") if flag then arguments[flag] = unquote(value or "") else flag = match(argument,"^%-+(.+)") if flag then arguments[flag] = true else files[#files+1] = argument end end end end environment.ownname = environment.ownname or arg[0] or 'unknown.lua' end function environment.setargument(name,value) environment.arguments[name] = value end -- todo: defaults, better checks e.g on type (boolean versus string) -- -- tricky: too many hits when we support partials unless we add -- a registration of arguments so from now on we have 'partial' function environment.argument(name,partial) local arguments, sortedflags = environment.arguments, environment.sortedflags if arguments[name] then return arguments[name] elseif partial then if not sortedflags then sortedflags = table.sortedkeys(arguments) for k=1,#sortedflags do sortedflags[k] = "^" .. sortedflags[k] end environment.sortedflags = sortedflags end -- example of potential clash: ^mode ^modefile for k=1,#sortedflags do local v = sortedflags[k] if find(name,v) then return arguments[sub(v,2,#v)] end end end return nil end function environment.split_arguments(separator) -- rather special, cut-off before separator local done, before, after = false, { }, { } local original_arguments = environment.original_arguments for k=1,#original_arguments do local v = original_arguments[k] if not done and v == separator then done = true elseif done then after[#after+1] = v else before[#before+1] = v end end return before, after end function environment.reconstruct_commandline(arg,noquote) arg = arg or environment.original_arguments if noquote and #arg == 1 then local a = arg[1] a = resolvers.resolve(a) a = unquote(a) return a elseif #arg > 0 then local result = { } for i=1,#arg do local a = arg[i] a = resolvers.resolve(a) a = unquote(a) a = gsub(a,'"','\\"') -- tricky if find(a," ") then result[#result+1] = quote(a) else result[#result+1] = a end end return table.join(result," ") else return "" end end if arg then -- new, reconstruct quoted snippets (maybe better just remove the " then and add them later) local newarg, instring = { }, false for index=1,#arg do local argument = arg[index] if find(argument,"^\"") then newarg[#newarg+1] = gsub(argument,"^\"","") if not find(argument,"\"$") then instring = true end elseif find(argument,"\"$") then newarg[#newarg] = newarg[#newarg] .. " " .. gsub(argument,"\"$","") instring = false elseif instring then newarg[#newarg] = newarg[#newarg] .. " " .. argument else newarg[#newarg+1] = argument end end for i=1,-5,-1 do newarg[i] = arg[i] end environment.initialize_arguments(newarg) environment.original_arguments = newarg environment.raw_arguments = arg arg = { } -- prevent duplicate handling end -- weird place ... depends on a not yet loaded module function environment.texfile(filename) return resolvers.find_file(filename,'tex') end function environment.luafile(filename) local resolved = resolvers.find_file(filename,'tex') or "" if resolved ~= "" then return resolved end resolved = resolvers.find_file(filename,'texmfscripts') or "" if resolved ~= "" then return resolved end return resolvers.find_file(filename,'luatexlibs') or "" end environment.loadedluacode = loadfile -- can be overloaded function environment.luafilechunk(filename,silent) -- used for loading lua bytecode in the format filename = file.replacesuffix(filename, "lua") local fullname = environment.luafile(filename) if fullname and fullname ~= "" then local data = environment.loadedluacode(fullname) if trace_locating then report_resolvers("loading file %s%s", fullname, not data and " failed" or "") elseif not silent then texio.write("<",data and "+ " or "- ",fullname,">") end return data else if trace_locating then report_resolvers("unknown file %s", filename) end return nil end end -- the next ones can use the previous ones / combine function environment.loadluafile(filename, version) local lucname, luaname, chunk local basename = file.removesuffix(filename) if basename == filename then lucname, luaname = basename .. ".luc", basename .. ".lua" else lucname, luaname = nil, basename -- forced suffix end -- when not overloaded by explicit suffix we look for a luc file first local fullname = (lucname and environment.luafile(lucname)) or "" if fullname ~= "" then if trace_locating then report_resolvers("loading %s", fullname) end chunk = loadfile(fullname) -- this way we don't need a file exists check end if chunk then assert(chunk)() if version then -- we check of the version number of this chunk matches local v = version -- can be nil if modules and modules[filename] then v = modules[filename].version -- new method elseif versions and versions[filename] then v = versions[filename] -- old method end if v == version then return true else if trace_locating then report_resolvers("version mismatch for %s: lua=%s, luc=%s", filename, v, version) end environment.loadluafile(filename) end else return true end end fullname = (luaname and environment.luafile(luaname)) or "" if fullname ~= "" then if trace_locating then report_resolvers("loading %s", fullname) end chunk = loadfile(fullname) -- this way we don't need a file exists check if not chunk then if trace_locating then report_resolvers("unknown file %s", filename) end else assert(chunk)() return true end end return false end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['lxml-tab'] = { version = 1.001, comment = "this module is the basis for the lxml-* ones", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- this module needs a cleanup: check latest lpeg, passing args, (sub)grammar, etc etc -- stripping spaces from e.g. cont-en.xml saves .2 sec runtime so it's not worth the -- trouble local trace_entities = false trackers.register("xml.entities", function(v) trace_entities = v end) local report_xml = logs.new("xml") --[[ldx--

The parser used here is inspired by the variant discussed in the lua book, but handles comment and processing instructions, has a different structure, provides parent access; a first version used different trickery but was less optimized to we went this route. First we had a find based parser, now we have an based one. The find based parser can be found in l-xml-edu.lua along with other older code.

Beware, the interface may change. For instance at, ns, tg, dt may get more verbose names. Once the code is stable we will also remove some tracing and optimize the code.

--ldx]]-- xml = xml or { } local concat, remove, insert = table.concat, table.remove, table.insert local type, next, setmetatable, getmetatable, tonumber = type, next, setmetatable, getmetatable, tonumber local format, lower, find, match, gsub = string.format, string.lower, string.find, string.match, string.gsub local utfchar = unicode.utf8.char local lpegmatch = lpeg.match local P, S, R, C, V, C, Cs = lpeg.P, lpeg.S, lpeg.R, lpeg.C, lpeg.V, lpeg.C, lpeg.Cs --[[ldx--

First a hack to enable namespace resolving. A namespace is characterized by a . The following function associates a namespace prefix with a pattern. We use , which in this case is more than twice as fast as a find based solution where we loop over an array of patterns. Less code and much cleaner.

--ldx]]-- xml.xmlns = xml.xmlns or { } local check = P(false) local parse = check --[[ldx--

The next function associates a namespace prefix with an . This normally happens independent of parsing.

xml.registerns("mml","mathml") --ldx]]-- function xml.registerns(namespace, pattern) -- pattern can be an lpeg check = check + C(P(lower(pattern))) / namespace parse = P { P(check) + 1 * V(1) } end --[[ldx--

The next function also registers a namespace, but this time we map a given namespace prefix onto a registered one, using the given . This used for attributes like xmlns:m.

xml.checkns("m","http://www.w3.org/mathml") --ldx]]-- function xml.checkns(namespace,url) local ns = lpegmatch(parse,lower(url)) if ns and namespace ~= ns then xml.xmlns[namespace] = ns end end --[[ldx--

Next we provide a way to turn an into a registered namespace. This used for the xmlns attribute.

resolvedns = xml.resolvens("http://www.w3.org/mathml") This returns mml. --ldx]]-- function xml.resolvens(url) return lpegmatch(parse,lower(url)) or "" end --[[ldx--

A namespace in an element can be remapped onto the registered one efficiently by using the xml.xmlns table.

--ldx]]-- --[[ldx--

This version uses . We follow the same approach as before, stack and top and such. This version is about twice as fast which is mostly due to the fact that we don't have to prepare the stream for cdata, doctype etc etc. This variant is is dedicated to Luigi Scarso, who challenged me with 40 megabyte files that took 12.5 seconds to load (1.5 for file io and the rest for tree building). With the implementation we got that down to less 7.3 seconds. Loading the 14 interface definition files (2.6 meg) went down from 1.05 seconds to 0.55.

Next comes the parser. The rather messy doctype definition comes in many disguises so it is no surprice that later on have to dedicate quite some code to it.

The code may look a bit complex but this is mostly due to the fact that we resolve namespaces and attach metatables. There is only one public function:

local x = xml.convert(somestring)

An optional second boolean argument tells this function not to create a root element.

Valid entities are:

--ldx]]-- -- not just one big nested table capture (lpeg overflow) local nsremap, resolvens = xml.xmlns, xml.resolvens local stack, top, dt, at, xmlns, errorstr, entities = { }, { }, { }, { }, { }, nil, { } local strip, cleanup, utfize, resolve, resolve_predefined, unify_predefined = false, false, false, false, false, false local dcache, hcache, acache = { }, { }, { } local mt = { } local function initialize_mt(root) mt = { __index = root } -- will be redefined later end function xml.setproperty(root,k,v) getmetatable(root).__index[k] = v end function xml.check_error(top,toclose) return "" end local function add_attribute(namespace,tag,value) if cleanup and #value > 0 then value = cleanup(value) -- new end if tag == "xmlns" then xmlns[#xmlns+1] = resolvens(value) at[tag] = value elseif namespace == "" then at[tag] = value elseif namespace == "xmlns" then xml.checkns(tag,value) at["xmlns:" .. tag] = value else -- for the moment this way: at[namespace .. ":" .. tag] = value end end local function add_empty(spacing, namespace, tag) if #spacing > 0 then dt[#dt+1] = spacing end local resolved = (namespace == "" and xmlns[#xmlns]) or nsremap[namespace] or namespace top = stack[#stack] dt = top.dt local t = { ns=namespace or "", rn=resolved, tg=tag, at=at, dt={}, __p__ = top } dt[#dt+1] = t setmetatable(t, mt) if at.xmlns then remove(xmlns) end at = { } end local function add_begin(spacing, namespace, tag) if #spacing > 0 then dt[#dt+1] = spacing end local resolved = (namespace == "" and xmlns[#xmlns]) or nsremap[namespace] or namespace top = { ns=namespace or "", rn=resolved, tg=tag, at=at, dt={}, __p__ = stack[#stack] } setmetatable(top, mt) dt = top.dt stack[#stack+1] = top at = { } end local function add_end(spacing, namespace, tag) if #spacing > 0 then dt[#dt+1] = spacing end local toclose = remove(stack) top = stack[#stack] if #stack < 1 then errorstr = format("nothing to close with %s %s", tag, xml.check_error(top,toclose) or "") elseif toclose.tg ~= tag then -- no namespace check errorstr = format("unable to close %s with %s %s", toclose.tg, tag, xml.check_error(top,toclose) or "") end dt = top.dt dt[#dt+1] = toclose -- dt[0] = top -- nasty circular reference when serializing table if toclose.at.xmlns then remove(xmlns) end end local function add_text(text) if cleanup and #text > 0 then dt[#dt+1] = cleanup(text) else dt[#dt+1] = text end end local function add_special(what, spacing, text) if #spacing > 0 then dt[#dt+1] = spacing end if strip and (what == "@cm@" or what == "@dt@") then -- forget it else dt[#dt+1] = { special=true, ns="", tg=what, dt={ text } } end end local function set_message(txt) errorstr = "garbage at the end of the file: " .. gsub(txt,"([ \n\r\t]*)","") end local reported_attribute_errors = { } local function attribute_value_error(str) if not reported_attribute_errors[str] then report_xml("invalid attribute value: %q",str) reported_attribute_errors[str] = true at._error_ = str end return str end local function attribute_specification_error(str) if not reported_attribute_errors[str] then report_xml("invalid attribute specification: %q",str) reported_attribute_errors[str] = true at._error_ = str end return str end function xml.unknown_dec_entity_format(str) return (str == "" and "&error;") or format("&%s;",str) end function xml.unknown_hex_entity_format(str) return format("&#x%s;",str) end function xml.unknown_any_entity_format(str) return format("&#x%s;",str) end local function fromhex(s) local n = tonumber(s,16) if n then return utfchar(n) else return format("h:%s",s), true end end local function fromdec(s) local n = tonumber(s) if n then return utfchar(n) else return format("d:%s",s), true end end -- one level expansion (simple case), no checking done local rest = (1-P(";"))^0 local many = P(1)^0 local parsedentity = P("&") * (P("#x")*(rest/fromhex) + P("#")*(rest/fromdec)) * P(";") * P(-1) + (P("#x")*(many/fromhex) + P("#")*(many/fromdec)) -- parsing in the xml file local predefined_unified = { [38] = "&", [42] = """, [47] = "'", [74] = "<", [76] = "&gr;", } local predefined_simplified = { [38] = "&", amp = "&", [42] = '"', quot = '"', [47] = "'", apos = "'", [74] = "<", lt = "<", [76] = ">", gt = ">", } local function handle_hex_entity(str) local h = hcache[str] if not h then local n = tonumber(str,16) h = unify_predefined and predefined_unified[n] if h then if trace_entities then report_xml("utfize, converting hex entity &#x%s; into %s",str,h) end elseif utfize then h = (n and utfchar(n)) or xml.unknown_hex_entity_format(str) or "" if not n then report_xml("utfize, ignoring hex entity &#x%s;",str) elseif trace_entities then report_xml("utfize, converting hex entity &#x%s; into %s",str,h) end else if trace_entities then report_xml("found entity &#x%s;",str) end h = "&#x" .. str .. ";" end hcache[str] = h end return h end local function handle_dec_entity(str) local d = dcache[str] if not d then local n = tonumber(str) d = unify_predefined and predefined_unified[n] if d then if trace_entities then report_xml("utfize, converting dec entity &#%s; into %s",str,d) end elseif utfize then d = (n and utfchar(n)) or xml.unknown_dec_entity_format(str) or "" if not n then report_xml("utfize, ignoring dec entity &#%s;",str) elseif trace_entities then report_xml("utfize, converting dec entity &#%s; into %s",str,h) end else if trace_entities then report_xml("found entity &#%s;",str) end d = "&#" .. str .. ";" end dcache[str] = d end return d end xml.parsedentitylpeg = parsedentity local function handle_any_entity(str) if resolve then local a = acache[str] -- per instance ! todo if not a then a = resolve_predefined and predefined_simplified[str] if a then -- one of the predefined elseif type(resolve) == "function" then a = resolve(str) or entities[str] else a = entities[str] end if a then if trace_entities then report_xml("resolved entity &%s; -> %s (internal)",str,a) end a = lpegmatch(parsedentity,a) or a else if xml.unknown_any_entity_format then a = xml.unknown_any_entity_format(str) or "" end if a then if trace_entities then report_xml("resolved entity &%s; -> %s (external)",str,a) end else if trace_entities then report_xml("keeping entity &%s;",str) end if str == "" then a = "&error;" else a = "&" .. str .. ";" end end end acache[str] = a elseif trace_entities then if not acache[str] then report_xml("converting entity &%s; into %s",str,a) acache[str] = a end end return a else local a = acache[str] if not a then if trace_entities then report_xml("found entity &%s;",str) end a = resolve_predefined and predefined_simplified[str] if a then -- one of the predefined acache[str] = a elseif str == "" then a = "&error;" acache[str] = a else a = "&" .. str .. ";" acache[str] = a end end return a end end local function handle_end_entity(chr) report_xml("error in entity, %q found instead of ';'",chr) end local space = S(' \r\n\t') local open = P('<') local close = P('>') local squote = S("'") local dquote = S('"') local equal = P('=') local slash = P('/') local colon = P(':') local semicolon = P(';') local ampersand = P('&') local valid = R('az', 'AZ', '09') + S('_-.') local name_yes = C(valid^1) * colon * C(valid^1) local name_nop = C(P(true)) * C(valid^1) local name = name_yes + name_nop local utfbom = lpeg.patterns.utfbom -- no capture local spacing = C(space^0) ----- entitycontent = (1-open-semicolon)^0 local anyentitycontent = (1-open-semicolon-space-close)^0 local hexentitycontent = R("AF","af","09")^0 local decentitycontent = R("09")^0 local parsedentity = P("#")/"" * ( P("x")/"" * (hexentitycontent/handle_hex_entity) + (decentitycontent/handle_dec_entity) ) + (anyentitycontent/handle_any_entity) local entity = ampersand/"" * parsedentity * ( (semicolon/"") + #(P(1)/handle_end_entity)) local text_unparsed = C((1-open)^1) local text_parsed = Cs(((1-open-ampersand)^1 + entity)^1) local somespace = space^1 local optionalspace = space^0 ----- value = (squote * C((1 - squote)^0) * squote) + (dquote * C((1 - dquote)^0) * dquote) -- ampersand and < also invalid in value local value = (squote * Cs((entity + (1 - squote))^0) * squote) + (dquote * Cs((entity + (1 - dquote))^0) * dquote) -- ampersand and < also invalid in value local endofattributes = slash * close + close -- recovery of flacky html local whatever = space * name * optionalspace * equal local wrongvalue = C(P(1-whatever-close)^1 + P(1-close)^1) / attribute_value_error ----- wrongvalue = C(P(1-whatever-endofattributes)^1 + P(1-endofattributes)^1) / attribute_value_error ----- wrongvalue = C(P(1-space-endofattributes)^1) / attribute_value_error local wrongvalue = Cs(P(entity + (1-space-endofattributes))^1) / attribute_value_error local attributevalue = value + wrongvalue local attribute = (somespace * name * optionalspace * equal * optionalspace * attributevalue) / add_attribute ----- attributes = (attribute)^0 local attributes = (attribute + somespace^-1 * (((1-endofattributes)^1)/attribute_specification_error))^0 local parsedtext = text_parsed / add_text local unparsedtext = text_unparsed / add_text local balanced = P { "[" * ((1 - S"[]") + V(1))^0 * "]" } -- taken from lpeg manual, () example local emptyelement = (spacing * open * name * attributes * optionalspace * slash * close) / add_empty local beginelement = (spacing * open * name * attributes * optionalspace * close) / add_begin local endelement = (spacing * open * slash * name * optionalspace * close) / add_end local begincomment = open * P("!--") local endcomment = P("--") * close local begininstruction = open * P("?") local endinstruction = P("?") * close local begincdata = open * P("![CDATA[") local endcdata = P("]]") * close local someinstruction = C((1 - endinstruction)^0) local somecomment = C((1 - endcomment )^0) local somecdata = C((1 - endcdata )^0) local function normalentity(k,v ) entities[k] = v end local function systementity(k,v,n) entities[k] = v end local function publicentity(k,v,n) entities[k] = v end local begindoctype = open * P("!DOCTYPE") local enddoctype = close local beginset = P("[") local endset = P("]") local doctypename = C((1-somespace-close)^0) local elementdoctype = optionalspace * P(" & cleanup = settings.text_cleanup stack, top, at, xmlns, errorstr, entities = { }, { }, { }, { }, nil, settings.entities or { } acache, hcache, dcache = { }, { }, { } -- not stored reported_attribute_errors = { } if settings.parent_root then mt = getmetatable(settings.parent_root) else initialize_mt(top) end stack[#stack+1] = top top.dt = { } dt = top.dt if not data or data == "" then errorstr = "empty xml file" elseif utfize or resolve then if lpegmatch(grammar_parsed_text,data) then errorstr = "" else errorstr = "invalid xml file - parsed text" end elseif type(data) == "string" then if lpegmatch(grammar_unparsed_text,data) then errorstr = "" else errorstr = "invalid xml file - unparsed text" end else errorstr = "invalid xml file - no text at all" end local result if errorstr and errorstr ~= "" then result = { dt = { { ns = "", tg = "error", dt = { errorstr }, at={ }, er = true } } } setmetatable(stack, mt) local error_handler = settings.error_handler if error_handler == false then -- no error message else error_handler = error_handler or xml.error_handler if error_handler then xml.error_handler("load",errorstr) end end else result = stack[1] end if not settings.no_root then result = { special = true, ns = "", tg = '@rt@', dt = result.dt, at={ }, entities = entities, settings = settings } setmetatable(result, mt) local rdt = result.dt for k=1,#rdt do local v = rdt[k] if type(v) == "table" and not v.special then -- always table -) result.ri = k -- rootindex v.__p__ = result -- new, experiment, else we cannot go back to settings, we need to test this ! break end end end if errorstr and errorstr ~= "" then result.error = true end return result end xml.convert = xmlconvert function xml.inheritedconvert(data,xmldata) local settings = xmldata.settings settings.parent_root = xmldata -- to be tested -- settings.no_root = true local xc = xmlconvert(data,settings) -- xc.settings = nil -- xc.entities = nil -- xc.special = nil -- xc.ri = nil -- print(xc.tg) return xc end --[[ldx--

Packaging data in an xml like table is done with the following function. Maybe it will go away (when not used).

--ldx]]-- function xml.is_valid(root) return root and root.dt and root.dt[1] and type(root.dt[1]) == "table" and not root.dt[1].er end function xml.package(tag,attributes,data) local ns, tg = match(tag,"^(.-):?([^:]+)$") local t = { ns = ns, tg = tg, dt = data or "", at = attributes or {} } setmetatable(t, mt) return t end function xml.is_valid(root) return root and not root.error end xml.error_handler = (logs and logs.report) or (input and logs.report) or print --[[ldx--

We cannot load an from a filehandle so we need to load the whole file first. The function accepts a string representing a filename or a file handle.

--ldx]]-- function xml.load(filename,settings) local data = "" if type(filename) == "string" then -- local data = io.loaddata(filename) - -todo: check type in io.loaddata local f = io.open(filename,'r') if f then data = f:read("*all") f:close() end elseif filename then -- filehandle data = filename:read("*all") end return xmlconvert(data,settings) end --[[ldx--

When we inject new elements, we need to convert strings to valid trees, which is what the next function does.

--ldx]]-- local no_root = { no_root = true } function xml.toxml(data) if type(data) == "string" then local root = { xmlconvert(data,no_root) } return (#root > 1 and root) or root[1] else return data end end --[[ldx--

For copying a tree we use a dedicated function instead of the generic table copier. Since we know what we're dealing with we can speed up things a bit. The second argument is not to be used!

--ldx]]-- local function copy(old,tables) if old then tables = tables or { } local new = { } if not tables[old] then tables[old] = new end for k,v in next, old do new[k] = (type(v) == "table" and (tables[v] or copy(v, tables))) or v end local mt = getmetatable(old) if mt then setmetatable(new,mt) end return new else return { } end end xml.copy = copy --[[ldx--

In serializing the tree or parts of the tree is a major actitivity which is why the following function is pretty optimized resulting in a few more lines of code than needed. The variant that uses the formatting function for all components is about 15% slower than the concatinating alternative.

--ldx]]-- -- todo: add when not present function xml.checkbom(root) -- can be made faster if root.ri then local dt, found = root.dt, false for k=1,#dt do local v = dt[k] if type(v) == "table" and v.special and v.tg == "@pi@" and find(v.dt[1],"xml.*version=") then found = true break end end if not found then insert(dt, 1, { special=true, ns="", tg="@pi@", dt = { "xml version='1.0' standalone='yes'"} } ) insert(dt, 2, "\n" ) end end end --[[ldx--

At the cost of some 25% runtime overhead you can first convert the tree to a string and then handle the lot.

--ldx]]-- -- new experimental reorganized serialize local function verbose_element(e,handlers) local handle = handlers.handle local serialize = handlers.serialize local ens, etg, eat, edt, ern = e.ns, e.tg, e.at, e.dt, e.rn local ats = eat and next(eat) and { } if ats then for k,v in next, eat do ats[#ats+1] = format('%s=%q',k,v) end end if ern and trace_entities and ern ~= ens then ens = ern end if ens ~= "" then if edt and #edt > 0 then if ats then handle("<",ens,":",etg," ",concat(ats," "),">") else handle("<",ens,":",etg,">") end for i=1,#edt do local e = edt[i] if type(e) == "string" then handle(e) else serialize(e,handlers) end end handle("") else if ats then handle("<",ens,":",etg," ",concat(ats," "),"/>") else handle("<",ens,":",etg,"/>") end end else if edt and #edt > 0 then if ats then handle("<",etg," ",concat(ats," "),">") else handle("<",etg,">") end for i=1,#edt do local ei = edt[i] if type(ei) == "string" then handle(ei) else serialize(ei,handlers) end end handle("") else if ats then handle("<",etg," ",concat(ats," "),"/>") else handle("<",etg,"/>") end end end end local function verbose_pi(e,handlers) handlers.handle("") end local function verbose_comment(e,handlers) handlers.handle("") end local function verbose_cdata(e,handlers) handlers.handle("") end local function verbose_doctype(e,handlers) handlers.handle("") end local function verbose_root(e,handlers) handlers.serialize(e.dt,handlers) end local function verbose_text(e,handlers) handlers.handle(e) end local function verbose_document(e,handlers) local serialize = handlers.serialize local functions = handlers.functions for i=1,#e do local ei = e[i] if type(ei) == "string" then functions["@tx@"](ei,handlers) else serialize(ei,handlers) end end end local function serialize(e,handlers,...) local initialize = handlers.initialize local finalize = handlers.finalize local functions = handlers.functions if initialize then local state = initialize(...) if not state == true then return state end end local etg = e.tg if etg then (functions[etg] or functions["@el@"])(e,handlers) -- elseif type(e) == "string" then -- functions["@tx@"](e,handlers) else functions["@dc@"](e,handlers) end if finalize then return finalize() end end local function xserialize(e,handlers) local functions = handlers.functions local etg = e.tg if etg then (functions[etg] or functions["@el@"])(e,handlers) -- elseif type(e) == "string" then -- functions["@tx@"](e,handlers) else functions["@dc@"](e,handlers) end end local handlers = { } local function newhandlers(settings) local t = table.copy(handlers.verbose or { }) -- merge if settings then for k,v in next, settings do if type(v) == "table" then local tk = t[k] if not tk then tk = { } t[k] = tk end for kk,vv in next, v do tk[kk] = vv end else t[k] = v end end if settings.name then handlers[settings.name] = t end end return t end local nofunction = function() end function xml.sethandlersfunction(handler,name,fnc) handler.functions[name] = fnc or nofunction end function xml.gethandlersfunction(handler,name) return handler.functions[name] end function xml.gethandlers(name) return handlers[name] end newhandlers { name = "verbose", initialize = false, -- faster than nil and mt lookup finalize = false, -- faster than nil and mt lookup serialize = xserialize, handle = print, functions = { ["@dc@"] = verbose_document, ["@dt@"] = verbose_doctype, ["@rt@"] = verbose_root, ["@el@"] = verbose_element, ["@pi@"] = verbose_pi, ["@cm@"] = verbose_comment, ["@cd@"] = verbose_cdata, ["@tx@"] = verbose_text, } } --[[ldx--

How you deal with saving data depends on your preferences. For a 40 MB database file the timing on a 2.3 Core Duo are as follows (time in seconds):

1.3 : load data from file to string 6.1 : convert string into tree 5.3 : saving in file using xmlsave 6.8 : converting to string using xml.tostring 3.6 : saving converted string in file

Beware, these were timing with the old routine but measurements will not be that much different I guess.

--ldx]]-- -- maybe this will move to lxml-xml local result local xmlfilehandler = newhandlers { name = "file", initialize = function(name) result = io.open(name,"wb") return result end, finalize = function() result:close() return true end, handle = function(...) result:write(...) end, } -- no checking on writeability here but not faster either -- -- local xmlfilehandler = newhandlers { -- initialize = function(name) io.output(name,"wb") return true end, -- finalize = function() io.close() return true end, -- handle = io.write, -- } function xml.save(root,name) serialize(root,xmlfilehandler,name) end local result local xmlstringhandler = newhandlers { name = "string", initialize = function() result = { } return result end, finalize = function() return concat(result) end, handle = function(...) result[#result+1] = concat { ... } end } local function xmltostring(root) -- 25% overhead due to collecting if root then if type(root) == 'string' then return root else -- if next(root) then -- next is faster than type (and >0 test) return serialize(root,xmlstringhandler) or "" end end return "" end local function xmltext(root) -- inline return (root and xmltostring(root)) or "" end initialize_mt = function(root) -- redefinition mt = { __tostring = xmltext, __index = root } end xml.defaulthandlers = handlers xml.newhandlers = newhandlers xml.serialize = serialize xml.tostring = xmltostring --[[ldx--

The next function operated on the content only and needs a handle function that accepts a string.

--ldx]]-- local function xmlstring(e,handle) if not handle or (e.special and e.tg ~= "@rt@") then -- nothing elseif e.tg then local edt = e.dt if edt then for i=1,#edt do xmlstring(edt[i],handle) end end else handle(e) end end xml.string = xmlstring --[[ldx--

A few helpers:

--ldx]]-- function xml.settings(e) while e do local s = e.settings if s then return s else e = e.__p__ end end return nil end function xml.root(e) local r = e while e do e = e.__p__ if e then r = e end end return r end function xml.parent(root) return root.__p__ end function xml.body(root) return (root.ri and root.dt[root.ri]) or root -- not ok yet end function xml.name(root) if not root then return "" elseif root.ns == "" then return root.tg else return root.ns .. ":" .. root.tg end end --[[ldx--

The next helper erases an element but keeps the table as it is, and since empty strings are not serialized (effectively) it does not harm. Copying the table would take more time. Usage:

--ldx]]-- function xml.erase(dt,k) if dt then if k then dt[k] = "" else for k=1,#dt do dt[1] = { "" } end end end end --[[ldx--

The next helper assigns a tree (or string). Usage:

dt[k] = xml.assign(root) or xml.assign(dt,k,root) --ldx]]-- function xml.assign(dt,k,root) if dt and k then dt[k] = (type(root) == "table" and xml.body(root)) or root return dt[k] else return xml.body(root) end end -- the following helpers may move --[[ldx--

The next helper assigns a tree (or string). Usage:

xml.tocdata(e) xml.tocdata(e,"error") --ldx]]-- function xml.tocdata(e,wrapper) local whatever = xmltostring(e.dt) if wrapper then whatever = format("<%s>%s",wrapper,whatever,wrapper) end local t = { special = true, ns = "", tg = "@cd@", at = {}, rn = "", dt = { whatever }, __p__ = e } setmetatable(t,getmetatable(e)) e.dt = { t } end function xml.makestandalone(root) if root.ri then local dt = root.dt for k=1,#dt do local v = dt[k] if type(v) == "table" and v.special and v.tg == "@pi@" then local txt = v.dt[1] if find(txt,"xml.*version=") then v.dt[1] = txt .. " standalone='yes'" break end end end end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['lxml-pth'] = { version = 1.001, comment = "this module is the basis for the lxml-* ones", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- e.ni is only valid after a filter run local concat, remove, insert = table.concat, table.remove, table.insert local type, next, tonumber, tostring, setmetatable, loadstring = type, next, tonumber, tostring, setmetatable, loadstring local format, upper, lower, gmatch, gsub, find, rep = string.format, string.upper, string.lower, string.gmatch, string.gsub, string.find, string.rep local lpegmatch = lpeg.match -- beware, this is not xpath ... e.g. position is different (currently) and -- we have reverse-sibling as reversed preceding sibling --[[ldx--

This module can be used stand alone but also inside in which case it hooks into the tracker code. Therefore we provide a few functions that set the tracers. Here we overload a previously defined function.

If I can get in the mood I will make a variant that is XSLT compliant but I wonder if it makes sense.

--ldx]]-- --[[ldx--

Expecially the lpath code is experimental, we will support some of xpath, but only things that make sense for us; as compensation it is possible to hook in your own functions. Apart from preprocessing content for we also need this module for process management, like handling and files.

a/b/c /*/c a/b/c/first() a/b/c/last() a/b/c/index(n) a/b/c/index(-n) a/b/c/text() a/b/c/text(1) a/b/c/text(-1) a/b/c/text(n) --ldx]]-- local trace_lpath = false if trackers then trackers.register("xml.path", function(v) trace_lpath = v end) end local trace_lparse = false if trackers then trackers.register("xml.parse", function(v) trace_lparse = v end) end local trace_lprofile = false if trackers then trackers.register("xml.profile", function(v) trace_lpath = v trace_lparse = v trace_lprofile = v end) end local report_lpath = logs.new("lpath") --[[ldx--

We've now arrived at an interesting part: accessing the tree using a subset of and since we're not compatible we call it . We will explain more about its usage in other documents.

--ldx]]-- local lpathcalls = 0 function xml.lpathcalls () return lpathcalls end local lpathcached = 0 function xml.lpathcached() return lpathcached end xml.functions = xml.functions or { } -- internal xml.expressions = xml.expressions or { } -- in expressions xml.finalizers = xml.finalizers or { } -- fast do-with ... (with return value other than collection) xml.specialhandler = xml.specialhandler or { } local functions = xml.functions local expressions = xml.expressions local finalizers = xml.finalizers finalizers.xml = finalizers.xml or { } finalizers.tex = finalizers.tex or { } local function fallback (t, name) local fn = finalizers[name] if fn then t[name] = fn else report_lpath("unknown sub finalizer '%s'",tostring(name)) fn = function() end end return fn end setmetatable(finalizers.xml, { __index = fallback }) setmetatable(finalizers.tex, { __index = fallback }) xml.defaultprotocol = "xml" -- as xsl does not follow xpath completely here we will also -- be more liberal especially with regards to the use of | and -- the rootpath: -- -- test : all 'test' under current -- /test : 'test' relative to current -- a|b|c : set of names -- (a|b|c) : idem -- ! : not -- -- after all, we're not doing transformations but filtering. in -- addition we provide filter functions (last bit) -- -- todo: optimizer -- -- .. : parent -- * : all kids -- / : anchor here -- // : /**/ -- ** : all in between -- -- so far we had (more practical as we don't transform) -- -- {/test} : kids 'test' under current node -- {test} : any kid with tag 'test' -- {//test} : same as above -- evaluator (needs to be redone, for the moment copied) -- todo: apply_axis(list,notable) and collection vs single local apply_axis = { } apply_axis['root'] = function(list) local collected = { } for l=1,#list do local ll = list[l] local rt = ll while ll do ll = ll.__p__ if ll then rt = ll end end collected[#collected+1] = rt end return collected end apply_axis['self'] = function(list) return list end apply_axis['child'] = function(list) local collected = { } for l=1,#list do local ll = list[l] local dt = ll.dt if dt then -- weird that this is needed local en = 0 for k=1,#dt do local dk = dt[k] if dk.tg then collected[#collected+1] = dk dk.ni = k -- refresh en = en + 1 dk.ei = en end end ll.en = en end end return collected end local function collect(list,collected) local dt = list.dt if dt then local en = 0 for k=1,#dt do local dk = dt[k] if dk.tg then collected[#collected+1] = dk dk.ni = k -- refresh en = en + 1 dk.ei = en collect(dk,collected) end end list.en = en end end apply_axis['descendant'] = function(list) local collected = { } for l=1,#list do collect(list[l],collected) end return collected end local function collect(list,collected) local dt = list.dt if dt then local en = 0 for k=1,#dt do local dk = dt[k] if dk.tg then collected[#collected+1] = dk dk.ni = k -- refresh en = en + 1 dk.ei = en collect(dk,collected) end end list.en = en end end apply_axis['descendant-or-self'] = function(list) local collected = { } for l=1,#list do local ll = list[l] if ll.special ~= true then -- catch double root collected[#collected+1] = ll end collect(ll,collected) end return collected end apply_axis['ancestor'] = function(list) local collected = { } for l=1,#list do local ll = list[l] while ll do ll = ll.__p__ if ll then collected[#collected+1] = ll end end end return collected end apply_axis['ancestor-or-self'] = function(list) local collected = { } for l=1,#list do local ll = list[l] collected[#collected+1] = ll while ll do ll = ll.__p__ if ll then collected[#collected+1] = ll end end end return collected end apply_axis['parent'] = function(list) local collected = { } for l=1,#list do local pl = list[l].__p__ if pl then collected[#collected+1] = pl end end return collected end apply_axis['attribute'] = function(list) return { } end apply_axis['namespace'] = function(list) return { } end apply_axis['following'] = function(list) -- incomplete return { } end apply_axis['preceding'] = function(list) -- incomplete return { } end apply_axis['following-sibling'] = function(list) local collected = { } for l=1,#list do local ll = list[l] local p = ll.__p__ local d = p.dt for i=ll.ni+1,#d do local di = d[i] if type(di) == "table" then collected[#collected+1] = di end end end return collected end apply_axis['preceding-sibling'] = function(list) local collected = { } for l=1,#list do local ll = list[l] local p = ll.__p__ local d = p.dt for i=1,ll.ni-1 do local di = d[i] if type(di) == "table" then collected[#collected+1] = di end end end return collected end apply_axis['reverse-sibling'] = function(list) -- reverse preceding local collected = { } for l=1,#list do local ll = list[l] local p = ll.__p__ local d = p.dt for i=ll.ni-1,1,-1 do local di = d[i] if type(di) == "table" then collected[#collected+1] = di end end end return collected end apply_axis['auto-descendant-or-self'] = apply_axis['descendant-or-self'] apply_axis['auto-descendant'] = apply_axis['descendant'] apply_axis['auto-child'] = apply_axis['child'] apply_axis['auto-self'] = apply_axis['self'] apply_axis['initial-child'] = apply_axis['child'] local function apply_nodes(list,directive,nodes) -- todo: nodes[1] etc ... negated node name in set ... when needed -- ... currently ignored local maxn = #nodes if maxn == 3 then --optimized loop local nns, ntg = nodes[2], nodes[3] if not nns and not ntg then -- wildcard if directive then return list else return { } end else local collected, m, p = { }, 0, nil if not nns then -- only check tag for l=1,#list do local ll = list[l] local ltg = ll.tg if ltg then if directive then if ntg == ltg then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end elseif ntg ~= ltg then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end end end elseif not ntg then -- only check namespace for l=1,#list do local ll = list[l] local lns = ll.rn or ll.ns if lns then if directive then if lns == nns then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end elseif lns ~= nns then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end end end else -- check both for l=1,#list do local ll = list[l] local ltg = ll.tg if ltg then local lns = ll.rn or ll.ns local ok = ltg == ntg and lns == nns if directive then if ok then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end elseif not ok then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end end end end return collected end else local collected, m, p = { }, 0, nil for l=1,#list do local ll = list[l] local ltg = ll.tg if ltg then local lns = ll.rn or ll.ns local ok = false for n=1,maxn,3 do local nns, ntg = nodes[n+1], nodes[n+2] ok = (not ntg or ltg == ntg) and (not nns or lns == nns) if ok then break end end if directive then if ok then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end elseif not ok then local llp = ll.__p__ ; if llp ~= p then p, m = llp, 1 else m = m + 1 end collected[#collected+1], ll.mi = ll, m end end end return collected end end local quit_expression = false local function apply_expression(list,expression,order) local collected = { } quit_expression = false for l=1,#list do local ll = list[l] if expression(list,ll,l,order) then -- nasty, order alleen valid als n=1 collected[#collected+1] = ll end if quit_expression then break end end return collected end local P, V, C, Cs, Cc, Ct, R, S, Cg, Cb = lpeg.P, lpeg.V, lpeg.C, lpeg.Cs, lpeg.Cc, lpeg.Ct, lpeg.R, lpeg.S, lpeg.Cg, lpeg.Cb local spaces = S(" \n\r\t\f")^0 local lp_space = S(" \n\r\t\f") local lp_any = P(1) local lp_noequal = P("!=") / "~=" + P("<=") + P(">=") + P("==") local lp_doequal = P("=") / "==" local lp_or = P("|") / " or " local lp_and = P("&") / " and " local lp_builtin = P ( P("firstindex") / "1" + P("lastindex") / "(#ll.__p__.dt or 1)" + P("firstelement") / "1" + P("lastelement") / "(ll.__p__.en or 1)" + P("first") / "1" + P("last") / "#list" + P("rootposition") / "order" + P("position") / "l" + -- is element in finalizer P("order") / "order" + P("element") / "(ll.ei or 1)" + P("index") / "(ll.ni or 1)" + P("match") / "(ll.mi or 1)" + P("text") / "(ll.dt[1] or '')" + -- P("name") / "(ll.ns~='' and ll.ns..':'..ll.tg)" + P("name") / "((ll.ns~='' and ll.ns..':'..ll.tg) or ll.tg)" + P("tag") / "ll.tg" + P("ns") / "ll.ns" ) * ((spaces * P("(") * spaces * P(")"))/"") local lp_attribute = (P("@") + P("attribute::")) / "" * Cc("(ll.at and ll.at['") * R("az","AZ","--","__")^1 * Cc("'])") local lp_fastpos_p = ((P("+")^0 * R("09")^1 * P(-1)) / function(s) return "l==" .. s end) local lp_fastpos_n = ((P("-") * R("09")^1 * P(-1)) / function(s) return "(" .. s .. "<0 and (#list+".. s .. "==l))" end) local lp_fastpos = lp_fastpos_n + lp_fastpos_p local lp_reserved = C("and") + C("or") + C("not") + C("div") + C("mod") + C("true") + C("false") local lp_lua_function = C(R("az","AZ","__")^1 * (P(".") * R("az","AZ","__")^1)^1) * ("(") / function(t) -- todo: better . handling return t .. "(" end local lp_function = C(R("az","AZ","__")^1) * P("(") / function(t) -- todo: better . handling if expressions[t] then return "expr." .. t .. "(" else return "expr.error(" end end local lparent = lpeg.P("(") local rparent = lpeg.P(")") local noparent = 1 - (lparent+rparent) local nested = lpeg.P{lparent * (noparent + lpeg.V(1))^0 * rparent} local value = lpeg.P(lparent * lpeg.C((noparent + nested)^0) * rparent) -- lpeg.P{"("*C(((1-S("()"))+V(1))^0)*")"} local lp_child = Cc("expr.child(ll,'") * R("az","AZ","--","__")^1 * Cc("')") local lp_number = S("+-") * R("09")^1 local lp_string = Cc("'") * R("az","AZ","--","__")^1 * Cc("'") local lp_content = (P("'") * (1-P("'"))^0 * P("'") + P('"') * (1-P('"'))^0 * P('"')) local cleaner local lp_special = (C(P("name")+P("text")+P("tag")+P("count")+P("child"))) * value / function(t,s) if expressions[t] then s = s and s ~= "" and lpegmatch(cleaner,s) if s and s ~= "" then return "expr." .. t .. "(ll," .. s ..")" else return "expr." .. t .. "(ll)" end else return "expr.error(" .. t .. ")" end end local content = lp_builtin + lp_attribute + lp_special + lp_noequal + lp_doequal + lp_or + lp_and + lp_reserved + lp_lua_function + lp_function + lp_content + -- too fragile lp_child + lp_any local converter = Cs ( lp_fastpos + (P { lparent * (V(1))^0 * rparent + content } )^0 ) cleaner = Cs ( ( lp_reserved + lp_number + lp_string + 1 )^1 ) local template_e = [[ local expr = xml.expressions return function(list,ll,l,order) return %s end ]] local template_f_y = [[ local finalizer = xml.finalizers['%s']['%s'] return function(collection) return finalizer(collection,%s) end ]] local template_f_n = [[ return xml.finalizers['%s']['%s'] ]] -- local register_self = { kind = "axis", axis = "self" } -- , apply = apply_axis["self"] } local register_parent = { kind = "axis", axis = "parent" } -- , apply = apply_axis["parent"] } local register_descendant = { kind = "axis", axis = "descendant" } -- , apply = apply_axis["descendant"] } local register_child = { kind = "axis", axis = "child" } -- , apply = apply_axis["child"] } local register_descendant_or_self = { kind = "axis", axis = "descendant-or-self" } -- , apply = apply_axis["descendant-or-self"] } local register_root = { kind = "axis", axis = "root" } -- , apply = apply_axis["root"] } local register_ancestor = { kind = "axis", axis = "ancestor" } -- , apply = apply_axis["ancestor"] } local register_ancestor_or_self = { kind = "axis", axis = "ancestor-or-self" } -- , apply = apply_axis["ancestor-or-self"] } local register_attribute = { kind = "axis", axis = "attribute" } -- , apply = apply_axis["attribute"] } local register_namespace = { kind = "axis", axis = "namespace" } -- , apply = apply_axis["namespace"] } local register_following = { kind = "axis", axis = "following" } -- , apply = apply_axis["following"] } local register_following_sibling = { kind = "axis", axis = "following-sibling" } -- , apply = apply_axis["following-sibling"] } local register_preceding = { kind = "axis", axis = "preceding" } -- , apply = apply_axis["preceding"] } local register_preceding_sibling = { kind = "axis", axis = "preceding-sibling" } -- , apply = apply_axis["preceding-sibling"] } local register_reverse_sibling = { kind = "axis", axis = "reverse-sibling" } -- , apply = apply_axis["reverse-sibling"] } local register_auto_descendant_or_self = { kind = "axis", axis = "auto-descendant-or-self" } -- , apply = apply_axis["auto-descendant-or-self"] } local register_auto_descendant = { kind = "axis", axis = "auto-descendant" } -- , apply = apply_axis["auto-descendant"] } local register_auto_self = { kind = "axis", axis = "auto-self" } -- , apply = apply_axis["auto-self"] } local register_auto_child = { kind = "axis", axis = "auto-child" } -- , apply = apply_axis["auto-child"] } local register_initial_child = { kind = "axis", axis = "initial-child" } -- , apply = apply_axis["initial-child"] } local register_all_nodes = { kind = "nodes", nodetest = true, nodes = { true, false, false } } local skip = { } local function errorrunner_e(str,cnv) if not skip[str] then report_lpath("error in expression: %s => %s",str,cnv) skip[str] = cnv or str end return false end local function errorrunner_f(str,arg) report_lpath("error in finalizer: %s(%s)",str,arg or "") return false end local function register_nodes(nodetest,nodes) return { kind = "nodes", nodetest = nodetest, nodes = nodes } end local function register_expression(expression) local converted = lpegmatch(converter,expression) local runner = loadstring(format(template_e,converted)) runner = (runner and runner()) or function() errorrunner_e(expression,converted) end return { kind = "expression", expression = expression, converted = converted, evaluator = runner } end local function register_finalizer(protocol,name,arguments) local runner if arguments and arguments ~= "" then runner = loadstring(format(template_f_y,protocol or xml.defaultprotocol,name,arguments)) else runner = loadstring(format(template_f_n,protocol or xml.defaultprotocol,name)) end runner = (runner and runner()) or function() errorrunner_f(name,arguments) end return { kind = "finalizer", name = name, arguments = arguments, finalizer = runner } end local expression = P { "ex", ex = "[" * C((V("sq") + V("dq") + (1 - S("[]")) + V("ex"))^0) * "]", sq = "'" * (1 - S("'"))^0 * "'", dq = '"' * (1 - S('"'))^0 * '"', } local arguments = P { "ar", ar = "(" * Cs((V("sq") + V("dq") + V("nq") + P(1-P(")")))^0) * ")", nq = ((1 - S("),'\""))^1) / function(s) return format("%q",s) end, sq = P("'") * (1 - P("'"))^0 * P("'"), dq = P('"') * (1 - P('"'))^0 * P('"'), } -- todo: better arg parser local function register_error(str) return { kind = "error", error = format("unparsed: %s",str) } end -- there is a difference in * and /*/ and so we need to catch a few special cases local special_1 = P("*") * Cc(register_auto_descendant) * Cc(register_all_nodes) -- last one not needed local special_2 = P("/") * Cc(register_auto_self) local special_3 = P("") * Cc(register_auto_self) local parser = Ct { "patterns", -- can be made a bit faster by moving pattern outside patterns = spaces * V("protocol") * spaces * ( ( V("special") * spaces * P(-1) ) + ( V("initial") * spaces * V("step") * spaces * (P("/") * spaces * V("step") * spaces)^0 ) ), protocol = Cg(V("letters"),"protocol") * P("://") + Cg(Cc(nil),"protocol"), -- the / is needed for // as descendant or self is somewhat special -- step = (V("shortcuts") + V("axis") * spaces * V("nodes")^0 + V("error")) * spaces * V("expressions")^0 * spaces * V("finalizer")^0, step = ((V("shortcuts") + P("/") + V("axis")) * spaces * V("nodes")^0 + V("error")) * spaces * V("expressions")^0 * spaces * V("finalizer")^0, axis = V("descendant") + V("child") + V("parent") + V("self") + V("root") + V("ancestor") + V("descendant_or_self") + V("following_sibling") + V("following") + V("reverse_sibling") + V("preceding_sibling") + V("preceding") + V("ancestor_or_self") + #(1-P(-1)) * Cc(register_auto_child), special = special_1 + special_2 + special_3, initial = (P("/") * spaces * Cc(register_initial_child))^-1, error = (P(1)^1) / register_error, shortcuts_a = V("s_descendant_or_self") + V("s_descendant") + V("s_child") + V("s_parent") + V("s_self") + V("s_root") + V("s_ancestor"), shortcuts = V("shortcuts_a") * (spaces * "/" * spaces * V("shortcuts_a"))^0, s_descendant_or_self = (P("***/") + P("/")) * Cc(register_descendant_or_self), --- *** is a bonus -- s_descendant_or_self = P("/") * Cc(register_descendant_or_self), s_descendant = P("**") * Cc(register_descendant), s_child = P("*") * #(1-P(":")) * Cc(register_child ), -- s_child = P("*") * #(P("/")+P(-1)) * Cc(register_child ), s_parent = P("..") * Cc(register_parent ), s_self = P("." ) * Cc(register_self ), s_root = P("^^") * Cc(register_root ), s_ancestor = P("^") * Cc(register_ancestor ), descendant = P("descendant::") * Cc(register_descendant ), child = P("child::") * Cc(register_child ), parent = P("parent::") * Cc(register_parent ), self = P("self::") * Cc(register_self ), root = P('root::') * Cc(register_root ), ancestor = P('ancestor::') * Cc(register_ancestor ), descendant_or_self = P('descendant-or-self::') * Cc(register_descendant_or_self ), ancestor_or_self = P('ancestor-or-self::') * Cc(register_ancestor_or_self ), -- attribute = P('attribute::') * Cc(register_attribute ), -- namespace = P('namespace::') * Cc(register_namespace ), following = P('following::') * Cc(register_following ), following_sibling = P('following-sibling::') * Cc(register_following_sibling ), preceding = P('preceding::') * Cc(register_preceding ), preceding_sibling = P('preceding-sibling::') * Cc(register_preceding_sibling ), reverse_sibling = P('reverse-sibling::') * Cc(register_reverse_sibling ), nodes = (V("nodefunction") * spaces * P("(") * V("nodeset") * P(")") + V("nodetest") * V("nodeset")) / register_nodes, expressions = expression / register_expression, letters = R("az")^1, name = (1-lpeg.S("/[]()|:*!"))^1, negate = P("!") * Cc(false), nodefunction = V("negate") + P("not") * Cc(false) + Cc(true), nodetest = V("negate") + Cc(true), nodename = (V("negate") + Cc(true)) * spaces * ((V("wildnodename") * P(":") * V("wildnodename")) + (Cc(false) * V("wildnodename"))), wildnodename = (C(V("name")) + P("*") * Cc(false)) * #(1-P("(")), nodeset = spaces * Ct(V("nodename") * (spaces * P("|") * spaces * V("nodename"))^0) * spaces, finalizer = (Cb("protocol") * P("/")^-1 * C(V("name")) * arguments * P(-1)) / register_finalizer, } local cache = { } local function nodesettostring(set,nodetest) local t = { } for i=1,#set,3 do local directive, ns, tg = set[i], set[i+1], set[i+2] if not ns or ns == "" then ns = "*" end if not tg or tg == "" then tg = "*" end tg = (tg == "@rt@" and "[root]") or format("%s:%s",ns,tg) t[#t+1] = (directive and tg) or format("not(%s)",tg) end if nodetest == false then return format("not(%s)",concat(t,"|")) else return concat(t,"|") end end local function tagstostring(list) if #list == 0 then return "no elements" else local t = { } for i=1, #list do local li = list[i] local ns, tg = li.ns, li.tg if not ns or ns == "" then ns = "*" end if not tg or tg == "" then tg = "*" end t[#t+1] = (tg == "@rt@" and "[root]") or format("%s:%s",ns,tg) end return concat(t," ") end end xml.nodesettostring = nodesettostring local parse_pattern -- we have a harmless kind of circular reference local function lshow(parsed) if type(parsed) == "string" then parsed = parse_pattern(parsed) end local s = table.serialize_functions -- ugly table.serialize_functions = false -- ugly report_lpath("%s://%s => %s",parsed.protocol or xml.defaultprotocol,parsed.pattern,table.serialize(parsed,false)) table.serialize_functions = s -- ugly end xml.lshow = lshow local function add_comment(p,str) local pc = p.comment if not pc then p.comment = { str } else pc[#pc+1] = str end end parse_pattern = function (pattern) -- the gain of caching is rather minimal lpathcalls = lpathcalls + 1 if type(pattern) == "table" then return pattern else local parsed = cache[pattern] if parsed then lpathcached = lpathcached + 1 else parsed = lpegmatch(parser,pattern) if parsed then parsed.pattern = pattern local np = #parsed if np == 0 then parsed = { pattern = pattern, register_self, state = "parsing error" } report_lpath("parsing error in '%s'",pattern) lshow(parsed) else -- we could have done this with a more complex parser but this -- is cleaner local pi = parsed[1] if pi.axis == "auto-child" then if false then add_comment(parsed, "auto-child replaced by auto-descendant-or-self") parsed[1] = register_auto_descendant_or_self else add_comment(parsed, "auto-child replaced by auto-descendant") parsed[1] = register_auto_descendant end elseif pi.axis == "initial-child" and np > 1 and parsed[2].axis then add_comment(parsed, "initial-child removed") -- we could also make it a auto-self remove(parsed,1) end local np = #parsed -- can have changed if np > 1 then local pnp = parsed[np] if pnp.kind == "nodes" and pnp.nodetest == true then local nodes = pnp.nodes if nodes[1] == true and nodes[2] == false and nodes[3] == false then add_comment(parsed, "redundant final wildcard filter removed") remove(parsed,np) end end end end else parsed = { pattern = pattern } end cache[pattern] = parsed if trace_lparse and not trace_lprofile then lshow(parsed) end end return parsed end end -- we can move all calls inline and then merge the trace back -- technically we can combine axis and the next nodes which is -- what we did before but this a bit cleaner (but slower too) -- but interesting is that it's not that much faster when we -- go inline -- -- beware: we need to return a collection even when we filter -- else the (simple) cache gets messed up -- caching found lookups saves not that much (max .1 sec on a 8 sec run) -- and it also messes up finalizers -- watch out: when there is a finalizer, it's always called as there -- can be cases that a finalizer returns (or does) something in case -- there is no match; an example of this is count() local profiled = { } xml.profiled = profiled local function profiled_apply(list,parsed,nofparsed,order) local p = profiled[parsed.pattern] if p then p.tested = p.tested + 1 else p = { tested = 1, matched = 0, finalized = 0 } profiled[parsed.pattern] = p end local collected = list for i=1,nofparsed do local pi = parsed[i] local kind = pi.kind if kind == "axis" then collected = apply_axis[pi.axis](collected) elseif kind == "nodes" then collected = apply_nodes(collected,pi.nodetest,pi.nodes) elseif kind == "expression" then collected = apply_expression(collected,pi.evaluator,order) elseif kind == "finalizer" then collected = pi.finalizer(collected) p.matched = p.matched + 1 p.finalized = p.finalized + 1 return collected end if not collected or #collected == 0 then local pn = i < nofparsed and parsed[nofparsed] if pn and pn.kind == "finalizer" then collected = pn.finalizer(collected) p.finalized = p.finalized + 1 return collected end return nil end end if collected then p.matched = p.matched + 1 end return collected end local function traced_apply(list,parsed,nofparsed,order) if trace_lparse then lshow(parsed) end report_lpath("collecting : %s",parsed.pattern) report_lpath(" root tags : %s",tagstostring(list)) report_lpath(" order : %s",order or "unset") local collected = list for i=1,nofparsed do local pi = parsed[i] local kind = pi.kind if kind == "axis" then collected = apply_axis[pi.axis](collected) report_lpath("% 10i : ax : %s",(collected and #collected) or 0,pi.axis) elseif kind == "nodes" then collected = apply_nodes(collected,pi.nodetest,pi.nodes) report_lpath("% 10i : ns : %s",(collected and #collected) or 0,nodesettostring(pi.nodes,pi.nodetest)) elseif kind == "expression" then collected = apply_expression(collected,pi.evaluator,order) report_lpath("% 10i : ex : %s -> %s",(collected and #collected) or 0,pi.expression,pi.converted) elseif kind == "finalizer" then collected = pi.finalizer(collected) report_lpath("% 10i : fi : %s : %s(%s)",(type(collected) == "table" and #collected) or 0,parsed.protocol or xml.defaultprotocol,pi.name,pi.arguments or "") return collected end if not collected or #collected == 0 then local pn = i < nofparsed and parsed[nofparsed] if pn and pn.kind == "finalizer" then collected = pn.finalizer(collected) report_lpath("% 10i : fi : %s : %s(%s)",(type(collected) == "table" and #collected) or 0,parsed.protocol or xml.defaultprotocol,pn.name,pn.arguments or "") return collected end return nil end end return collected end local function normal_apply(list,parsed,nofparsed,order) local collected = list for i=1,nofparsed do local pi = parsed[i] local kind = pi.kind if kind == "axis" then local axis = pi.axis if axis ~= "self" then collected = apply_axis[axis](collected) end elseif kind == "nodes" then collected = apply_nodes(collected,pi.nodetest,pi.nodes) elseif kind == "expression" then collected = apply_expression(collected,pi.evaluator,order) elseif kind == "finalizer" then return pi.finalizer(collected) end if not collected or #collected == 0 then local pf = i < nofparsed and parsed[nofparsed].finalizer if pf then return pf(collected) -- can be anything end return nil end end return collected end local function parse_apply(list,pattern) -- we avoid an extra call local parsed = cache[pattern] if parsed then lpathcalls = lpathcalls + 1 lpathcached = lpathcached + 1 elseif type(pattern) == "table" then lpathcalls = lpathcalls + 1 parsed = pattern else parsed = parse_pattern(pattern) or pattern end if not parsed then return end local nofparsed = #parsed if nofparsed == 0 then return -- something is wrong end local one = list[1] if not one then return -- something is wrong elseif not trace_lpath then return normal_apply(list,parsed,nofparsed,one.mi) elseif trace_lprofile then return profiled_apply(list,parsed,nofparsed,one.mi) else return traced_apply(list,parsed,nofparsed,one.mi) end end -- internal (parsed) expressions.child = function(e,pattern) return parse_apply({ e },pattern) -- todo: cache end expressions.count = function(e,pattern) local collected = parse_apply({ e },pattern) -- todo: cache return (collected and #collected) or 0 end -- external expressions.oneof = function(s,...) -- slow local t = {...} for i=1,#t do if s == t[i] then return true end end return false end expressions.error = function(str) xml.error_handler("unknown function in lpath expression",tostring(str or "?")) return false end expressions.undefined = function(s) return s == nil end expressions.quit = function(s) if s or s == nil then quit_expression = true end return true end expressions.print = function(...) print(...) return true end expressions.contains = find expressions.find = find expressions.upper = upper expressions.lower = lower expressions.number = tonumber expressions.boolean = toboolean -- user interface local function traverse(root,pattern,handle) report_lpath("use 'xml.selection' instead for '%s'",pattern) local collected = parse_apply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] local r = e.__p__ handle(r,r.dt,e.ni) end end end local function selection(root,pattern,handle) local collected = parse_apply({ root },pattern) if collected then if handle then for c=1,#collected do handle(collected[c]) end else return collected end end end xml.parse_parser = parser xml.parse_pattern = parse_pattern xml.parse_apply = parse_apply xml.traverse = traverse -- old method, r, d, k xml.selection = selection -- new method, simple handle local lpath = parse_pattern xml.lpath = lpath function xml.cached_patterns() return cache end -- generic function finalizer (independant namespace) local function dofunction(collected,fnc) if collected then local f = functions[fnc] if f then for c=1,#collected do f(collected[c]) end else report_lpath("unknown function '%s'",fnc) end end end xml.finalizers.xml["function"] = dofunction xml.finalizers.tex["function"] = dofunction -- functions expressions.text = function(e,n) local rdt = e.__p__.dt return (rdt and rdt[n]) or "" end expressions.name = function(e,n) -- ns + tg local found = false n = tonumber(n) or 0 if n == 0 then found = type(e) == "table" and e elseif n < 0 then local d, k = e.__p__.dt, e.ni for i=k-1,1,-1 do local di = d[i] if type(di) == "table" then if n == -1 then found = di break else n = n + 1 end end end else local d, k = e.__p__.dt, e.ni for i=k+1,#d,1 do local di = d[i] if type(di) == "table" then if n == 1 then found = di break else n = n - 1 end end end end if found then local ns, tg = found.rn or found.ns or "", found.tg if ns ~= "" then return ns .. ":" .. tg else return tg end else return "" end end expressions.tag = function(e,n) -- only tg if not e then return "" else local found = false n = tonumber(n) or 0 if n == 0 then found = (type(e) == "table") and e -- seems to fail elseif n < 0 then local d, k = e.__p__.dt, e.ni for i=k-1,1,-1 do local di = d[i] if type(di) == "table" then if n == -1 then found = di break else n = n + 1 end end end else local d, k = e.__p__.dt, e.ni for i=k+1,#d,1 do local di = d[i] if type(di) == "table" then if n == 1 then found = di break else n = n - 1 end end end end return (found and found.tg) or "" end end --[[ldx--

This is the main filter function. It returns whatever is asked for.

--ldx]]-- function xml.filter(root,pattern) -- no longer funny attribute handling here return parse_apply({ root },pattern) end --[[ldx--

Often using an iterators looks nicer in the code than passing handler functions. The book describes how to use coroutines for that purpose (). This permits code like:

for r, d, k in xml.elements(xml.load('text.xml'),"title") do print(d[k]) -- old method end for e in xml.collected(xml.load('text.xml'),"title") do print(e) -- new one end --ldx]]-- local wrap, yield = coroutine.wrap, coroutine.yield function xml.elements(root,pattern,reverse) -- r, d, k local collected = parse_apply({ root },pattern) if collected then if reverse then return wrap(function() for c=#collected,1,-1 do local e = collected[c] local r = e.__p__ yield(r,r.dt,e.ni) end end) else return wrap(function() for c=1,#collected do local e = collected[c] local r = e.__p__ yield(r,r.dt,e.ni) end end) end end return wrap(function() end) end function xml.collected(root,pattern,reverse) -- e local collected = parse_apply({ root },pattern) if collected then if reverse then return wrap(function() for c=#collected,1,-1 do yield(collected[c]) end end) else return wrap(function() for c=1,#collected do yield(collected[c]) end end) end end return wrap(function() end) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['lxml-mis'] = { version = 1.001, comment = "this module is the basis for the lxml-* ones", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local concat = table.concat local type, next, tonumber, tostring, setmetatable, loadstring = type, next, tonumber, tostring, setmetatable, loadstring local format, gsub, match = string.format, string.gsub, string.match local lpegmatch = lpeg.match --[[ldx--

The following helper functions best belong to the lxml-ini module. Some are here because we need then in the mk document and other manuals, others came up when playing with this module. Since this module is also used in we've put them here instead of loading mode modules there then needed.

--ldx]]-- local function xmlgsub(t,old,new) -- will be replaced local dt = t.dt if dt then for k=1,#dt do local v = dt[k] if type(v) == "string" then dt[k] = gsub(v,old,new) else xmlgsub(v,old,new) end end end end function xml.strip_leading_spaces(dk,d,k) -- cosmetic, for manual if d and k then local dkm = d[k-1] if dkm and type(dkm) == "string" then local s = match(dkm,"\n(%s+)") xmlgsub(dk,"\n"..rep(" ",#s),"\n") end end end local P, S, R, C, V, Cc, Cs = lpeg.P, lpeg.S, lpeg.R, lpeg.C, lpeg.V, lpeg.Cc, lpeg.Cs -- 100 * 2500 * "oeps< oeps> oeps&" : gsub:lpeg|lpeg|lpeg -- -- 1021:0335:0287:0247 -- 10 * 1000 * "oeps< oeps> oeps& asfjhalskfjh alskfjh alskfjh alskfjh ;al J;LSFDJ" -- -- 1559:0257:0288:0190 (last one suggested by roberto) -- escaped = Cs((S("<&>") / xml.escapes + 1)^0) -- escaped = Cs((S("<")/"<" + S(">")/">" + S("&")/"&" + 1)^0) local normal = (1 - S("<&>"))^0 local special = P("<")/"<" + P(">")/">" + P("&")/"&" local escaped = Cs(normal * (special * normal)^0) -- 100 * 1000 * "oeps< oeps> oeps&" : gsub:lpeg == 0153:0280:0151:0080 (last one by roberto) local normal = (1 - S"&")^0 local special = P("<")/"<" + P(">")/">" + P("&")/"&" local unescaped = Cs(normal * (special * normal)^0) -- 100 * 5000 * "oeps oeps oeps " : gsub:lpeg == 623:501 msec (short tags, less difference) local cleansed = Cs(((P("<") * (1-P(">"))^0 * P(">"))/"" + 1)^0) xml.escaped_pattern = escaped xml.unescaped_pattern = unescaped xml.cleansed_pattern = cleansed function xml.escaped (str) return lpegmatch(escaped,str) end function xml.unescaped(str) return lpegmatch(unescaped,str) end function xml.cleansed (str) return lpegmatch(cleansed,str) end -- this might move function xml.fillin(root,pattern,str,check) local e = xml.first(root,pattern) if e then local n = #e.dt if not check or n == 0 or (n == 1 and e.dt[1] == "") then e.dt = { str } end end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['lxml-aux'] = { version = 1.001, comment = "this module is the basis for the lxml-* ones", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- not all functions here make sense anymore vbut we keep them for -- compatibility reasons local trace_manipulations = false trackers.register("lxml.manipulations", function(v) trace_manipulations = v end) local report_xml = logs.new("xml") local xmlparseapply, xmlconvert, xmlcopy, xmlname = xml.parse_apply, xml.convert, xml.copy, xml.name local xmlinheritedconvert = xml.inheritedconvert local type = type local insert, remove = table.insert, table.remove local gmatch, gsub = string.gmatch, string.gsub local function report(what,pattern,c,e) report_xml("%s element '%s' (root: '%s', position: %s, index: %s, pattern: %s)",what,xmlname(e),xmlname(e.__p__),c,e.ni,pattern) end local function withelements(e,handle,depth) if e and handle then local edt = e.dt if edt then depth = depth or 0 for i=1,#edt do local e = edt[i] if type(e) == "table" then handle(e,depth) withelements(e,handle,depth+1) end end end end end xml.withelements = withelements function xml.withelement(e,n,handle) -- slow if e and n ~= 0 and handle then local edt = e.dt if edt then if n > 0 then for i=1,#edt do local ei = edt[i] if type(ei) == "table" then if n == 1 then handle(ei) return else n = n - 1 end end end elseif n < 0 then for i=#edt,1,-1 do local ei = edt[i] if type(ei) == "table" then if n == -1 then handle(ei) return else n = n + 1 end end end end end end end xml.elements_only = xml.collected function xml.each_element(root,pattern,handle,reverse) local collected = xmlparseapply({ root },pattern) if collected then if reverse then for c=#collected,1,-1 do handle(collected[c]) end else for c=1,#collected do handle(collected[c]) end end return collected end end xml.process_elements = xml.each_element function xml.process_attributes(root,pattern,handle) local collected = xmlparseapply({ root },pattern) if collected and handle then for c=1,#collected do handle(collected[c].at) end end return collected end --[[ldx--

The following functions collect elements and texts.

--ldx]]-- -- are these still needed -> lxml-cmp.lua function xml.collect_elements(root, pattern) return xmlparseapply({ root },pattern) end function xml.collect_texts(root, pattern, flatten) -- todo: variant with handle local collected = xmlparseapply({ root },pattern) if collected and flatten then local xmltostring = xml.tostring for c=1,#collected do collected[c] = xmltostring(collected[c].dt) end end return collected or { } end function xml.collect_tags(root, pattern, nonamespace) local collected = xmlparseapply({ root },pattern) if collected then local t = { } for c=1,#collected do local e = collected[c] local ns, tg = e.ns, e.tg if nonamespace then t[#t+1] = tg elseif ns == "" then t[#t+1] = tg else t[#t+1] = ns .. ":" .. tg end end return t end end --[[ldx--

We've now arrived at the functions that manipulate the tree.

--ldx]]-- local no_root = { no_root = true } function xml.redo_ni(d) for k=1,#d do local dk = d[k] if type(dk) == "table" then dk.ni = k end end end local function xmltoelement(whatever,root) if not whatever then return nil end local element if type(whatever) == "string" then element = xmlinheritedconvert(whatever,root) else element = whatever -- we assume a table end if element.error then return whatever -- string end if element then end return element end xml.toelement = xmltoelement local function copiedelement(element,newparent) if type(element) == "string" then return element else element = xmlcopy(element).dt if newparent and type(element) == "table" then element.__p__ = newparent end return element end end function xml.delete_element(root,pattern) local collected = xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] local p = e.__p__ if p then if trace_manipulations then report('deleting',pattern,c,e) end local d = p.dt remove(d,e.ni) xml.redo_ni(d) -- can be made faster and inlined end end end end function xml.replace_element(root,pattern,whatever) local element = root and xmltoelement(whatever,root) local collected = element and xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] local p = e.__p__ if p then if trace_manipulations then report('replacing',pattern,c,e) end local d = p.dt d[e.ni] = copiedelement(element,p) xml.redo_ni(d) -- probably not needed end end end end local function inject_element(root,pattern,whatever,prepend) local element = root and xmltoelement(whatever,root) local collected = element and xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] local r = e.__p__ local d, k, rri = r.dt, e.ni, r.ri local edt = (rri and d[rri].dt) or (d and d[k] and d[k].dt) if edt then local be, af local cp = copiedelement(element,e) if prepend then be, af = cp, edt else be, af = edt, cp end for i=1,#af do be[#be+1] = af[i] end if rri then r.dt[rri].dt = be else d[k].dt = be end xml.redo_ni(d) end end end end local function insert_element(root,pattern,whatever,before) -- todo: element als functie local element = root and xmltoelement(whatever,root) local collected = element and xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] local r = e.__p__ local d, k = r.dt, e.ni if not before then k = k + 1 end insert(d,k,copiedelement(element,r)) xml.redo_ni(d) end end end xml.insert_element = insert_element xml.insert_element_after = insert_element xml.insert_element_before = function(r,p,e) insert_element(r,p,e,true) end xml.inject_element = inject_element xml.inject_element_after = inject_element xml.inject_element_before = function(r,p,e) inject_element(r,p,e,true) end local function include(xmldata,pattern,attribute,recursive,loaddata) -- parse="text" (default: xml), encoding="" (todo) -- attribute = attribute or 'href' pattern = pattern or 'include' loaddata = loaddata or io.loaddata local collected = xmlparseapply({ xmldata },pattern) if collected then for c=1,#collected do local ek = collected[c] local name = nil local ekdt = ek.dt local ekat = ek.at local epdt = ek.__p__.dt if not attribute or attribute == "" then name = (type(ekdt) == "table" and ekdt[1]) or ekdt -- ckeck, probably always tab or str end if not name then for a in gmatch(attribute or "href","([^|]+)") do name = ekat[a] if name then break end end end local data = (name and name ~= "" and loaddata(name)) or "" if data == "" then epdt[ek.ni] = "" -- xml.empty(d,k) elseif ekat["parse"] == "text" then -- for the moment hard coded epdt[ek.ni] = xml.escaped(data) -- d[k] = xml.escaped(data) else local xi = xmlinheritedconvert(data,xmldata) if not xi then epdt[ek.ni] = "" -- xml.empty(d,k) else if recursive then include(xi,pattern,attribute,recursive,loaddata) end epdt[ek.ni] = xml.body(xi) -- xml.assign(d,k,xi) end end end end end xml.include = include function xml.strip_whitespace(root, pattern, nolines) -- strips all leading and trailing space ! local collected = xmlparseapply({ root },pattern) if collected then for i=1,#collected do local e = collected[i] local edt = e.dt if edt then local t = { } for i=1,#edt do local str = edt[i] if type(str) == "string" then if str == "" then -- stripped else if nolines then str = gsub(str,"[ \n\r\t]+"," ") end if str == "" then -- stripped else t[#t+1] = str end end else t[#t+1] = str end end e.dt = t end end end end function xml.strip_whitespace(root, pattern, nolines, anywhere) -- strips all leading and trailing spacing local collected = xmlparseapply({ root },pattern) -- beware, indices no longer are valid now if collected then for i=1,#collected do local e = collected[i] local edt = e.dt if edt then if anywhere then local t = { } for e=1,#edt do local str = edt[e] if type(str) ~= "string" then t[#t+1] = str elseif str ~= "" then -- todo: lpeg for each case if nolines then str = gsub(str,"%s+"," ") end str = gsub(str,"^%s*(.-)%s*$","%1") if str ~= "" then t[#t+1] = str end end end e.dt = t else -- we can assume a regular sparse xml table with no successive strings -- otherwise we should use a while loop if #edt > 0 then -- strip front local str = edt[1] if type(str) ~= "string" then -- nothing elseif str == "" then remove(edt,1) else if nolines then str = gsub(str,"%s+"," ") end str = gsub(str,"^%s+","") if str == "" then remove(edt,1) else edt[1] = str end end end if #edt > 1 then -- strip end local str = edt[#edt] if type(str) ~= "string" then -- nothing elseif str == "" then remove(edt) else if nolines then str = gsub(str,"%s+"," ") end str = gsub(str,"%s+$","") if str == "" then remove(edt) else edt[#edt] = str end end end end end end end end local function rename_space(root, oldspace, newspace) -- fast variant local ndt = #root.dt for i=1,ndt or 0 do local e = root[i] if type(e) == "table" then if e.ns == oldspace then e.ns = newspace if e.rn then e.rn = newspace end end local edt = e.dt if edt then rename_space(edt, oldspace, newspace) end end end end xml.rename_space = rename_space function xml.remap_tag(root, pattern, newtg) local collected = xmlparseapply({ root },pattern) if collected then for c=1,#collected do collected[c].tg = newtg end end end function xml.remap_namespace(root, pattern, newns) local collected = xmlparseapply({ root },pattern) if collected then for c=1,#collected do collected[c].ns = newns end end end function xml.check_namespace(root, pattern, newns) local collected = xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] if (not e.rn or e.rn == "") and e.ns == "" then e.rn = newns end end end end function xml.remap_name(root, pattern, newtg, newns, newrn) local collected = xmlparseapply({ root },pattern) if collected then for c=1,#collected do local e = collected[c] e.tg, e.ns, e.rn = newtg, newns, newrn end end end --[[ldx--

Here are a few synonyms.

--ldx]]-- xml.each = xml.each_element xml.process = xml.process_element xml.strip = xml.strip_whitespace xml.collect = xml.collect_elements xml.all = xml.collect_elements xml.insert = xml.insert_element_after xml.inject = xml.inject_element_after xml.after = xml.insert_element_after xml.before = xml.insert_element_before xml.delete = xml.delete_element xml.replace = xml.replace_element end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['lxml-xml'] = { version = 1.001, comment = "this module is the basis for the lxml-* ones", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local finalizers = xml.finalizers.xml local xmlfilter = xml.filter -- we could inline this one for speed local xmltostring = xml.tostring local xmlserialize = xml.serialize local xmlcollected = xml.collected local function first(collected) -- wrong ? return collected and collected[1] end local function last(collected) return collected and collected[#collected] end local function all(collected) return collected end local function reverse(collected) if collected then local reversed = { } for c=#collected,1,-1 do reversed[#reversed+1] = collected[c] end return reversed end end local function attribute(collected,name) if collected and #collected > 0 then local at = collected[1].at return at and at[name] end end local function att(id,name) local at = id.at return at and at[name] end local function count(collected) return (collected and #collected) or 0 end local function position(collected,n) if collected then n = tonumber(n) or 0 if n < 0 then return collected[#collected + n + 1] elseif n > 0 then return collected[n] else return collected[1].mi or 0 end end end local function match(collected) return (collected and collected[1].mi) or 0 -- match end local function index(collected) if collected then return collected[1].ni end end local function attributes(collected,arguments) if collected then local at = collected[1].at if arguments then return at[arguments] elseif next(at) then return at -- all of them end end end local function chainattribute(collected,arguments) -- todo: optional levels if collected then local e = collected[1] while e do local at = e.at if at then local a = at[arguments] if a then return a end else break -- error end e = e.__p__ end end return "" end local function raw(collected) -- hybrid if collected then local e = collected[1] or collected return (e and xmlserialize(e)) or "" -- only first as we cannot concat function else return "" end end local function text(collected) -- hybrid if collected then local e = collected[1] or collected return (e and xmltostring(e.dt)) or "" else return "" end end local function texts(collected) if collected then local t = { } for c=1,#collected do local e = collection[c] if e and e.dt then t[#t+1] = e.dt end end return t end end local function tag(collected,n) if collected then local c if n == 0 or not n then c = collected[1] elseif n > 1 then c = collected[n] else c = collected[#collected-n+1] end return c and c.tg end end local function name(collected,n) if collected then local c if n == 0 or not n then c = collected[1] elseif n > 1 then c = collected[n] else c = collected[#collected-n+1] end if c then if c.ns == "" then return c.tg else return c.ns .. ":" .. c.tg end end end end local function tags(collected,nonamespace) if collected then local t = { } for c=1,#collected do local e = collected[c] local ns, tg = e.ns, e.tg if nonamespace or ns == "" then t[#t+1] = tg else t[#t+1] = ns .. ":" .. tg end end return t end end local function empty(collected) if collected then for c=1,#collected do local e = collected[c] if e then local edt = e.dt if edt then local n = #edt if n == 1 then local edk = edt[1] local typ = type(edk) if typ == "table" then return false elseif edk ~= "" then -- maybe an extra tester for spacing only return false end elseif n > 1 then return false end end end end end return true end finalizers.first = first finalizers.last = last finalizers.all = all finalizers.reverse = reverse finalizers.elements = all finalizers.default = all finalizers.attribute = attribute finalizers.att = att finalizers.count = count finalizers.position = position finalizers.match = match finalizers.index = index finalizers.attributes = attributes finalizers.chainattribute = chainattribute finalizers.text = text finalizers.texts = texts finalizers.tag = tag finalizers.name = name finalizers.tags = tags finalizers.empty = empty -- shortcuts -- we could support xmlfilter(id,pattern,first) function xml.first(id,pattern) return first(xmlfilter(id,pattern)) end function xml.last(id,pattern) return last(xmlfilter(id,pattern)) end function xml.count(id,pattern) return count(xmlfilter(id,pattern)) end function xml.attribute(id,pattern,a,default) return attribute(xmlfilter(id,pattern),a,default) end function xml.raw(id,pattern) if pattern then return raw(xmlfilter(id,pattern)) else return raw(id) end end function xml.text(id,pattern) if pattern then -- return text(xmlfilter(id,pattern)) local collected = xmlfilter(id,pattern) return (collected and xmltostring(collected[1].dt)) or "" elseif id then -- return text(id) return xmltostring(id.dt) or "" else return "" end end xml.content = text function xml.position(id,pattern,n) -- element return position(xmlfilter(id,pattern),n) end function xml.match(id,pattern) -- number return match(xmlfilter(id,pattern)) end function xml.empty(id,pattern) return empty(xmlfilter(id,pattern)) end xml.all = xml.filter xml.index = xml.position xml.found = xml.filter -- a nice one: local function totable(x) local t = { } for e in xmlcollected(x[1] or x,"/*") do t[e.tg] = xmltostring(e.dt) or "" end return next(t) and t or nil end xml.table = totable finalizers.table = totable end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-ini'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files", } local gsub, find, gmatch = string.gsub, string.find, string.gmatch local concat = table.concat local next, type = next, type local filedirname, filebasename, fileextname, filejoin = file.dirname, file.basename, file.extname, file.join local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local trace_detail = false trackers.register("resolvers.details", function(v) trace_detail = v end) local trace_expansions = false trackers.register("resolvers.expansions", function(v) trace_expansions = v end) local report_resolvers = logs.new("resolvers") local ostype, osname, ossetenv, osgetenv = os.type, os.name, os.setenv, os.getenv -- The code here used to be part of a data-res but for convenience -- we now split it over multiple files. As this file is now the -- starting point we introduce resolvers here. resolvers = resolvers or { } -- We don't want the kpse library to kick in. Also, we want to be able to -- execute programs. Control over execution is implemented later. texconfig.kpse_init = false texconfig.shell_escape = 't' kpse = { original = kpse } setmetatable(kpse, { __index = function(kp,name) local r = resolvers[name] if not r then r = function (...) report_resolvers("not supported: %s(%s)",name,concat(...)) end rawset(kp,name,r) end return r end } ) -- First we check a couple of environment variables. Some might be -- set already but we need then later on. We start with the system -- font path. do local osfontdir = osgetenv("OSFONTDIR") if osfontdir and osfontdir ~= "" then -- ok elseif osname == "windows" then ossetenv("OSFONTDIR","c:/windows/fonts//") elseif osname == "macosx" then ossetenv("OSFONTDIR","$HOME/Library/Fonts//;/Library/Fonts//;/System/Library/Fonts//") end end -- Next comes the user's home path. We need this as later on we have -- to replace ~ with its value. do local homedir = osgetenv(ostype == "windows" and 'USERPROFILE' or 'HOME') or '~' homedir = file.collapse_path(homedir) ossetenv("HOME", homedir) -- can be used in unix cnf files ossetenv("USERPROFILE",homedir) -- can be used in windows cnf files environment.homedir = homedir end -- The following code sets the name of the own binary and its -- path. This is fallback code as we have os.selfdir now. do local args = environment.original_arguments or arg -- this needs a cleanup local ownbin = environment.ownbin or args[-2] or arg[-2] or args[-1] or arg[-1] or arg[0] or "luatex" local ownpath = environment.ownpath or os.selfdir ownbin = file.collapse_path(ownbin) ownpath = file.collapse_path(ownpath) if not ownpath or ownpath == "" or ownpath == "unset" then ownpath = args[-1] or arg[-1] ownpath = ownpath and filedirname(gsub(ownpath,"\\","/")) if not ownpath or ownpath == "" then ownpath = args[-0] or arg[-0] ownpath = ownpath and filedirname(gsub(ownpath,"\\","/")) end local binary = ownbin if not ownpath or ownpath == "" then ownpath = ownpath and filedirname(binary) end if not ownpath or ownpath == "" then if os.binsuffix ~= "" then binary = file.replacesuffix(binary,os.binsuffix) end local path = osgetenv("PATH") if path then for p in gmatch(path,"[^"..io.pathseparator.."]+") do local b = filejoin(p,binary) if lfs.isfile(b) then -- we assume that after changing to the path the currentdir function -- resolves to the real location and use this side effect here; this -- trick is needed because on the mac installations use symlinks in the -- path instead of real locations local olddir = lfs.currentdir() if lfs.chdir(p) then local pp = lfs.currentdir() if trace_locating and p ~= pp then report_resolvers("following symlink '%s' to '%s'",p,pp) end ownpath = pp lfs.chdir(olddir) else if trace_locating then report_resolvers("unable to check path '%s'",p) end ownpath = p end break end end end end if not ownpath or ownpath == "" then ownpath = "." report_resolvers("forcing fallback ownpath .") elseif trace_locating then report_resolvers("using ownpath '%s'",ownpath) end end environment.ownbin = ownbin environment.ownpath = ownpath end resolvers.ownpath = environment.ownpath function resolvers.getownpath() return environment.ownpath end -- The self variables permit us to use only a few (or even no) -- environment variables. do local ownpath = environment.ownpath or dir.current() if ownpath then ossetenv('SELFAUTOLOC', file.collapse_path(ownpath)) ossetenv('SELFAUTODIR', file.collapse_path(ownpath .. "/..")) ossetenv('SELFAUTOPARENT', file.collapse_path(ownpath .. "/../..")) else report_resolvers("error: unable to locate ownpath") os.exit() end end -- The running os: -- todo: check is context sits here os.platform is more trustworthy -- that the bin check as mtx-update runs from another path local texos = environment.texos or osgetenv("TEXOS") local texmfos = environment.texmfos or osgetenv('SELFAUTODIR') if not texos or texos == "" then texos = file.basename(texmfos) end ossetenv('TEXMFOS', texmfos) -- full bin path ossetenv('TEXOS', texos) -- partial bin parent ossetenv('SELFAUTOSYSTEM',os.platform) -- bonus environment.texos = texos environment.texmfos = texmfos -- The current root: local texroot = environment.texroot or osgetenv("TEXROOT") if not texroot or texroot == "" then texroot = osgetenv('SELFAUTOPARENT') ossetenv('TEXROOT',texroot) end environment.texroot = file.collapse_path(texroot) -- Tracing. Todo ... function resolvers.settrace(n) -- no longer number but: 'locating' or 'detail' if n then trackers.disable("resolvers.*") trackers.enable("resolvers."..n) end end resolvers.settrace(osgetenv("MTX_INPUT_TRACE")) -- todo: -- if profiler and osgetenv("MTX_PROFILE_RUN") == "YES" then -- profiler.start("luatex-profile.log") -- end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-exp'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files", } local format, gsub, find, gmatch, lower = string.format, string.gsub, string.find, string.gmatch, string.lower local concat, sort = table.concat, table.sort local lpegmatch, lpegpatterns = lpeg.match, lpeg.patterns local lpegCt, lpegCs, lpegP, lpegC, lpegS = lpeg.Ct, lpeg.Cs, lpeg.P, lpeg.C, lpeg.S local type, next = type, next local ostype = os.type local collapse_path = file.collapse_path local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local trace_expansions = false trackers.register("resolvers.expansions", function(v) trace_expansions = v end) local report_resolvers = logs.new("resolvers") -- As this bit of code is somewhat special it gets its own module. After -- all, when working on the main resolver code, I don't want to scroll -- past this every time. -- {a,b,c,d} -- a,b,c/{p,q,r},d -- a,b,c/{p,q,r}/d/{x,y,z}// -- a,b,c/{p,q/{x,y,z},r},d/{p,q,r} -- a,b,c/{p,q/{x,y,z},r},d/{p,q,r} -- a{b,c}{d,e}f -- {a,b,c,d} -- {a,b,c/{p,q,r},d} -- {a,b,c/{p,q,r}/d/{x,y,z}//} -- {a,b,c/{p,q/{x,y,z}},d/{p,q,r}} -- {a,b,c/{p,q/{x,y,z},w}v,d/{p,q,r}} -- {$SELFAUTODIR,$SELFAUTOPARENT}{,{/share,}/texmf{-local,.local,}/web2c} -- this one is better and faster, but it took me a while to realize -- that this kind of replacement is cleaner than messy parsing and -- fuzzy concatenating we can probably gain a bit with selectively -- applying lpeg, but experiments with lpeg parsing this proved not to -- work that well; the parsing is ok, but dealing with the resulting -- table is a pain because we need to work inside-out recursively local dummy_path_expr = "^!*unset/*$" local function do_first(a,b) local t = { } for s in gmatch(b,"[^,]+") do t[#t+1] = a .. s end return "{" .. concat(t,",") .. "}" end local function do_second(a,b) local t = { } for s in gmatch(a,"[^,]+") do t[#t+1] = s .. b end return "{" .. concat(t,",") .. "}" end local function do_both(a,b) local t = { } for sa in gmatch(a,"[^,]+") do for sb in gmatch(b,"[^,]+") do t[#t+1] = sa .. sb end end return "{" .. concat(t,",") .. "}" end local function do_three(a,b,c) return a .. b.. c end local stripper_1 = lpeg.stripper("{}@") local replacer_1 = lpeg.replacer { { ",}", ",@}" }, { "{,", "{@," }, } local function splitpathexpr(str, newlist, validate) -- no need for further optimization as it is only called a -- few times, we can use lpeg for the sub if trace_expansions then report_resolvers("expanding variable '%s'",str) end local t, ok, done = newlist or { }, false, false str = lpegmatch(replacer_1,str) while true do done = false while true do str, ok = gsub(str,"([^{},]+){([^{}]+)}",do_first) if ok > 0 then done = true else break end end while true do str, ok = gsub(str,"{([^{}]+)}([^{},]+)",do_second) if ok > 0 then done = true else break end end while true do str, ok = gsub(str,"{([^{}]+)}{([^{}]+)}",do_both) if ok > 0 then done = true else break end end str, ok = gsub(str,"({[^{}]*){([^{}]+)}([^{}]*})",do_three) if ok > 0 then done = true end if not done then break end end str = lpegmatch(stripper_1,str) if validate then for s in gmatch(str,"[^,]+") do s = validate(s) if s then t[#t+1] = s end end else for s in gmatch(str,"[^,]+") do t[#t+1] = s end end if trace_expansions then for k=1,#t do report_resolvers("% 4i: %s",k,t[k]) end end return t end local function validate(s) local isrecursive = find(s,"//$") s = collapse_path(s) if isrecursive then s = s .. "//" end return s ~= "" and not find(s,dummy_path_expr) and s end resolvers.validated_path = validate -- keeps the trailing // function resolvers.expanded_path_from_list(pathlist) -- maybe not a list, just a path -- a previous version fed back into pathlist local newlist, ok = { }, false for k=1,#pathlist do if find(pathlist[k],"[{}]") then ok = true break end end if ok then for k=1,#pathlist do splitpathexpr(pathlist[k],newlist,validate) end else for k=1,#pathlist do for p in gmatch(pathlist[k],"([^,]+)") do p = validate(p) if p ~= "" then newlist[#newlist+1] = p end end end end return newlist end -- We also put some cleanup code here. local cleanup -- used recursively cleanup = lpeg.replacer { { "!", "" }, { "\\", "/" }, { "~" , function() return lpegmatch(cleanup,environment.homedir) end }, } function resolvers.clean_path(str) return str and lpegmatch(cleanup,str) end -- This one strips quotes and funny tokens. local expandhome = lpegP("~") / "$HOME" -- environment.homedir local dodouble = lpegP('"')/"" * (expandhome + (1 - lpegP('"')))^0 * lpegP('"')/"" local dosingle = lpegP("'")/"" * (expandhome + (1 - lpegP("'")))^0 * lpegP("'")/"" local dostring = (expandhome + 1 )^0 local stripper = lpegCs( lpegpatterns.unspacer * (dosingle + dodouble + dostring) * lpegpatterns.unspacer ) function resolvers.checked_variable(str) -- assumes str is a string return lpegmatch(stripper,str) or str end -- The path splitter: -- A config (optionally) has the paths split in tables. Internally -- we join them and split them after the expansion has taken place. This -- is more convenient. local cache = { } local splitter = lpegCt(lpeg.splitat(lpegS(ostype == "windows" and ";" or ":;"))) -- maybe add , local function split_configuration_path(str) -- beware, this can be either a path or a { specification } if str then local found = cache[str] if not found then if str == "" then found = { } else str = gsub(str,"\\","/") local split = lpegmatch(splitter,str) found = { } for i=1,#split do local s = split[i] if not find(s,"^{*unset}*") then found[#found+1] = s end end if trace_expansions then report_resolvers("splitting path specification '%s'",str) for k=1,#found do report_resolvers("% 4i: %s",k,found[k]) end end cache[str] = found end end return found end end resolvers.split_configuration_path = split_configuration_path function resolvers.split_path(str) if type(str) == 'table' then return str else return split_configuration_path(str) end end function resolvers.join_path(str) if type(str) == 'table' then return file.join_path(str) else return str end end -- The next function scans directories and returns a hash where the -- entries are either strings or tables. -- starting with . or .. etc or funny char local weird = lpegP(".")^1 + lpeg.anywhere(lpegS("~`!#$%^&*()={}[]:;\"\'||<>,?\n\r\t")) function resolvers.scan_files(specification) if trace_locating then report_resolvers("scanning path '%s'",specification) end local attributes, directory = lfs.attributes, lfs.dir local files = { __path__ = specification } local n, m, r = 0, 0, 0 local function scan(spec,path) local full = (path == "" and spec) or (spec .. path .. '/') local dirs = { } for name in directory(full) do if not lpegmatch(weird,name) then local mode = attributes(full..name,'mode') if mode == 'file' then n = n + 1 local f = files[name] if f then if type(f) == 'string' then files[name] = { f, path } else f[#f+1] = path end else -- probably unique anyway files[name] = path local lower = lower(name) if name ~= lower then files["remap:"..lower] = name r = r + 1 end end elseif mode == 'directory' then m = m + 1 if path ~= "" then dirs[#dirs+1] = path..'/'..name else dirs[#dirs+1] = name end end end end if #dirs > 0 then sort(dirs) for i=1,#dirs do scan(spec,dirs[i]) end end end scan(specification .. '/',"") files.__files__, files.__directories__, files.__remappings__ = n, m, r if trace_locating then report_resolvers("%s files found on %s directories with %s uppercase remappings",n,m,r) end return files end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-env'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files", } local formats = { } resolvers.formats = formats local suffixes = { } resolvers.suffixes = suffixes local dangerous = { } resolvers.dangerous = dangerous local suffixmap = { } resolvers.suffixmap = suffixmap local alternatives = { } resolvers.alternatives = alternatives formats['afm'] = 'AFMFONTS' suffixes['afm'] = { 'afm' } formats['enc'] = 'ENCFONTS' suffixes['enc'] = { 'enc' } formats['fmt'] = 'TEXFORMATS' suffixes['fmt'] = { 'fmt' } formats['map'] = 'TEXFONTMAPS' suffixes['map'] = { 'map' } formats['mp'] = 'MPINPUTS' suffixes['mp'] = { 'mp' } formats['ocp'] = 'OCPINPUTS' suffixes['ocp'] = { 'ocp' } formats['ofm'] = 'OFMFONTS' suffixes['ofm'] = { 'ofm', 'tfm' } formats['otf'] = 'OPENTYPEFONTS' suffixes['otf'] = { 'otf' } formats['opl'] = 'OPLFONTS' suffixes['opl'] = { 'opl' } formats['otp'] = 'OTPINPUTS' suffixes['otp'] = { 'otp' } formats['ovf'] = 'OVFFONTS' suffixes['ovf'] = { 'ovf', 'vf' } formats['ovp'] = 'OVPFONTS' suffixes['ovp'] = { 'ovp' } formats['tex'] = 'TEXINPUTS' suffixes['tex'] = { 'tex' } formats['tfm'] = 'TFMFONTS' suffixes['tfm'] = { 'tfm' } formats['ttf'] = 'TTFONTS' suffixes['ttf'] = { 'ttf', 'ttc', 'dfont' } formats['pfb'] = 'T1FONTS' suffixes['pfb'] = { 'pfb', 'pfa' } formats['vf'] = 'VFFONTS' suffixes['vf'] = { 'vf' } formats['fea'] = 'FONTFEATURES' suffixes['fea'] = { 'fea' } formats['cid'] = 'FONTCIDMAPS' suffixes['cid'] = { 'cid', 'cidmap' } formats['icc'] = 'ICCPROFILES' suffixes['icc'] = { 'icc' } formats['texmfscripts'] = 'TEXMFSCRIPTS' suffixes['texmfscripts'] = { 'rb', 'pl', 'py' } formats['lua'] = 'LUAINPUTS' suffixes['lua'] = { 'lua', 'luc', 'tma', 'tmc' } formats['lib'] = 'CLUAINPUTS' suffixes['lib'] = (os.libsuffix and { os.libsuffix }) or { 'dll', 'so' } -- backward compatible ones alternatives['map files'] = 'map' alternatives['enc files'] = 'enc' alternatives['cid maps'] = 'cid' -- great, why no cid files alternatives['font feature files'] = 'fea' -- and fea files here alternatives['opentype fonts'] = 'otf' alternatives['truetype fonts'] = 'ttf' alternatives['truetype collections'] = 'ttc' alternatives['truetype dictionary'] = 'dfont' alternatives['type1 fonts'] = 'pfb' alternatives['icc profiles'] = 'icc' --[[ldx--

If you wondered about some of the previous mappings, how about the next bunch:

--ldx]]-- -- kpse specific ones (a few omitted) .. we only add them for locating -- files that we don't use anyway formats['base'] = 'MFBASES' suffixes['base'] = { 'base', 'bas' } formats['bib'] = '' suffixes['bib'] = { 'bib' } formats['bitmap font'] = '' suffixes['bitmap font'] = { } formats['bst'] = '' suffixes['bst'] = { 'bst' } formats['cmap files'] = 'CMAPFONTS' suffixes['cmap files'] = { 'cmap' } formats['cnf'] = '' suffixes['cnf'] = { 'cnf' } formats['cweb'] = '' suffixes['cweb'] = { 'w', 'web', 'ch' } formats['dvips config'] = '' suffixes['dvips config'] = { } formats['gf'] = '' suffixes['gf'] = { 'gf' } formats['graphic/figure'] = '' suffixes['graphic/figure'] = { 'eps', 'epsi' } formats['ist'] = '' suffixes['ist'] = { 'ist' } formats['lig files'] = 'LIGFONTS' suffixes['lig files'] = { 'lig' } formats['ls-R'] = '' suffixes['ls-R'] = { } formats['mem'] = 'MPMEMS' suffixes['mem'] = { 'mem' } formats['MetaPost support'] = '' suffixes['MetaPost support'] = { } formats['mf'] = 'MFINPUTS' suffixes['mf'] = { 'mf' } formats['mft'] = '' suffixes['mft'] = { 'mft' } formats['misc fonts'] = '' suffixes['misc fonts'] = { } formats['other text files'] = '' suffixes['other text files'] = { } formats['other binary files'] = '' suffixes['other binary files'] = { } formats['pdftex config'] = 'PDFTEXCONFIG' suffixes['pdftex config'] = { } formats['pk'] = '' suffixes['pk'] = { 'pk' } formats['PostScript header'] = 'TEXPSHEADERS' suffixes['PostScript header'] = { 'pro' } formats['sfd'] = 'SFDFONTS' suffixes['sfd'] = { 'sfd' } formats['TeX system documentation'] = '' suffixes['TeX system documentation'] = { } formats['TeX system sources'] = '' suffixes['TeX system sources'] = { } formats['Troff fonts'] = '' suffixes['Troff fonts'] = { } formats['type42 fonts'] = 'T42FONTS' suffixes['type42 fonts'] = { } formats['web'] = '' suffixes['web'] = { 'web', 'ch' } formats['web2c files'] = 'WEB2C' suffixes['web2c files'] = { } formats['fontconfig files'] = 'FONTCONFIG_PATH' suffixes['fontconfig files'] = { } -- not unique alternatives['subfont definition files'] = 'sfd' -- A few accessors, mostly for command line tool. function resolvers.suffix_of_format(str) local s = suffixes[str] return s and s[1] or "" end function resolvers.suffixes_of_format(str) return suffixes[str] or { } end -- As we don't register additional suffixes anyway, we can as well -- freeze the reverse map here. for name, suffixlist in next, suffixes do for i=1,#suffixlist do suffixmap[suffixlist[i]] = name end end setmetatable(suffixes, { __newindex = function(suffixes,name,suffixlist) rawset(suffixes,name,suffixlist) suffixes[name] = suffixlist for i=1,#suffixlist do suffixmap[suffixlist[i]] = name end end } ) for name, format in next, formats do dangerous[name] = true end -- because vf searching is somewhat dangerous, we want to prevent -- too liberal searching esp because we do a lookup on the current -- path anyway; only tex (or any) is safe dangerous.tex = nil -- more helpers function resolvers.format_of_var(str) return formats[str] or formats[alternatives[str]] or '' end function resolvers.format_of_suffix(str) -- of file return suffixmap[file.extname(str)] or 'tex' end function resolvers.variable_of_format(str) return formats[str] or formats[alternatives[str]] or '' end function resolvers.var_of_format_or_suffix(str) local v = formats[str] if v then return v end v = formats[alternatives[str]] if v then return v end v = suffixmap[fileextname(str)] if v then return formats[v] end return '' end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-tmp'] = { version = 1.100, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } --[[ldx--

This module deals with caching data. It sets up the paths and implements loaders and savers for tables. Best is to set the following variable. When not set, the usual paths will be checked. Personally I prefer the (users) temporary path.

TEXMFCACHE=$TMP;$TEMP;$TMPDIR;$TEMPDIR;$HOME;$TEXMFVAR;$VARTEXMF;.

Currently we do no locking when we write files. This is no real problem because most caching involves fonts and the chance of them being written at the same time is small. We also need to extend luatools with a recache feature.

--ldx]]-- local format, lower, gsub, concat = string.format, string.lower, string.gsub, table.concat local mkdirs, isdir = dir.mkdirs, lfs.isdir local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local trace_cache = false trackers.register("resolvers.cache", function(v) trace_cache = v end) local report_cache = logs.new("cache") local report_resolvers = logs.new("resolvers") caches = caches or { } caches.base = caches.base or "luatex-cache" caches.more = caches.more or "context" caches.direct = false -- true is faster but may need huge amounts of memory caches.tree = false caches.force = true caches.ask = false caches.defaults = { "TMPDIR", "TEMPDIR", "TMP", "TEMP", "HOME", "HOMEPATH" } local writable, readables, usedreadables = nil, { }, { } -- we could use a metatable for writable and readable but not yet local function identify() -- Combining the loops makes it messy. First we check the format cache path -- and when the last component is not present we try to create it. local texmfcaches = resolvers.clean_path_list("TEXMFCACHE") if texmfcaches then for k=1,#texmfcaches do local cachepath = texmfcaches[k] if cachepath ~= "" then cachepath = resolvers.clean_path(cachepath) cachepath = file.collapse_path(cachepath) local valid = isdir(cachepath) if valid then if file.isreadable(cachepath) then readables[#readables+1] = cachepath if not writable and file.iswritable(cachepath) then writable = cachepath end end elseif not writable and caches.force then local cacheparent = file.dirname(cachepath) if file.iswritable(cacheparent) then if not caches.ask or io.ask(format("\nShould I create the cache path %s?",cachepath), "no", { "yes", "no" }) == "yes" then mkdirs(cachepath) if isdir(cachepath) and file.iswritable(cachepath) then report_cache("created: %s",cachepath) writable = cachepath readables[#readables+1] = cachepath end end end end end end end -- As a last resort we check some temporary paths but this time we don't -- create them. local texmfcaches = caches.defaults if texmfcaches then for k=1,#texmfcaches do local cachepath = texmfcaches[k] cachepath = resolvers.getenv(cachepath) if cachepath ~= "" then cachepath = resolvers.clean_path(cachepath) local valid = isdir(cachepath) if valid and file.isreadable(cachepath) then if not writable and file.iswritable(cachepath) then readables[#readables+1] = cachepath writable = cachepath break end end end end end -- Some extra checking. If we have no writable or readable path then we simply -- quit. if not writable then report_cache("fatal error: there is no valid writable cache path defined") os.exit() elseif #readables == 0 then report_cache("fatal error: there is no valid readable cache path defined") os.exit() end -- why here writable = dir.expand_name(resolvers.clean_path(writable)) -- just in case -- moved here local base, more, tree = caches.base, caches.more, caches.tree or caches.treehash() -- we have only one writable tree if tree then caches.tree = tree writable = mkdirs(writable,base,more,tree) for i=1,#readables do readables[i] = file.join(readables[i],base,more,tree) end else writable = mkdirs(writable,base,more) for i=1,#readables do readables[i] = file.join(readables[i],base,more) end end -- end if trace_cache then for i=1,#readables do report_cache("using readable path '%s' (order %s)",readables[i],i) end report_cache("using writable path '%s'",writable) end identify = function() return writable, readables end return writable, readables end function caches.usedpaths() local writable, readables = identify() if #readables > 1 then local result = { } for i=1,#readables do local readable = readables[i] if usedreadables[i] or readable == writable then result[#result+1] = format("readable: '%s' (order %s)",readable,i) end end result[#result+1] = format("writable: '%s'",writable) return result else return writable end end function caches.configfiles() return table.concat(resolvers.instance.specification,";") end function caches.hashed(tree) return md5.hex(gsub(lower(tree),"[\\\/]+","/")) end function caches.treehash() local tree = caches.configfiles() if not tree or tree == "" then return false else return caches.hashed(tree) end end local r_cache, w_cache = { }, { } -- normally w in in r but who cares local function getreadablepaths(...) -- we can optimize this as we have at most 2 tags local tags = { ... } local hash = concat(tags,"/") local done = r_cache[hash] if not done then local writable, readables = identify() -- exit if not found if #tags > 0 then done = { } for i=1,#readables do done[i] = file.join(readables[i],...) end else done = readables end r_cache[hash] = done end return done end local function getwritablepath(...) local tags = { ... } local hash = concat(tags,"/") local done = w_cache[hash] if not done then local writable, readables = identify() -- exit if not found if #tags > 0 then done = mkdirs(writable,...) else done = writable end w_cache[hash] = done end return done end caches.getreadablepaths = getreadablepaths caches.getwritablepath = getwritablepath function caches.getfirstreadablefile(filename,...) local rd = getreadablepaths(...) for i=1,#rd do local path = rd[i] local fullname = file.join(path,filename) if file.isreadable(fullname) then usedreadables[i] = true return fullname, path end end return caches.setfirstwritablefile(filename,...) end function caches.setfirstwritablefile(filename,...) local wr = getwritablepath(...) local fullname = file.join(wr,filename) return fullname, wr end function caches.define(category,subcategory) -- for old times sake return function() return getwritablepath(category,subcategory) end end function caches.setluanames(path,name) return path .. "/" .. name .. ".tma", path .. "/" .. name .. ".tmc" end function caches.loaddata(readables,name) if type(readables) == "string" then readables = { readables } end for i=1,#readables do local path = readables[i] local tmaname, tmcname = caches.setluanames(path,name) local loader = loadfile(tmcname) or loadfile(tmaname) if loader then loader = loader() collectgarbage("step") return loader end end return false end function caches.iswritable(filepath,filename) local tmaname, tmcname = caches.setluanames(filepath,filename) return file.iswritable(tmaname) end function caches.savedata(filepath,filename,data,raw) local tmaname, tmcname = caches.setluanames(filepath,filename) local reduce, simplify = true, true if raw then reduce, simplify = false, false end data.cache_uuid = os.uuid() if caches.direct then file.savedata(tmaname, table.serialize(data,'return',false,true,false)) -- no hex else table.tofile(tmaname, data,'return',false,true,false) -- maybe not the last true end local cleanup = resolvers.boolean_variable("PURGECACHE", false) local strip = resolvers.boolean_variable("LUACSTRIP", true) utils.lua.compile(tmaname, tmcname, cleanup, strip) end -- moved from data-res: local content_state = { } function caches.contentstate() return content_state or { } end function caches.loadcontent(cachename,dataname) local name = caches.hashed(cachename) local full, path = caches.getfirstreadablefile(name ..".lua","trees") local filename = file.join(path,name) local blob = loadfile(filename .. ".luc") or loadfile(filename .. ".lua") if blob then local data = blob() if data and data.content and data.type == dataname and data.version == resolvers.cacheversion then content_state[#content_state+1] = data.uuid if trace_locating then report_resolvers("loading '%s' for '%s' from '%s'",dataname,cachename,filename) end return data.content elseif trace_locating then report_resolvers("skipping '%s' for '%s' from '%s'",dataname,cachename,filename) end elseif trace_locating then report_resolvers("skipping '%s' for '%s' from '%s'",dataname,cachename,filename) end end function caches.collapsecontent(content) for k, v in next, content do if type(v) == "table" and #v == 1 then content[k] = v[1] end end end function caches.savecontent(cachename,dataname,content) local name = caches.hashed(cachename) local full, path = caches.setfirstwritablefile(name ..".lua","trees") local filename = file.join(path,name) -- is full local luaname, lucname = filename .. ".lua", filename .. ".luc" if trace_locating then report_resolvers("preparing '%s' for '%s'",dataname,cachename) end local data = { type = dataname, root = cachename, version = resolvers.cacheversion, date = os.date("%Y-%m-%d"), time = os.date("%H:%M:%S"), content = content, uuid = os.uuid(), } local ok = io.savedata(luaname,table.serialize(data,true)) if ok then if trace_locating then report_resolvers("category '%s', cachename '%s' saved in '%s'",dataname,cachename,luaname) end if utils.lua.compile(luaname,lucname,false,true) then -- no cleanup but strip if trace_locating then report_resolvers("'%s' compiled to '%s'",dataname,lucname) end return true else if trace_locating then report_resolvers("compiling failed for '%s', deleting file '%s'",dataname,lucname) end os.remove(lucname) end elseif trace_locating then report_resolvers("unable to save '%s' in '%s' (access error)",dataname,luaname) end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-met'] = { version = 1.100, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local find = string.find local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") resolvers.locators = { notfound = { nil } } -- locate databases resolvers.hashers = { notfound = { nil } } -- load databases resolvers.generators = { notfound = { nil } } -- generate databases function resolvers.splitmethod(filename) if not filename then return { } -- safeguard elseif type(filename) == "table" then return filename -- already split elseif not find(filename,"://") then return { scheme="file", path = filename, original = filename } -- quick hack else return url.hashed(filename) end end function resolvers.methodhandler(what, filename, filetype) -- ... filename = file.collapse_path(filename) local specification = (type(filename) == "string" and resolvers.splitmethod(filename)) or filename -- no or { }, let it bomb local scheme = specification.scheme local resolver = resolvers[what] if resolver[scheme] then if trace_locating then report_resolvers("handler '%s' -> '%s' -> '%s'",specification.original,what,table.sequenced(specification)) end return resolver[scheme](filename,filetype) else return resolver.tex(filename,filetype) -- todo: specification end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-res'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files", } -- In practice we will work within one tds tree, but i want to keep -- the option open to build tools that look at multiple trees, which is -- why we keep the tree specific data in a table. We used to pass the -- instance but for practical purposes we now avoid this and use a -- instance variable. We always have one instance active (sort of global). -- todo: cache:/// home:/// local format, gsub, find, lower, upper, match, gmatch = string.format, string.gsub, string.find, string.lower, string.upper, string.match, string.gmatch local concat, insert, sortedkeys = table.concat, table.insert, table.sortedkeys local next, type = next, type local lpegP, lpegS, lpegR, lpegC, lpegCc, lpegCs, lpegCt = lpeg.P, lpeg.S, lpeg.R, lpeg.C, lpeg.Cc, lpeg.Cs, lpeg.Ct local lpegmatch, lpegpatterns = lpeg.match, lpeg.patterns local filedirname, filebasename, fileextname, filejoin = file.dirname, file.basename, file.extname, file.join local collapse_path = file.collapse_path local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local trace_detail = false trackers.register("resolvers.details", function(v) trace_detail = v end) local trace_expansions = false trackers.register("resolvers.expansions", function(v) trace_expansions = v end) local report_resolvers = logs.new("resolvers") local expanded_path_from_list = resolvers.expanded_path_from_list local checked_variable = resolvers.checked_variable local split_configuration_path = resolvers.split_configuration_path local ostype, osname, osenv, ossetenv, osgetenv = os.type, os.name, os.env, os.setenv, os.getenv resolvers.cacheversion = '1.0.1' resolvers.configbanner = '' resolvers.homedir = environment.homedir resolvers.criticalvars = { "SELFAUTOLOC", "SELFAUTODIR", "SELFAUTOPARENT", "TEXMFCNF", "TEXMF", "TEXOS" } resolvers.luacnfspec = '{$SELFAUTODIR,$SELFAUTOPARENT}{,{/share,}/texmf{-local,}/web2c}' -- rubish path resolvers.luacnfname = 'texmfcnf.lua' resolvers.luacnfstate = "unknown" local unset_variable = "unset" local formats = resolvers.formats local suffixes = resolvers.suffixes local dangerous = resolvers.dangerous local suffixmap = resolvers.suffixmap local alternatives = resolvers.alternatives resolvers.instance = resolvers.instance or nil -- the current one (slow access) local instance = resolvers.instance or nil -- the current one (fast access) function resolvers.newinstance() local newinstance = { progname = 'context', engine = 'luatex', format = '', environment = { }, variables = { }, expansions = { }, files = { }, setups = { }, order = { }, found = { }, foundintrees = { }, origins = { }, hashes = { }, specification = { }, lists = { }, remember = true, diskcache = true, renewcache = false, loaderror = false, savelists = true, allresults = false, pattern = nil, -- lists data = { }, -- only for loading force_suffixes = true, fakepaths = { }, } local ne = newinstance.environment for k, v in next, osenv do ne[upper(k)] = checked_variable(v) end return newinstance end function resolvers.setinstance(someinstance) instance = someinstance resolvers.instance = someinstance return someinstance end function resolvers.reset() return resolvers.setinstance(resolvers.newinstance()) end local function reset_hashes() instance.lists = { } instance.found = { } end function resolvers.setenv(key,value) if instance then instance.environment[key] = value ossetenv(key,value) end end function resolvers.getenv(key) local value = instance.environment[key] if value and value ~= "" then return value else local e = osgetenv(key) return e ~= nil and e ~= "" and checked_variable(e) or "" end end resolvers.env = resolvers.getenv local function expand_vars(lst) -- simple vars local variables, getenv = instance.variables, resolvers.getenv local function resolve(a) local va = variables[a] or "" return (va ~= "" and va) or getenv(a) or "" end for k=1,#lst do local var = lst[k] var = gsub(var,"%$([%a%d%_%-]+)",resolve) var = gsub(var,";+",";") var = gsub(var,";[!{}/\\]+;",";") lst[k] = var end end local function resolve(key) local value = instance.variables[key] if value and value ~= "" then return value end local value = instance.environment[key] if value and value ~= "" then return value end local e = osgetenv(key) return e ~= nil and e ~= "" and checked_variable(e) or "" end local function expanded_var(var) -- simple vars var = gsub(var,"%$([%a%d%_%-]+)",resolve) var = gsub(var,";+",";") var = gsub(var,";[!{}/\\]+;",";") return var end local function entry(entries,name) if name and name ~= "" then name = gsub(name,'%$','') local result = entries[name..'.'..instance.progname] or entries[name] if result then return result else result = resolvers.getenv(name) if result then instance.variables[name] = result resolvers.expand_variables() return instance.expansions[name] or "" end end end return "" end local function is_entry(entries,name) if name and name ~= "" then name = gsub(name,'%$','') return (entries[name..'.'..instance.progname] or entries[name]) ~= nil else return false end end function resolvers.report_critical_variables() if trace_locating then for i=1,#resolvers.criticalvars do local v = resolvers.criticalvars[i] report_resolvers("variable '%s' set to '%s'",v,resolvers.getenv(v) or "unknown") end report_resolvers() end resolvers.report_critical_variables = function() end end local function identify_configuration_files() local specification = instance.specification if #specification == 0 then local cnfspec = resolvers.getenv('TEXMFCNF') if cnfspec == "" then cnfspec = resolvers.luacnfspec resolvers.luacnfstate = "default" else resolvers.luacnfstate = "environment" end resolvers.report_critical_variables() resolvers.expand_variables() local cnfpaths = expanded_path_from_list(resolvers.split_path(cnfspec)) expand_vars(cnfpaths) --- hm local luacnfname = resolvers.luacnfname for i=1,#cnfpaths do local filename = collapse_path(filejoin(cnfpaths[i],luacnfname)) if lfs.isfile(filename) then specification[#specification+1] = filename end end end end local function load_configuration_files() local specification = instance.specification if #specification > 0 then local luacnfname = resolvers.luacnfname for i=1,#specification do local filename = specification[i] local pathname = filedirname(filename) local filename = filejoin(pathname,luacnfname) local blob = loadfile(filename) if blob then local data = blob() data = data and data.content local setups = instance.setups if data then if trace_locating then report_resolvers("loading configuration file '%s'",filename) report_resolvers() end -- flattening is easier to deal with as we need to collapse local t = { } for k, v in next, data do -- v = progname if v ~= unset_variable then local kind = type(v) if kind == "string" then t[k] = v elseif kind == "table" then -- this operates on the table directly setters.initialize(filename,k,v) -- this doesn't (maybe metatables some day) for kk, vv in next, v do -- vv = variable if vv ~= unset_variable then if type(vv) == "string" then t[kk.."."..k] = vv end end end else -- report_resolvers("strange key '%s' in configuration file '%s'",k,filename) end end end setups[pathname] = t if resolvers.luacnfstate == "default" then -- the following code is not tested local cnfspec = t["TEXMFCNF"] if cnfspec then -- we push the value into the main environment (osenv) so -- that it takes precedence over the default one and therefore -- also over following definitions resolvers.setenv('TEXMFCNF',cnfspec) -- we now identify and load the specified configuration files instance.specification = { } identify_configuration_files() load_configuration_files() -- we prevent further overload of the configuration variable resolvers.luacnfstate = "configuration" -- we quit the outer loop break end end else if trace_locating then report_resolvers("skipping configuration file '%s'",filename) end setups[pathname] = { } instance.loaderror = true end elseif trace_locating then report_resolvers("skipping configuration file '%s'",filename) end instance.order[#instance.order+1] = instance.setups[pathname] if instance.loaderror then break end end elseif trace_locating then report_resolvers("warning: no lua configuration files found") end end local function collapse_configuration_data() -- potential optimization: pass start index (setup and configuration are shared) local order, variables, environment, origins = instance.order, instance.variables, instance.environment, instance.origins for i=1,#order do local c = order[i] for k,v in next, c do if variables[k] then -- okay else local ek = environment[k] if ek and ek ~= "" then variables[k], origins[k] = ek, "env" else local bv = checked_variable(v) variables[k], origins[k] = bv, "cnf" end end end end end -- database loading -- locators function resolvers.locatedatabase(specification) return resolvers.methodhandler('locators', specification) end function resolvers.locators.tex(specification) if specification and specification ~= '' and lfs.isdir(specification) then if trace_locating then report_resolvers("tex locator '%s' found",specification) end resolvers.append_hash('file',specification,filename,true) -- cache elseif trace_locating then report_resolvers("tex locator '%s' not found",specification) end end -- hashers function resolvers.hashdatabase(tag,name) return resolvers.methodhandler('hashers',tag,name) end local function load_file_databases() instance.loaderror, instance.files = false, { } if not instance.renewcache then local hashes = instance.hashes for k=1,#hashes do local hash = hashes[k] resolvers.hashdatabase(hash.tag,hash.name) if instance.loaderror then break end end end end function resolvers.hashers.tex(tag,name) -- used where? local content = caches.loadcontent(tag,'files') if content then instance.files[tag] = content else instance.files[tag] = { } instance.loaderror = true end end local function locate_file_databases() -- todo: cache:// and tree:// (runtime) local texmfpaths = resolvers.expanded_path_list('TEXMF') for i=1,#texmfpaths do local path = collapse_path(texmfpaths[i]) local stripped = gsub(path,"^!!","") local runtime = stripped == path path = resolvers.clean_path(path) if stripped ~= "" then if lfs.isdir(path) then local spec = resolvers.splitmethod(stripped) if spec.scheme == "cache" then stripped = spec.path elseif runtime and (spec.noscheme or spec.scheme == "file") then stripped = "tree:///" .. stripped end if trace_locating then if runtime then report_resolvers("locating list of '%s' (runtime)",path) else report_resolvers("locating list of '%s' (cached)",path) end end resolvers.locatedatabase(stripped) -- nothing done with result else if trace_locating then if runtime then report_resolvers("skipping list of '%s' (runtime)",path) else report_resolvers("skipping list of '%s' (cached)",path) end end end end end if trace_locating then report_resolvers() end end local function generate_file_databases() local hashes = instance.hashes for i=1,#hashes do resolvers.methodhandler('generators',hashes[i].tag) end if trace_locating then report_resolvers() end end local function save_file_databases() -- will become cachers for i=1,#instance.hashes do local hash = instance.hashes[i] local cachename = hash.tag if hash.cache then local content = instance.files[cachename] caches.collapsecontent(content) caches.savecontent(cachename,"files",content) elseif trace_locating then report_resolvers("not saving runtime tree '%s'",cachename) end end end local function load_databases() locate_file_databases() if instance.diskcache and not instance.renewcache then load_file_databases() if instance.loaderror then generate_file_databases() save_file_databases() end else generate_file_databases() if instance.renewcache then save_file_databases() end end end function resolvers.append_hash(type,tag,name,cache) if trace_locating then report_resolvers("hash '%s' appended",tag) end insert(instance.hashes, { type = type, tag = tag, name = name, cache = cache } ) end function resolvers.prepend_hash(type,tag,name,cache) if trace_locating then report_resolvers("hash '%s' prepended",tag) end insert(instance.hashes, 1, { type = type, tag = tag, name = name, cache = cache } ) end function resolvers.extend_texmf_var(specification) -- crap, we could better prepend the hash -- local t = resolvers.expanded_path_list('TEXMF') -- full expansion local t = resolvers.split_path(resolvers.getenv('TEXMF')) insert(t,1,specification) local newspec = concat(t,";") if instance.environment["TEXMF"] then instance.environment["TEXMF"] = newspec elseif instance.variables["TEXMF"] then instance.variables["TEXMF"] = newspec else -- weird end resolvers.expand_variables() reset_hashes() end function resolvers.generators.tex(specification,tag) instance.files[tag or specification] = resolvers.scan_files(specification) end function resolvers.splitexpansions() local ie = instance.expansions for k,v in next, ie do local t, h, p = { }, { }, split_configuration_path(v) for kk=1,#p do local vv = p[kk] if vv ~= "" and not h[vv] then t[#t+1] = vv h[vv] = true end end if #t > 1 then ie[k] = t else ie[k] = t[1] end end end -- end of split/join code -- we used to have 'files' and 'configurations' so therefore the following -- shared function function resolvers.data_state() return caches.contentstate() end function resolvers.expand_variables() local expansions, environment, variables = { }, instance.environment, instance.variables local getenv = resolvers.getenv instance.expansions = expansions local engine, progname = instance.engine, instance.progname if type(engine) ~= "string" then instance.engine, engine = "", "" end if type(progname) ~= "string" then instance.progname, progname = "", "" end if engine ~= "" then environment['engine'] = engine end if progname ~= "" then environment['progname'] = progname end for k,v in next, environment do local a, b = match(k,"^(%a+)%_(.*)%s*$") if a and b then expansions[a..'.'..b] = v else expansions[k] = v end end for k,v in next, environment do -- move environment to expansions (variables are already in there) if not expansions[k] then expansions[k] = v end end for k,v in next, variables do -- move variables to expansions if not expansions[k] then expansions[k] = v end end local busy = false local function resolve(a) busy = true return expansions[a] or getenv(a) end while true do busy = false for k,v in next, expansions do local s, n = gsub(v,"%$([%a%d%_%-]+)",resolve) local s, m = gsub(s,"%$%{([%a%d%_%-]+)%}",resolve) if n > 0 or m > 0 then s = gsub(s,";+",";") s = gsub(s,";[!{}/\\]+;",";") expansions[k]= s end end if not busy then break end end for k,v in next, expansions do expansions[k] = gsub(v,"\\", '/') end end function resolvers.variable(name) return entry(instance.variables,name) end function resolvers.expansion(name) return entry(instance.expansions,name) end function resolvers.is_variable(name) return is_entry(instance.variables,name) end function resolvers.is_expansion(name) return is_entry(instance.expansions,name) end function resolvers.unexpanded_path_list(str) local pth = resolvers.variable(str) local lst = resolvers.split_path(pth) return expanded_path_from_list(lst) end function resolvers.unexpanded_path(str) return file.join_path(resolvers.unexpanded_path_list(str)) end local done = { } function resolvers.reset_extra_path() local ep = instance.extra_paths if not ep then ep, done = { }, { } instance.extra_paths = ep elseif #ep > 0 then instance.lists, done = { }, { } end end function resolvers.register_extra_path(paths,subpaths) local ep = instance.extra_paths or { } local n = #ep if paths and paths ~= "" then if subpaths and subpaths ~= "" then for p in gmatch(paths,"[^,]+") do -- we gmatch each step again, not that fast, but used seldom for s in gmatch(subpaths,"[^,]+") do local ps = p .. "/" .. s if not done[ps] then ep[#ep+1] = resolvers.clean_path(ps) done[ps] = true end end end else for p in gmatch(paths,"[^,]+") do if not done[p] then ep[#ep+1] = resolvers.clean_path(p) done[p] = true end end end elseif subpaths and subpaths ~= "" then for i=1,n do -- we gmatch each step again, not that fast, but used seldom for s in gmatch(subpaths,"[^,]+") do local ps = ep[i] .. "/" .. s if not done[ps] then ep[#ep+1] = resolvers.clean_path(ps) done[ps] = true end end end end if #ep > 0 then instance.extra_paths = ep -- register paths end if #ep > n then instance.lists = { } -- erase the cache end end local function made_list(instance,list) local ep = instance.extra_paths if not ep or #ep == 0 then return list else local done, new = { }, { } -- honour . .. ../.. but only when at the start for k=1,#list do local v = list[k] if not done[v] then if find(v,"^[%.%/]$") then done[v] = true new[#new+1] = v else break end end end -- first the extra paths for k=1,#ep do local v = ep[k] if not done[v] then done[v] = true new[#new+1] = v end end -- next the formal paths for k=1,#list do local v = list[k] if not done[v] then done[v] = true new[#new+1] = v end end return new end end function resolvers.clean_path_list(str) local t = resolvers.expanded_path_list(str) if t then for i=1,#t do t[i] = collapse_path(resolvers.clean_path(t[i])) end end return t end function resolvers.expand_path(str) return file.join_path(resolvers.expanded_path_list(str)) end function resolvers.expanded_path_list(str) if not str then return ep or { } -- ep ? elseif instance.savelists then -- engine+progname hash str = gsub(str,"%$","") if not instance.lists[str] then -- cached local lst = made_list(instance,resolvers.split_path(resolvers.expansion(str))) instance.lists[str] = expanded_path_from_list(lst) end return instance.lists[str] else local lst = resolvers.split_path(resolvers.expansion(str)) return made_list(instance,expanded_path_from_list(lst)) end end function resolvers.expanded_path_list_from_var(str) -- brrr local tmp = resolvers.var_of_format_or_suffix(gsub(str,"%$","")) if tmp ~= "" then return resolvers.expanded_path_list(tmp) else return resolvers.expanded_path_list(str) end end function resolvers.expand_path_from_var(str) return file.join_path(resolvers.expanded_path_list_from_var(str)) end function resolvers.expand_braces(str) -- output variable and brace expansion of STRING local ori = resolvers.variable(str) local pth = expanded_path_from_list(resolvers.split_path(ori)) return file.join_path(pth) end resolvers.isreadable = { } function resolvers.isreadable.file(name) local readable = lfs.isfile(name) -- brrr if trace_detail then if readable then report_resolvers("file '%s' is readable",name) else report_resolvers("file '%s' is not readable", name) end end return readable end resolvers.isreadable.tex = resolvers.isreadable.file -- name -- name/name local function collect_files(names) local filelist = { } for k=1,#names do local fname = names[k] if trace_detail then report_resolvers("checking name '%s'",fname) end local bname = filebasename(fname) local dname = filedirname(fname) if dname == "" or find(dname,"^%.") then dname = false else dname = "/" .. dname .. "$" end local hashes = instance.hashes for h=1,#hashes do local hash = hashes[h] local blobpath = hash.tag local files = blobpath and instance.files[blobpath] if files then if trace_detail then report_resolvers("deep checking '%s' (%s)",blobpath,bname) end local blobfile = files[bname] if not blobfile then local rname = "remap:"..bname blobfile = files[rname] if blobfile then bname = files[rname] blobfile = files[bname] end end if blobfile then local blobroot = files.__path__ or blobpath if type(blobfile) == 'string' then if not dname or find(blobfile,dname) then local kind = hash.type local search = filejoin(blobpath,blobfile,bname) local result = resolvers.concatinators[hash.type](blobroot,blobfile,bname) if trace_detail then report_resolvers("match: kind '%s', search '%s', result '%s'",kind,search,result) end filelist[#filelist+1] = { kind, search, result } end else for kk=1,#blobfile do local vv = blobfile[kk] if not dname or find(vv,dname) then local kind = hash.type local search = filejoin(blobpath,vv,bname) local result = resolvers.concatinators[hash.type](blobroot,vv,bname) if trace_detail then report_resolvers("match: kind '%s', search '%s', result '%s'",kind,search,result) end filelist[#filelist+1] = { kind, search, result } end end end end elseif trace_locating then report_resolvers("no match in '%s' (%s)",blobpath,bname) end end end return #filelist > 0 and filelist or nil end function resolvers.register_in_trees(name) if not find(name,"^%.") then instance.foundintrees[name] = (instance.foundintrees[name] or 0) + 1 -- maybe only one end end -- split the next one up for readability (bu this module needs a cleanup anyway) local function can_be_dir(name) -- can become local local fakepaths = instance.fakepaths if not fakepaths[name] then if lfs.isdir(name) then fakepaths[name] = 1 -- directory else fakepaths[name] = 2 -- no directory end end return fakepaths[name] == 1 end local function collect_instance_files(filename,collected) -- todo : plugin (scanners, checkers etc) local result = collected or { } local stamp = nil filename = collapse_path(filename) -- speed up / beware: format problem if instance.remember then stamp = filename .. "--" .. instance.engine .. "--" .. instance.progname .. "--" .. instance.format if instance.found[stamp] then if trace_locating then report_resolvers("remembering file '%s'",filename) end resolvers.register_in_trees(filename) -- for tracing used files return instance.found[stamp] end end if not dangerous[instance.format or "?"] then if resolvers.isreadable.file(filename) then if trace_detail then report_resolvers("file '%s' found directly",filename) end instance.found[stamp] = { filename } return { filename } end end if find(filename,'%*') then if trace_locating then report_resolvers("checking wildcard '%s'", filename) end result = resolvers.find_wildcard_files(filename) elseif file.is_qualified_path(filename) then if resolvers.isreadable.file(filename) then if trace_locating then report_resolvers("qualified name '%s'", filename) end result = { filename } else local forcedname, ok, suffix = "", false, fileextname(filename) if suffix == "" then -- why if instance.format == "" then forcedname = filename .. ".tex" if resolvers.isreadable.file(forcedname) then if trace_locating then report_resolvers("no suffix, forcing standard filetype 'tex'") end result, ok = { forcedname }, true end else local format_suffixes = suffixes[instance.format] if format_suffixes then for i=1,#format_suffixes do local s = format_suffixes[i] forcedname = filename .. "." .. s if resolvers.isreadable.file(forcedname) then if trace_locating then report_resolvers("no suffix, forcing format filetype '%s'", s) end result, ok = { forcedname }, true break end end end end end if not ok and suffix ~= "" then -- try to find in tree (no suffix manipulation), here we search for the -- matching last part of the name local basename = filebasename(filename) local pattern = gsub(filename .. "$","([%.%-])","%%%1") local savedformat = instance.format local format = savedformat or "" if format == "" then instance.format = resolvers.format_of_suffix(suffix) end if not format then instance.format = "othertextfiles" -- kind of everything, maybe texinput is better end -- if basename ~= filename then local resolved = collect_instance_files(basename) if #result == 0 then local lowered = lower(basename) if filename ~= lowered then resolved = collect_instance_files(lowered) end end resolvers.format = savedformat -- for r=1,#resolved do local rr = resolved[r] if find(rr,pattern) then result[#result+1], ok = rr, true end end end -- a real wildcard: -- -- if not ok then -- local filelist = collect_files({basename}) -- for f=1,#filelist do -- local ff = filelist[f][3] or "" -- if find(ff,pattern) then -- result[#result+1], ok = ff, true -- end -- end -- end end if not ok and trace_locating then report_resolvers("qualified name '%s'", filename) end end else -- search spec local filetype, extra, done, wantedfiles, ext = '', nil, false, { }, fileextname(filename) -- tricky as filename can be bla.1.2.3 if ext == "" then if not instance.force_suffixes then wantedfiles[#wantedfiles+1] = filename end else wantedfiles[#wantedfiles+1] = filename end if instance.format == "" then if ext == "" or not suffixmap[ext] then local forcedname = filename .. '.tex' wantedfiles[#wantedfiles+1] = forcedname filetype = resolvers.format_of_suffix(forcedname) if trace_locating then report_resolvers("forcing filetype '%s'",filetype) end else filetype = resolvers.format_of_suffix(filename) if trace_locating then report_resolvers("using suffix based filetype '%s'",filetype) end end else if ext == "" or not suffixmap[ext] then local format_suffixes = suffixes[instance.format] if format_suffixes then for i=1,#format_suffixes do wantedfiles[#wantedfiles+1] = filename .. "." .. format_suffixes[i] end end end filetype = instance.format if trace_locating then report_resolvers("using given filetype '%s'",filetype) end end local typespec = resolvers.variable_of_format(filetype) local pathlist = resolvers.expanded_path_list(typespec) if not pathlist or #pathlist == 0 then -- no pathlist, access check only / todo == wildcard if trace_detail then report_resolvers("checking filename '%s', filetype '%s', wanted files '%s'",filename, filetype or '?',concat(wantedfiles," | ")) end for k=1,#wantedfiles do local fname = wantedfiles[k] if fname and resolvers.isreadable.file(fname) then filename, done = fname, true result[#result+1] = filejoin('.',fname) break end end -- this is actually 'other text files' or 'any' or 'whatever' local filelist = collect_files(wantedfiles) local fl = filelist and filelist[1] if fl then filename = fl[3] result[#result+1] = filename done = true end else -- list search local filelist = collect_files(wantedfiles) local dirlist = { } if filelist then for i=1,#filelist do dirlist[i] = filedirname(filelist[i][3]) .. "/" -- was [2] .. gamble end end if trace_detail then report_resolvers("checking filename '%s'",filename) end -- a bit messy ... esp the doscan setting here local doscan for k=1,#pathlist do local path = pathlist[k] if find(path,"^!!") then doscan = false else doscan = true end local pathname = gsub(path,"^!+", '') done = false -- using file list if filelist then local expression -- compare list entries with permitted pattern -- /xx /xx// if not find(pathname,"/$") then expression = pathname .. "/" else expression = pathname end expression = gsub(expression,"([%-%.])","%%%1") -- this also influences expression = gsub(expression,"//+$", '/.*') -- later usage of pathname expression = gsub(expression,"//", '/.-/') -- not ok for /// but harmless expression = "^" .. expression .. "$" if trace_detail then report_resolvers("using pattern '%s' for path '%s'",expression,pathname) end for k=1,#filelist do local fl = filelist[k] local f = fl[2] local d = dirlist[k] if find(d,expression) then --- todo, test for readable result[#result+1] = fl[3] done = true if instance.allresults then if trace_detail then report_resolvers("match to '%s' in hash for file '%s' and path '%s', continue scanning",expression,f,d) end else if trace_detail then report_resolvers("match to '%s' in hash for file '%s' and path '%s', quit scanning",expression,f,d) end break end elseif trace_detail then report_resolvers("no match to '%s' in hash for file '%s' and path '%s'",expression,f,d) end end end if not done and doscan then -- check if on disk / unchecked / does not work at all / also zips if resolvers.splitmethod(pathname).scheme == 'file' then -- ? local pname = gsub(pathname,"%.%*$",'') if not find(pname,"%*") then local ppname = gsub(pname,"/+$","") if can_be_dir(ppname) then for k=1,#wantedfiles do local w = wantedfiles[k] local fname = filejoin(ppname,w) if resolvers.isreadable.file(fname) then if trace_detail then report_resolvers("found '%s' by scanning",fname) end result[#result+1] = fname done = true if not instance.allresults then break end end end else -- no access needed for non existing path, speedup (esp in large tree with lots of fake) end end end end if not done and doscan then -- todo: slow path scanning ... although we now have tree:// supported in $TEXMF end if done and not instance.allresults then break end end end end for k=1,#result do local rk = collapse_path(result[k]) result[k] = rk resolvers.register_in_trees(rk) -- for tracing used files end if instance.remember then instance.found[stamp] = result end return result end if not resolvers.concatinators then resolvers.concatinators = { } end resolvers.concatinators.tex = filejoin resolvers.concatinators.file = resolvers.concatinators.tex function resolvers.find_files(filename,filetype,mustexist) if type(mustexist) == boolean then -- all set elseif type(filetype) == 'boolean' then filetype, mustexist = nil, false elseif type(filetype) ~= 'string' then filetype, mustexist = nil, false end instance.format = filetype or '' local result = collect_instance_files(filename) if #result == 0 then local lowered = lower(filename) if filename ~= lowered then return collect_instance_files(lowered) end end instance.format = '' return result end function resolvers.find_file(filename,filetype,mustexist) return (resolvers.find_files(filename,filetype,mustexist)[1] or "") end function resolvers.find_path(filename,filetype) local path = resolvers.find_files(filename,filetype)[1] or "" -- todo return current path return file.dirname(path) end function resolvers.find_given_files(filename) local bname, result = filebasename(filename), { } local hashes = instance.hashes for k=1,#hashes do local hash = hashes[k] local files = instance.files[hash.tag] or { } local blist = files[bname] if not blist then local rname = "remap:"..bname blist = files[rname] if blist then bname = files[rname] blist = files[bname] end end if blist then if type(blist) == 'string' then result[#result+1] = resolvers.concatinators[hash.type](hash.tag,blist,bname) or "" if not instance.allresults then break end else for kk=1,#blist do local vv = blist[kk] result[#result+1] = resolvers.concatinators[hash.type](hash.tag,vv,bname) or "" if not instance.allresults then break end end end end end return result end function resolvers.find_given_file(filename) return (resolvers.find_given_files(filename)[1] or "") end local function doit(path,blist,bname,tag,kind,result,allresults) local done = false if blist and kind then if type(blist) == 'string' then -- make function and share code if find(lower(blist),path) then result[#result+1] = resolvers.concatinators[kind](tag,blist,bname) or "" done = true end else for kk=1,#blist do local vv = blist[kk] if find(lower(vv),path) then result[#result+1] = resolvers.concatinators[kind](tag,vv,bname) or "" done = true if not allresults then break end end end end end return done end function resolvers.find_wildcard_files(filename) -- todo: remap: and lpeg local result = { } local bname, dname = filebasename(filename), filedirname(filename) local path = gsub(dname,"^*/","") path = gsub(path,"*",".*") path = gsub(path,"-","%%-") if dname == "" then path = ".*" end local name = bname name = gsub(name,"*",".*") name = gsub(name,"-","%%-") path = lower(path) name = lower(name) local files, allresults, done = instance.files, instance.allresults, false if find(name,"%*") then local hashes = instance.hashes for k=1,#hashes do local hash = hashes[k] local tag, kind = hash.tag, hash.type for kk, hh in next, files[hash.tag] do if not find(kk,"^remap:") then if find(lower(kk),name) then if doit(path,hh,kk,tag,kind,result,allresults) then done = true end if done and not allresults then break end end end end end else local hashes = instance.hashes for k=1,#hashes do local hash = hashes[k] local tag, kind = hash.tag, hash.type if doit(path,files[tag][bname],bname,tag,kind,result,allresults) then done = true end if done and not allresults then break end end end -- we can consider also searching the paths not in the database, but then -- we end up with a messy search (all // in all path specs) return result end function resolvers.find_wildcard_file(filename) return (resolvers.find_wildcard_files(filename)[1] or "") end -- main user functions function resolvers.automount() -- implemented later end function resolvers.load(option) statistics.starttiming(instance) identify_configuration_files() load_configuration_files() collapse_configuration_data() resolvers.expand_variables() if option ~= "nofiles" then load_databases() resolvers.automount() end statistics.stoptiming(instance) local files = instance.files return files and next(files) and true end function resolvers.for_files(command, files, filetype, mustexist) if files and #files > 0 then local function report(str) if trace_locating then report_resolvers(str) -- has already verbose else print(str) end end if trace_locating then report('') -- ? end for f=1,#files do local file = files[f] local result = command(file,filetype,mustexist) if type(result) == 'string' then report(result) else for i=1,#result do report(result[i]) -- could be unpack end end end end end -- strtab resolvers.var_value = resolvers.variable -- output the value of variable $STRING. resolvers.expand_var = resolvers.expansion -- output variable expansion of STRING. function resolvers.show_path(str) -- output search path for file type NAME return file.join_path(resolvers.expanded_path_list(resolvers.format_of_var(str))) end -- resolvers.find_file(filename) -- resolvers.find_file(filename, filetype, mustexist) -- resolvers.find_file(filename, mustexist) -- resolvers.find_file(filename, filetype) function resolvers.register_file(files, name, path) if files[name] then if type(files[name]) == 'string' then files[name] = { files[name], path } else files[name] = path end else files[name] = path end end function resolvers.do_with_path(name,func) local pathlist = resolvers.expanded_path_list(name) for i=1,#pathlist do func("^"..resolvers.clean_path(pathlist[i])) end end function resolvers.do_with_var(name,func) func(expanded_var(name)) end function resolvers.locate_format(name) local barename = gsub(name,"%.%a+$","") local fmtname = caches.getfirstreadablefile(barename..".fmt","formats") or "" if fmtname == "" then fmtname = resolvers.find_files(barename..".fmt")[1] or "" fmtname = resolvers.clean_path(fmtname) end if fmtname ~= "" then local barename = file.removesuffix(fmtname) local luaname, lucname, luiname = barename .. ".lua", barename .. ".luc", barename .. ".lui" if lfs.isfile(luiname) then return barename, luiname elseif lfs.isfile(lucname) then return barename, lucname elseif lfs.isfile(luaname) then return barename, luaname end end return nil, nil end function resolvers.boolean_variable(str,default) local b = resolvers.expansion(str) if b == "" then return default else b = toboolean(b) return (b == nil and default) or b end end function resolvers.with_files(pattern,handle,before,after) -- can be a nice iterator instead local instance = resolvers.instance local hashes = instance.hashes for i=1,#hashes do local hash = hashes[i] local blobtype = hash.type local blobpath = hash.tag if blobpath then if before then before(blobtype,blobpath,pattern) end local files = instance.files[blobpath] local total, checked, done = 0, 0, 0 if files then for k,v in next, files do total = total + 1 if find(k,"^remap:") then k = files[k] v = k -- files[k] -- chained end if find(k,pattern) then if type(v) == "string" then checked = checked + 1 if handle(blobtype,blobpath,v,k) then done = done + 1 end else checked = checked + #v for i=1,#v do if handle(blobtype,blobpath,v[i],k) then done = done + 1 end end end end end end if after then after(blobtype,blobpath,pattern,total,checked,done) end end end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-pre'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local upper, lower, gsub = string.upper, string.lower, string.gsub local prefixes = { } local getenv = resolvers.getenv prefixes.environment = function(str) -- getenv is case insensitive anyway return resolvers.clean_path(getenv(str) or getenv(upper(str)) or getenv(lower(str)) or "") end prefixes.relative = function(str,n) if io.exists(str) then -- nothing elseif io.exists("./" .. str) then str = "./" .. str else local p = "../" for i=1,n or 2 do if io.exists(p .. str) then str = p .. str break else p = p .. "../" end end end return resolvers.clean_path(str) end prefixes.auto = function(str) local fullname = prefixes.relative(str) if not lfs.isfile(fullname) then fullname = prefixes.locate(str) end return fullname end prefixes.locate = function(str) local fullname = resolvers.find_given_file(str) or "" return resolvers.clean_path((fullname ~= "" and fullname) or str) end prefixes.filename = function(str) local fullname = resolvers.find_given_file(str) or "" return resolvers.clean_path(file.basename((fullname ~= "" and fullname) or str)) end prefixes.pathname = function(str) local fullname = resolvers.find_given_file(str) or "" return resolvers.clean_path(file.dirname((fullname ~= "" and fullname) or str)) end prefixes.env = prefixes.environment prefixes.rel = prefixes.relative prefixes.loc = prefixes.locate prefixes.kpse = prefixes.locate prefixes.full = prefixes.locate prefixes.file = prefixes.filename prefixes.path = prefixes.pathname function resolvers.allprefixes(separator) local all = table.sortedkeys(prefixes) if separator then for i=1,#all do all[i] = all[i] .. ":" end end return all end local function _resolve_(method,target) if prefixes[method] then return prefixes[method](target) else return method .. ":" .. target end end local function resolve(str) if type(str) == "table" then for k=1,#str do local v = str[k] str[k] = resolve(v) or v end elseif str and str ~= "" then str = gsub(str,"([a-z]+):([^ \"\']*)",_resolve_) end return str end resolvers.resolve = resolve if os.uname then for k, v in next, os.uname() do if not prefixes[k] then prefixes[k] = function() return v end end end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-inp'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } resolvers.finders = resolvers.finders or { } resolvers.openers = resolvers.openers or { } resolvers.loaders = resolvers.loaders or { } resolvers.finders.notfound = { nil } resolvers.openers.notfound = { nil } resolvers.loaders.notfound = { false, nil, 0 } end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-out'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } outputs = outputs or { } end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-con'] = { version = 1.100, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local format, lower, gsub = string.format, string.lower, string.gsub local trace_cache = false trackers.register("resolvers.cache", function(v) trace_cache = v end) local trace_containers = false trackers.register("resolvers.containers", function(v) trace_containers = v end) local trace_storage = false trackers.register("resolvers.storage", function(v) trace_storage = v end) --[[ldx--

Once we found ourselves defining similar cache constructs several times, containers were introduced. Containers are used to collect tables in memory and reuse them when possible based on (unique) hashes (to be provided by the calling function).

Caching to disk is disabled by default. Version numbers are stored in the saved table which makes it possible to change the table structures without bothering about the disk cache.

Examples of usage can be found in the font related code.

--ldx]]-- containers = containers or { } containers.usecache = true local report_cache = logs.new("cache") local function report(container,tag,name) if trace_cache or trace_containers then report_cache("container: %s, tag: %s, name: %s",container.subcategory,tag,name or 'invalid') end end local allocated = { } local mt = { __index = function(t,k) if k == "writable" then local writable = caches.getwritablepath(t.category,t.subcategory) or { "." } t.writable = writable return writable elseif k == "readables" then local readables = caches.getreadablepaths(t.category,t.subcategory) or { "." } t.readables = readables return readables end end } function containers.define(category, subcategory, version, enabled) if category and subcategory then local c = allocated[category] if not c then c = { } allocated[category] = c end local s = c[subcategory] if not s then s = { category = category, subcategory = subcategory, storage = { }, enabled = enabled, version = version or math.pi, -- after all, this is TeX trace = false, -- writable = caches.getwritablepath and caches.getwritablepath (category,subcategory) or { "." }, -- readables = caches.getreadablepaths and caches.getreadablepaths(category,subcategory) or { "." }, } setmetatable(s,mt) c[subcategory] = s end return s end end function containers.is_usable(container, name) return container.enabled and caches and caches.iswritable(container.writable, name) end function containers.is_valid(container, name) if name and name ~= "" then local storage = container.storage[name] return storage and storage.cache_version == container.version else return false end end function containers.read(container,name) local storage = container.storage local stored = storage[name] if not stored and container.enabled and caches and containers.usecache then stored = caches.loaddata(container.readables,name) if stored and stored.cache_version == container.version then report(container,"loaded",name) else stored = nil end storage[name] = stored elseif stored then report(container,"reusing",name) end return stored end function containers.write(container, name, data) if data then data.cache_version = container.version if container.enabled and caches then local unique, shared = data.unique, data.shared data.unique, data.shared = nil, nil caches.savedata(container.writable, name, data) report(container,"saved",name) data.unique, data.shared = unique, shared end report(container,"stored",name) container.storage[name] = data end return data end function containers.content(container,name) return container.storage[name] end function containers.cleanname(name) return (gsub(lower(name),"[^%w%d]+","-")) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-use'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local format, lower, gsub, find = string.format, string.lower, string.gsub, string.find local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") -- we will make a better format, maybe something xml or just text or lua resolvers.automounted = resolvers.automounted or { } function resolvers.automount(usecache) local mountpaths = resolvers.clean_path_list(resolvers.expansion('TEXMFMOUNT')) if (not mountpaths or #mountpaths == 0) and usecache then mountpaths = caches.getreadablepaths("mount") end if mountpaths and #mountpaths > 0 then statistics.starttiming(resolvers.instance) for k=1,#mountpaths do local root = mountpaths[k] local f = io.open(root.."/url.tmi") if f then for line in f:lines() do if line then if find(line,"^[%%#%-]") then -- or %W -- skip elseif find(line,"^zip://") then if trace_locating then report_resolvers("mounting %s",line) end table.insert(resolvers.automounted,line) resolvers.usezipfile(line) end end end f:close() end end statistics.stoptiming(resolvers.instance) end end -- status info statistics.register("used config file", function() return caches.configfiles() end) statistics.register("used cache path", function() return caches.usedpaths() end) -- experiment (code will move) function statistics.save_fmt_status(texname,formatbanner,sourcefile) -- texname == formatname local enginebanner = status.list().banner if formatbanner and enginebanner and sourcefile then local luvname = file.replacesuffix(texname,"luv") local luvdata = { enginebanner = enginebanner, formatbanner = formatbanner, sourcehash = md5.hex(io.loaddata(resolvers.find_file(sourcefile)) or "unknown"), sourcefile = sourcefile, } io.savedata(luvname,table.serialize(luvdata,true)) end end function statistics.check_fmt_status(texname) local enginebanner = status.list().banner if enginebanner and texname then local luvname = file.replacesuffix(texname,"luv") if lfs.isfile(luvname) then local luv = dofile(luvname) if luv and luv.sourcefile then local sourcehash = md5.hex(io.loaddata(resolvers.find_file(luv.sourcefile)) or "unknown") local luvbanner = luv.enginebanner or "?" if luvbanner ~= enginebanner then return format("engine mismatch (luv: %s <> bin: %s)",luvbanner,enginebanner) end local luvhash = luv.sourcehash or "?" if luvhash ~= sourcehash then return format("source mismatch (luv: %s <> bin: %s)",luvhash,sourcehash) end else return "invalid status file" end else return "missing status file" end end return true end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-zip'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local format, find, match = string.format, string.find, string.match local unpack = unpack or table.unpack local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") -- zip:///oeps.zip?name=bla/bla.tex -- zip:///oeps.zip?tree=tex/texmf-local -- zip:///texmf.zip?tree=/tex/texmf -- zip:///texmf.zip?tree=/tex/texmf-local -- zip:///texmf-mine.zip?tree=/tex/texmf-projects zip = zip or { } zip.archives = zip.archives or { } zip.registeredfiles = zip.registeredfiles or { } local finders, openers, loaders = resolvers.finders, resolvers.openers, resolvers.loaders local locators, hashers, concatinators = resolvers.locators, resolvers.hashers, resolvers.concatinators local archives = zip.archives local function validzip(str) -- todo: use url splitter if not find(str,"^zip://") then return "zip:///" .. str else return str end end function zip.openarchive(name) if not name or name == "" then return nil else local arch = archives[name] if not arch then local full = resolvers.find_file(name) or "" arch = (full ~= "" and zip.open(full)) or false archives[name] = arch end return arch end end function zip.closearchive(name) if not name or (name == "" and archives[name]) then zip.close(archives[name]) archives[name] = nil end end function locators.zip(specification) -- where is this used? startup zips (untested) specification = resolvers.splitmethod(specification) local zipfile = specification.path local zfile = zip.openarchive(name) -- tricky, could be in to be initialized tree if trace_locating then if zfile then report_resolvers("zip locator, archive '%s' found",specification.original) else report_resolvers("zip locator, archive '%s' not found",specification.original) end end end function hashers.zip(tag,name) if trace_locating then report_resolvers("loading zip file '%s' as '%s'",name,tag) end resolvers.usezipfile(format("%s?tree=%s",tag,name)) end function concatinators.zip(tag,path,name) if not path or path == "" then return format('%s?name=%s',tag,name) else return format('%s?name=%s/%s',tag,path,name) end end function resolvers.isreadable.zip(name) return true end function finders.zip(specification,filetype) specification = resolvers.splitmethod(specification) if specification.path then local q = url.query(specification.query) if q.name then local zfile = zip.openarchive(specification.path) if zfile then if trace_locating then report_resolvers("zip finder, archive '%s' found",specification.path) end local dfile = zfile:open(q.name) if dfile then dfile = zfile:close() if trace_locating then report_resolvers("zip finder, file '%s' found",q.name) end return specification.original elseif trace_locating then report_resolvers("zip finder, file '%s' not found",q.name) end elseif trace_locating then report_resolvers("zip finder, unknown archive '%s'",specification.path) end end end if trace_locating then report_resolvers("zip finder, '%s' not found",filename) end return unpack(finders.notfound) end function openers.zip(specification) local zipspecification = resolvers.splitmethod(specification) if zipspecification.path then local q = url.query(zipspecification.query) if q.name then local zfile = zip.openarchive(zipspecification.path) if zfile then if trace_locating then report_resolvers("zip opener, archive '%s' opened",zipspecification.path) end local dfile = zfile:open(q.name) if dfile then logs.show_open(specification) if trace_locating then report_resolvers("zip opener, file '%s' found",q.name) end return openers.text_opener(specification,dfile,'zip') elseif trace_locating then report_resolvers("zip opener, file '%s' not found",q.name) end elseif trace_locating then report_resolvers("zip opener, unknown archive '%s'",zipspecification.path) end end end if trace_locating then report_resolvers("zip opener, '%s' not found",filename) end return unpack(openers.notfound) end function loaders.zip(specification) specification = resolvers.splitmethod(specification) if specification.path then local q = url.query(specification.query) if q.name then local zfile = zip.openarchive(specification.path) if zfile then if trace_locating then report_resolvers("zip loader, archive '%s' opened",specification.path) end local dfile = zfile:open(q.name) if dfile then logs.show_load(filename) if trace_locating then report_resolvers("zip loader, file '%s' loaded",filename) end local s = dfile:read("*all") dfile:close() return true, s, #s elseif trace_locating then report_resolvers("zip loader, file '%s' not found",q.name) end elseif trace_locating then report_resolvers("zip loader, unknown archive '%s'",specification.path) end end end if trace_locating then report_resolvers("zip loader, '%s' not found",filename) end return unpack(openers.notfound) end -- zip:///somefile.zip -- zip:///somefile.zip?tree=texmf-local -> mount function resolvers.usezipfile(zipname) zipname = validzip(zipname) local specification = resolvers.splitmethod(zipname) local zipfile = specification.path if zipfile and not zip.registeredfiles[zipname] then local tree = url.query(specification.query).tree or "" local z = zip.openarchive(zipfile) if z then local instance = resolvers.instance if trace_locating then report_resolvers("zip registering, registering archive '%s'",zipname) end statistics.starttiming(instance) resolvers.prepend_hash('zip',zipname,zipfile) resolvers.extend_texmf_var(zipname) -- resets hashes too zip.registeredfiles[zipname] = z instance.files[zipname] = resolvers.register_zip_file(z,tree or "") statistics.stoptiming(instance) elseif trace_locating then report_resolvers("zip registering, unknown archive '%s'",zipname) end elseif trace_locating then report_resolvers("zip registering, '%s' not found",zipname) end end function resolvers.register_zip_file(z,tree) local files, filter = { }, "" if tree == "" then filter = "^(.+)/(.-)$" else filter = format("^%s/(.+)/(.-)$",tree) end if trace_locating then report_resolvers("zip registering, using filter '%s'",filter) end local register, n = resolvers.register_file, 0 for i in z:files() do local path, name = match(i.filename,filter) if path then if name and name ~= '' then register(files, name, path) n = n + 1 else -- directory end else register(files, i.filename, '') n = n + 1 end end report_resolvers("zip registering, %s files registered",n) return files end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-tre'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- \input tree://oeps1/**/oeps.tex local find, gsub, format = string.find, string.gsub, string.format local unpack = unpack or table.unpack local report_resolvers = logs.new("resolvers") local done, found, notfound = { }, { }, resolvers.finders.notfound function resolvers.finders.tree(specification,filetype) local fnd = found[specification] if not fnd then local spec = resolvers.splitmethod(specification).path or "" if spec ~= "" then local path, name = file.dirname(spec), file.basename(spec) if path == "" then path = "." end local hash = done[path] if not hash then local pattern = path .. "/*" -- we will use the proper splitter hash = dir.glob(pattern) done[path] = hash end local pattern = "/" .. gsub(name,"([%.%-%+])", "%%%1") .. "$" for k=1,#hash do local v = hash[k] if find(v,pattern) then found[specification] = v return v end end end fnd = unpack(notfound) -- unpack ? why not just notfound[1] found[specification] = fnd end return fnd end function resolvers.locators.tree(specification) local spec = resolvers.splitmethod(specification) local path = spec.path if path ~= '' and lfs.isdir(path) then if trace_locating then report_resolvers("tree locator '%s' found (%s)",path,specification) end resolvers.append_hash('tree',specification,path,false) -- don't cache elseif trace_locating then report_resolvers("tree locator '%s' not found",path) end end function resolvers.hashers.tree(tag,name) if trace_locating then report_resolvers("analysing tree '%s' as '%s'",name,tag) end -- todo: maybe share with done above local spec = resolvers.splitmethod(tag) local path = spec.path resolvers.generators.tex(path,tag) -- we share this with the normal tree analyzer end function resolvers.generators.tree(tag) local spec = resolvers.splitmethod(tag) local path = spec.path resolvers.generators.tex(path,tag) -- we share this with the normal tree analyzer end function resolvers.concatinators.tree(tag,path,name) return file.join(tag,path,name) end resolvers.isreadable.tree = file.isreadable resolvers.openers.tree = resolvers.openers.generic resolvers.loaders.tree = resolvers.loaders.generic end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-crl'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- this one is replaced by data-sch.lua -- curl = curl or { } local gsub = string.gsub local finders, openers, loaders = resolvers.finders, resolvers.openers, resolvers.loaders local cached = { } function curl.fetch(protocol, name) -- todo: use socket library local cleanname = gsub(name,"[^%a%d%.]+","-") local cachename = caches.setfirstwritablefile(cleanname,"curl") if not cached[name] then if not io.exists(cachename) then cached[name] = cachename local command = "curl --silent --create-dirs --output " .. cachename .. " " .. name -- no protocol .. "://" os.spawn(command) end if io.exists(cachename) then cached[name] = cachename else cached[name] = "" end end return cached[name] end function finders.curl(protocol,filename) local foundname = curl.fetch(protocol, filename) return finders.generic(protocol,foundname,filetype) end function openers.curl(protocol,filename) return openers.generic(protocol,filename) end function loaders.curl(protocol,filename) return loaders.generic(protocol,filename) end -- todo: metamethod function curl.install(protocol) finders[protocol] = function (filename,filetype) return finders.curl(protocol,filename) end openers[protocol] = function (filename) return openers.curl(protocol,filename) end loaders[protocol] = function (filename) return loaders.curl(protocol,filename) end end curl.install('http') curl.install('https') curl.install('ftp') end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-lua'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- some loading stuff ... we might move this one to slot 2 depending -- on the developments (the loaders must not trigger kpse); we could -- of course use a more extensive lib path spec local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") local gsub, insert = string.gsub, table.insert local unpack = unpack or table.unpack local libformats = { 'luatexlibs', 'tex', 'texmfscripts', 'othertextfiles' } -- 'luainputs' local clibformats = { 'lib' } local _path_, libpaths, _cpath_, clibpaths function package.libpaths() if not _path_ or package.path ~= _path_ then _path_ = package.path libpaths = file.split_path(_path_,";") end return libpaths end function package.clibpaths() if not _cpath_ or package.cpath ~= _cpath_ then _cpath_ = package.cpath clibpaths = file.split_path(_cpath_,";") end return clibpaths end local function thepath(...) local t = { ... } t[#t+1] = "?.lua" local path = file.join(unpack(t)) if trace_locating then report_resolvers("! appending '%s' to 'package.path'",path) end return path end local p_libpaths, a_libpaths = { }, { } function package.append_libpath(...) insert(a_libpath,thepath(...)) end function package.prepend_libpath(...) insert(p_libpaths,1,thepath(...)) end -- beware, we need to return a loadfile result ! local function loaded(libpaths,name,simple) for i=1,#libpaths do -- package.path, might become option local libpath = libpaths[i] local resolved = gsub(libpath,"%?",simple) if trace_locating then -- more detail report_resolvers("! checking for '%s' on 'package.path': '%s' => '%s'",simple,libpath,resolved) end if resolvers.isreadable.file(resolved) then if trace_locating then report_resolvers("! lib '%s' located via 'package.path': '%s'",name,resolved) end return loadfile(resolved) end end end package.loaders[2] = function(name) -- was [#package.loaders+1] if trace_locating then -- mode detail report_resolvers("! locating '%s'",name) end for i=1,#libformats do local format = libformats[i] local resolved = resolvers.find_file(name,format) or "" if trace_locating then -- mode detail report_resolvers("! checking for '%s' using 'libformat path': '%s'",name,format) end if resolved ~= "" then if trace_locating then report_resolvers("! lib '%s' located via environment: '%s'",name,resolved) end return loadfile(resolved) end end -- libpaths local libpaths, clibpaths = package.libpaths(), package.clibpaths() local simple = gsub(name,"%.lua$","") local simple = gsub(simple,"%.","/") local resolved = loaded(p_libpaths,name,simple) or loaded(libpaths,name,simple) or loaded(a_libpaths,name,simple) if resolved then return resolved end -- local libname = file.addsuffix(simple,os.libsuffix) for i=1,#clibformats do -- better have a dedicated loop local format = clibformats[i] local paths = resolvers.expanded_path_list_from_var(format) for p=1,#paths do local path = paths[p] local resolved = file.join(path,libname) if trace_locating then -- mode detail report_resolvers("! checking for '%s' using 'clibformat path': '%s'",libname,path) end if resolvers.isreadable.file(resolved) then if trace_locating then report_resolvers("! lib '%s' located via 'clibformat': '%s'",libname,resolved) end return package.loadlib(resolved,name) end end end for i=1,#clibpaths do -- package.path, might become option local libpath = clibpaths[i] local resolved = gsub(libpath,"?",simple) if trace_locating then -- more detail report_resolvers("! checking for '%s' on 'package.cpath': '%s'",simple,libpath) end if resolvers.isreadable.file(resolved) then if trace_locating then report_resolvers("! lib '%s' located via 'package.cpath': '%s'",name,resolved) end return package.loadlib(resolved,name) end end -- just in case the distribution is messed up if trace_loading then -- more detail report_resolvers("! checking for '%s' using 'luatexlibs': '%s'",name) end local resolved = resolvers.find_file(file.basename(name),'luatexlibs') or "" if resolved ~= "" then if trace_locating then report_resolvers("! lib '%s' located by basename via environment: '%s'",name,resolved) end return loadfile(resolved) end if trace_locating then report_resolvers('? unable to locate lib: %s',name) end -- return "unable to locate " .. name end resolvers.loadlualib = require end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-aux'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } local find = string.find local type, next = type, next local trace_locating = false trackers.register("resolvers.locating", function(v) trace_locating = v end) local report_resolvers = logs.new("resolvers") function resolvers.update_script(oldname,newname) -- oldname -> own.name, not per se a suffix local scriptpath = "scripts/context/lua" newname = file.addsuffix(newname,"lua") local oldscript = resolvers.clean_path(oldname) if trace_locating then report_resolvers("to be replaced old script %s", oldscript) end local newscripts = resolvers.find_files(newname) or { } if #newscripts == 0 then if trace_locating then report_resolvers("unable to locate new script") end else for i=1,#newscripts do local newscript = resolvers.clean_path(newscripts[i]) if trace_locating then report_resolvers("checking new script %s", newscript) end if oldscript == newscript then if trace_locating then report_resolvers("old and new script are the same") end elseif not find(newscript,scriptpath) then if trace_locating then report_resolvers("new script should come from %s",scriptpath) end elseif not (find(oldscript,file.removesuffix(newname).."$") or find(oldscript,newname.."$")) then if trace_locating then report_resolvers("invalid new script name") end else local newdata = io.loaddata(newscript) if newdata then if trace_locating then report_resolvers("old script content replaced by new content") end io.savedata(oldscript,newdata) break elseif trace_locating then report_resolvers("unable to load new script") end end end end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-tmf'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- = << -- ? ?? -- < += -- > =+ function resolvers.load_tree(tree) if type(tree) == "string" and tree ~= "" then local getenv, setenv = resolvers.getenv, resolvers.setenv -- later might listen to the raw osenv var as well local texos = "texmf-" .. os.platform local oldroot = environment.texroot local newroot = file.collapse_path(tree) local newtree = file.join(newroot,texos) local newpath = file.join(newtree,"bin") if not lfs.isdir(newtree) then logs.simple("no '%s' under tree %s",texos,tree) os.exit() end if not lfs.isdir(newpath) then logs.simple("no '%s/bin' under tree %s",texos,tree) os.exit() end local texmfos = newtree environment.texroot = newroot environment.texos = texos environment.texmfos = texmfos setenv('SELFAUTOPARENT', newroot) setenv('SELFAUTODIR', newtree) setenv('SELFAUTOLOC', newpath) setenv('TEXROOT', newroot) setenv('TEXOS', texos) setenv('TEXMFOS', texmfos) setenv('TEXROOT', newroot) setenv('TEXMFCNF', resolvers.luacnfspec) setenv("PATH", newpath .. io.pathseparator .. getenv("PATH")) logs.simple("changing from root '%s' to '%s'",oldroot,newroot) logs.simple("prepending '%s' to binary path",newpath) logs.simple() end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['data-lst'] = { version = 1.001, comment = "companion to luat-lib.mkiv", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- used in mtxrun local find, concat, upper, format = string.find, table.concat, string.upper, string.format resolvers.listers = resolvers.listers or { } local function tabstr(str) if type(str) == 'table' then return concat(str," | ") else return str end end local function list(list,report,pattern) pattern = pattern and pattern ~= "" and upper(pattern) or "" local instance = resolvers.instance local report = report or texio.write_nl local sorted = table.sortedkeys(list) for i=1,#sorted do local key = sorted[i] if pattern == "" or find(upper(key),pattern) then report(format('%s %s=%s',instance.origins[key] or "---",key,tabstr(list[key]))) end end end function resolvers.listers.variables (report,pattern) list(resolvers.instance.variables, report,pattern) end function resolvers.listers.expansions(report,pattern) list(resolvers.instance.expansions,report,pattern) end function resolvers.listers.configurations(report) local configurations = resolvers.instance.specification local report = report or texio.write_nl for i=1,#configurations do report(configurations[i]) end end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['luat-sta'] = { version = 1.001, author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- this code is used in the updater local gmatch, match = string.gmatch, string.match local type = type states = states or { } states.data = states.data or { } states.hash = states.hash or { } states.tag = states.tag or "" states.filename = states.filename or "" function states.save(filename,tag) tag = tag or states.tag filename = file.addsuffix(filename or states.filename,'lus') io.savedata(filename, "-- generator : luat-sta.lua\n" .. "-- state tag : " .. tag .. "\n\n" .. table.serialize(states.data[tag or states.tag] or {},true) ) end function states.load(filename,tag) states.filename = filename states.tag = tag or "whatever" states.filename = file.addsuffix(states.filename,'lus') states.data[states.tag], states.hash[states.tag] = (io.exists(filename) and dofile(filename)) or { }, { } end function states.set_by_tag(tag,key,value,default,persistent) local d, h = states.data[tag], states.hash[tag] if d then if type(d) == "table" then local dkey, hkey = key, key local pre, post = match(key,"(.+)%.([^%.]+)$") if pre and post then for k in gmatch(pre,"[^%.]+") do local dk = d[k] if not dk then dk = { } d[k] = dk elseif type(dk) == "string" then -- invalid table, unable to upgrade structure -- hope for the best or delete the state file break end d = dk end dkey, hkey = post, key end if type(value) == nil then value = value or default elseif persistent then value = value or d[dkey] or default else value = value or default end d[dkey], h[hkey] = value, value elseif type(d) == "string" then -- weird states.data[tag], states.hash[tag] = value, value end end end function states.get_by_tag(tag,key,default) local h = states.hash[tag] if h and h[key] then return h[key] else local d = states.data[tag] if d then for k in gmatch(key,"[^%.]+") do local dk = d[k] if dk then d = dk else return default end end return d or default end end end function states.set(key,value,default,persistent) states.set_by_tag(states.tag,key,value,default,persistent) end function states.get(key,default) return states.get_by_tag(states.tag,key,default) end end -- of closure do -- create closure to overcome 200 locals limit if not modules then modules = { } end modules ['luat-fmt'] = { version = 1.001, comment = "companion to mtxrun", author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", copyright = "PRAGMA ADE / ConTeXt Development Team", license = "see context related readme files" } -- helper for mtxrun local quote = string.quote local function primaryflags() local trackers = environment.argument("trackers") local directives = environment.argument("directives") local flags = "" if trackers and trackers ~= "" then flags = flags .. "--trackers=" .. quote(trackers) end if directives and directives ~= "" then flags = flags .. "--directives=" .. quote(directives) end return flags end function environment.make_format(name) -- change to format path (early as we need expanded paths) local olddir = lfs.currentdir() local path = caches.getwritablepath("formats") or "" -- maybe platform if path ~= "" then lfs.chdir(path) end logs.simple("format path: %s",lfs.currentdir()) -- check source file local texsourcename = file.addsuffix(name,"tex") local fulltexsourcename = resolvers.find_file(texsourcename,"tex") or "" if fulltexsourcename == "" then logs.simple("no tex source file with name: %s",texsourcename) lfs.chdir(olddir) return else logs.simple("using tex source file: %s",fulltexsourcename) end local texsourcepath = dir.expand_name(file.dirname(fulltexsourcename)) -- really needed -- check specification local specificationname = file.replacesuffix(fulltexsourcename,"lus") local fullspecificationname = resolvers.find_file(specificationname,"tex") or "" if fullspecificationname == "" then specificationname = file.join(texsourcepath,"context.lus") fullspecificationname = resolvers.find_file(specificationname,"tex") or "" end if fullspecificationname == "" then logs.simple("unknown stub specification: %s",specificationname) lfs.chdir(olddir) return end local specificationpath = file.dirname(fullspecificationname) -- load specification local usedluastub = nil local usedlualibs = dofile(fullspecificationname) if type(usedlualibs) == "string" then usedluastub = file.join(file.dirname(fullspecificationname),usedlualibs) elseif type(usedlualibs) == "table" then logs.simple("using stub specification: %s",fullspecificationname) local texbasename = file.basename(name) local luastubname = file.addsuffix(texbasename,"lua") local lucstubname = file.addsuffix(texbasename,"luc") -- pack libraries in stub logs.simple("creating initialization file: %s",luastubname) utils.merger.selfcreate(usedlualibs,specificationpath,luastubname) -- compile stub file (does not save that much as we don't use this stub at startup any more) local strip = resolvers.boolean_variable("LUACSTRIP", true) if utils.lua.compile(luastubname,lucstubname,false,strip) and lfs.isfile(lucstubname) then logs.simple("using compiled initialization file: %s",lucstubname) usedluastub = lucstubname else logs.simple("using uncompiled initialization file: %s",luastubname) usedluastub = luastubname end else logs.simple("invalid stub specification: %s",fullspecificationname) lfs.chdir(olddir) return end -- generate format local command = string.format("luatex --ini %s --lua=%s %s %sdump",primaryflags(),quote(usedluastub),quote(fulltexsourcename),os.platform == "unix" and "\\\\" or "\\") logs.simple("running command: %s\n",command) os.spawn(command) -- remove related mem files local pattern = file.removesuffix(file.basename(usedluastub)).."-*.mem" -- logs.simple("removing related mplib format with pattern '%s'", pattern) local mp = dir.glob(pattern) if mp then for i=1,#mp do local name = mp[i] logs.simple("removing related mplib format %s", file.basename(name)) os.remove(name) end end lfs.chdir(olddir) end function environment.run_format(name,data,more) -- hm, rather old code here; we can now use the file.whatever functions if name and name ~= "" then local barename = file.removesuffix(name) local fmtname = caches.getfirstreadablefile(file.addsuffix(barename,"fmt"),"formats") if fmtname == "" then fmtname = resolvers.find_file(file.addsuffix(barename,"fmt")) or "" end fmtname = resolvers.clean_path(fmtname) if fmtname == "" then logs.simple("no format with name: %s",name) else local barename = file.removesuffix(name) -- expanded name local luaname = file.addsuffix(barename,"luc") if not lfs.isfile(luaname) then luaname = file.addsuffix(barename,"lua") end if not lfs.isfile(luaname) then logs.simple("using format name: %s",fmtname) logs.simple("no luc/lua with name: %s",barename) else local q = string.quote local command = string.format("luatex %s --fmt=%s --lua=%s %s %s",primaryflags(),quote(barename),quote(luaname),quote(data),more ~= "" and quote(more) or "") logs.simple("running command: %s",command) os.spawn(command) end end end end end -- of closure -- end library merge own = { } -- not local, might change own.libs = { -- order can be made better 'l-string.lua', 'l-lpeg.lua', 'l-table.lua', 'l-io.lua', 'l-number.lua', 'l-set.lua', 'l-os.lua', 'l-file.lua', 'l-md5.lua', 'l-url.lua', 'l-dir.lua', 'l-boolean.lua', 'l-unicode.lua', 'l-math.lua', 'l-utils.lua', 'l-aux.lua', 'trac-inf.lua', 'trac-set.lua', 'trac-tra.lua', 'trac-log.lua', 'trac-pro.lua', 'luat-env.lua', -- can come before inf (as in mkiv) 'lxml-tab.lua', 'lxml-lpt.lua', -- 'lxml-ent.lua', 'lxml-mis.lua', 'lxml-aux.lua', 'lxml-xml.lua', 'data-ini.lua', 'data-exp.lua', 'data-env.lua', 'data-tmp.lua', 'data-met.lua', 'data-res.lua', 'data-pre.lua', 'data-inp.lua', 'data-out.lua', 'data-con.lua', 'data-use.lua', -- 'data-tex.lua', -- 'data-bin.lua', 'data-zip.lua', 'data-tre.lua', 'data-crl.lua', 'data-lua.lua', 'data-aux.lua', -- updater 'data-tmf.lua', 'data-lst.lua', 'luat-sta.lua', 'luat-fmt.lua', } -- We need this hack till luatex is fixed. -- -- for k,v in pairs(arg) do print(k,v) end if arg and (arg[0] == 'luatex' or arg[0] == 'luatex.exe') and arg[1] == "--luaonly" then arg[-1]=arg[0] arg[0]=arg[2] for k=3,#arg do arg[k-2]=arg[k] end arg[#arg]=nil arg[#arg]=nil end -- End of hack. own.name = (environment and environment.ownname) or arg[0] or 'mtxrun.lua' own.path = string.gsub(string.match(own.name,"^(.+)[\\/].-$") or ".","\\","/") local ownpath, owntree = own.path, environment and environment.ownpath or own.path own.list = { '.', ownpath , ownpath .. "/../sources", -- HH's development path owntree .. "/../../texmf-local/tex/context/base", owntree .. "/../../texmf-context/tex/context/base", owntree .. "/../../texmf-dist/tex/context/base", owntree .. "/../../texmf/tex/context/base", owntree .. "/../../../texmf-local/tex/context/base", owntree .. "/../../../texmf-context/tex/context/base", owntree .. "/../../../texmf-dist/tex/context/base", owntree .. "/../../../texmf/tex/context/base", } if own.path == "." then table.remove(own.list,1) end local function locate_libs() for l=1,#own.libs do local lib = own.libs[l] for p =1,#own.list do local pth = own.list[p] local filename = pth .. "/" .. lib local found = lfs.isfile(filename) if found then return pth end end end end local function load_libs() local found = locate_libs() if found then for l=1,#own.libs do local filename = found .. "/" .. own.libs[l] local codeblob = loadfile(filename) if codeblob then codeblob() end end else resolvers = nil end end if not resolvers then load_libs() end if not resolvers then print("") print("Mtxrun is unable to start up due to lack of libraries. You may") print("try to run 'lua mtxrun.lua --selfmerge' in the path where this") print("script is located (normally under ..../scripts/context/lua) which") print("will make this script library independent.") os.exit() end logs.setprogram('MTXrun',"TDS Runner Tool 1.26") if environment.arguments["verbose"] then trackers.enable("resolvers.locating") end local instance = resolvers.reset() local trackspec = environment.argument("trackers") or environment.argument("track") if trackspec then trackers.enable(trackspec) end runners = runners or { } -- global messages = messages or { } messages.help = [[ --script run an mtx script (lua prefered method) (--noquotes), no script gives list --execute run a script or program (texmfstart method) (--noquotes) --resolve resolve prefixed arguments --ctxlua run internally (using preloaded libs) --internal run script using built in libraries (same as --ctxlua) --locate locate given filename --autotree use texmf tree cf. env 'texmfstart_tree' or 'texmfstarttree' --tree=pathtotree use given texmf tree (default file: 'setuptex.tmf') --environment=name use given (tmf) environment file --path=runpath go to given path before execution --ifchanged=filename only execute when given file has changed (md checksum) --iftouched=old,new only execute when given file has changed (time stamp) --makestubs create stubs for (context related) scripts --removestubs remove stubs (context related) scripts --stubpath=binpath paths where stubs wil be written --windows create windows (mswin) stubs --unix create unix (linux) stubs --verbose give a bit more info --trackers=list enable given trackers --engine=str target engine --progname=str format or backend --edit launch editor with found file --launch (--all) launch files like manuals, assumes os support --timedrun run a script an time its run --autogenerate regenerate databases if needed (handy when used to run context in an editor) --usekpse use kpse as fallback (when no mkiv and cache installed, often slower) --forcekpse force using kpse (handy when no mkiv and cache installed but less functionality) --prefixes show supported prefixes --generate generate file database --variables show configuration variables --expansions show expanded variables --configurations show configuration order --expand-braces expand complex variable --expand-path expand variable (resolve paths) --expand-var expand variable (resolve references) --show-path show path expansion of ... --var-value report value of variable --find-file report file location --find-path report path of file --pattern=str filter variables ]] runners.applications = { ["lua"] = "luatex --luaonly", ["luc"] = "luatex --luaonly", ["pl"] = "perl", ["py"] = "python", ["rb"] = "ruby", } runners.suffixes = { 'rb', 'lua', 'py', 'pl' } runners.registered = { texexec = { 'texexec.rb', false }, -- context mkii runner (only tool not to be luafied) texutil = { 'texutil.rb', true }, -- old perl based index sorter for mkii (old versions need it) texfont = { 'texfont.pl', true }, -- perl script that makes mkii font metric files texfind = { 'texfind.pl', false }, -- perltk based tex searching tool, mostly used at pragma texshow = { 'texshow.pl', false }, -- perltk based context help system, will be luafied -- texwork = { 'texwork.pl', false }, -- perltk based editing environment, only used at pragma makempy = { 'makempy.pl', true }, mptopdf = { 'mptopdf.pl', true }, pstopdf = { 'pstopdf.rb', true }, -- converts ps (and some more) images, does some cleaning (replaced) -- examplex = { 'examplex.rb', false }, concheck = { 'concheck.rb', false }, runtools = { 'runtools.rb', true }, textools = { 'textools.rb', true }, tmftools = { 'tmftools.rb', true }, ctxtools = { 'ctxtools.rb', true }, rlxtools = { 'rlxtools.rb', true }, pdftools = { 'pdftools.rb', true }, mpstools = { 'mpstools.rb', true }, -- exatools = { 'exatools.rb', true }, xmltools = { 'xmltools.rb', true }, -- luatools = { 'luatools.lua', true }, mtxtools = { 'mtxtools.rb', true }, pdftrimwhite = { 'pdftrimwhite.pl', false } } runners.launchers = { windows = { }, unix = { } } -- like runners.libpath("framework"): looks on script's subpath function runners.libpath(...) package.prepend_libpath(file.dirname(environment.ownscript),...) package.prepend_libpath(file.dirname(environment.ownname) ,...) end function runners.prepare() local checkname = environment.argument("ifchanged") local verbose = environment.argument("verbose") if type(checkname) == "string" and checkname ~= "" then local oldchecksum = file.loadchecksum(checkname) local newchecksum = file.checksum(checkname) if oldchecksum == newchecksum then if verbose then logs.simple("file '%s' is unchanged",checkname) end return "skip" elseif verbose then logs.simple("file '%s' is changed, processing started",checkname) end file.savechecksum(checkname) end local touchname = environment.argument("iftouched") if type(touchname) == "string" and touchname ~= "" then local oldname, newname = string.split(touchname, ",") if oldname and newname and oldname ~= "" and newname ~= "" then if not file.needs_updating(oldname,newname) then if verbose then logs.simple("file '%s' and '%s' have same age",oldname,newname) end return "skip" elseif verbose then logs.simple("file '%s' is older than '%s'",oldname,newname) end end end local runpath = environment.argument("path") if type(runpath) == "string" and not lfs.chdir(runpath) then logs.simple("unable to change to path '%s'",runpath) return "error" end runners.prepare = function() end return "run" end function runners.execute_script(fullname,internal,nosplit) local noquote = environment.argument("noquotes") if fullname and fullname ~= "" then local state = runners.prepare() if state == 'error' then return false elseif state == 'skip' then return true elseif state == "run" then instance.progname = environment.argument("progname") or instance.progname instance.format = environment.argument("format") or instance.format local path, name, suffix, result = file.dirname(fullname), file.basename(fullname), file.extname(fullname), "" if path ~= "" then result = fullname elseif name then name = name:gsub("^int[%a]*:",function() internal = true return "" end ) name = name:gsub("^script:","") if suffix == "" and runners.registered[name] and runners.registered[name][1] then name = runners.registered[name][1] suffix = file.extname(name) end if suffix == "" then -- loop over known suffixes for _,s in pairs(runners.suffixes) do result = resolvers.find_file(name .. "." .. s, 'texmfscripts') if result ~= "" then break end end elseif runners.applications[suffix] then result = resolvers.find_file(name, 'texmfscripts') else -- maybe look on path result = resolvers.find_file(name, 'other text files') end end if result and result ~= "" then if not no_split then local before, after = environment.split_arguments(fullname) -- already done environment.arguments_before, environment.arguments_after = before, after end if internal then arg = { } for _,v in pairs(environment.arguments_after) do arg[#arg+1] = v end environment.ownscript = result dofile(result) else local binary = runners.applications[file.extname(result)] if binary and binary ~= "" then result = binary .. " " .. result end local command = result .. " " .. environment.reconstruct_commandline(environment.arguments_after,noquote) if logs.verbose then logs.simpleline() logs.simple("executing: %s",command) logs.simpleline() logs.simpleline() io.flush() end -- no os.exec because otherwise we get the wrong return value local code = os.execute(command) -- maybe spawn if code == 0 then return true else if binary then binary = file.addsuffix(binary,os.binsuffix) for p in string.gmatch(os.getenv("PATH"),"[^"..io.pathseparator.."]+") do if lfs.isfile(file.join(p,binary)) then return false end end logs.simpleline() logs.simple("This script needs '%s' which seems not to be installed.",binary) logs.simpleline() end return false end end end end end return false end function runners.execute_program(fullname) local noquote = environment.argument("noquotes") if fullname and fullname ~= "" then local state = runners.prepare() if state == 'error' then return false elseif state == 'skip' then return true elseif state == "run" then local before, after = environment.split_arguments(fullname) environment.initialize_arguments(after) fullname = fullname:gsub("^bin:","") local command = fullname .. " " .. (environment.reconstruct_commandline(after or "",noquote) or "") logs.simpleline() logs.simple("executing: %s",command) logs.simpleline() logs.simpleline() io.flush() local code = os.exec(command) -- (fullname,unpack(after)) does not work / maybe spawn return code == 0 end end return false end -- the --usekpse flag will fallback (not default) on kpse (hm, we can better update mtx-stubs) local windows_stub = '@echo off\013\010setlocal\013\010set ownpath=%%~dp0%%\013\010texlua "%%ownpath%%mtxrun.lua" --usekpse --execute %s %%*\013\010endlocal\013\010' local unix_stub = '#!/bin/sh\010mtxrun --usekpse --execute %s \"$@\"\010' function runners.handle_stubs(create) local stubpath = environment.argument('stubpath') or '.' -- 'auto' no longer subpathssupported local windows = environment.argument('windows') or environment.argument('mswin') or false local unix = environment.argument('unix') or environment.argument('linux') or false if not windows and not unix then if os.platform == "unix" then unix = true else windows = true end end for _,v in pairs(runners.registered) do local name, doit = v[1], v[2] if doit then local base = string.gsub(file.basename(name), "%.(.-)$", "") if create then if windows then io.savedata(file.join(stubpath,base..".bat"),string.format(windows_stub,name)) logs.simple("windows stub for '%s' created",base) end if unix then io.savedata(file.join(stubpath,base),string.format(unix_stub,name)) logs.simple("unix stub for '%s' created",base) end else if windows and (os.remove(file.join(stubpath,base..'.bat')) or os.remove(file.join(stubpath,base..'.cmd'))) then logs.simple("windows stub for '%s' removed", base) end if unix and (os.remove(file.join(stubpath,base)) or os.remove(file.join(stubpath,base..'.sh'))) then logs.simple("unix stub for '%s' removed",base) end end end end end function runners.resolve_string(filename) if filename and filename ~= "" then runners.report_location(resolvers.resolve(filename)) end end function runners.locate_file(filename) -- differs from texmfstart where locate appends .com .exe .bat ... todo if filename and filename ~= "" then runners.report_location(resolvers.find_given_file(filename)) end end function runners.locate_platform() runners.report_location(os.platform) end function runners.report_location(result) if logs.verbose then logs.simpleline() if result and result ~= "" then logs.simple(result) else logs.simple("not found") end else io.write(result) end end function runners.edit_script(filename) -- we assume that gvim is present on most systems (todo: also in cnf file) local editor = os.getenv("MTXRUN_EDITOR") or os.getenv("TEXMFSTART_EDITOR") or os.getenv("EDITOR") or 'gvim' local rest = resolvers.resolve(filename) if rest ~= "" then local command = editor .. " " .. rest if logs.verbose then logs.simpleline() logs.simple("starting editor: %s",command) logs.simple_line() logs.simple_line() end os.launch(command) end end function runners.save_script_session(filename, list) local t = { } for i=1,#list do local key = list[i] t[key] = environment.arguments[key] end io.savedata(filename,table.serialize(t,true)) end function runners.load_script_session(filename) if lfs.isfile(filename) then local t = io.loaddata(filename) if t then t = loadstring(t) if t then t = t() end for key, value in pairs(t) do environment.arguments[key] = value end end end end function resolvers.launch(str) -- maybe we also need to test on mtxrun.launcher.suffix environment -- variable or on windows consult the assoc and ftype vars and such local launchers = runners.launchers[os.platform] if launchers then local suffix = file.extname(str) if suffix then local runner = launchers[suffix] if runner then str = runner .. " " .. str end end end os.launch(str) end function runners.launch_file(filename) instance.allresults = true trackers.enable("resolvers.locating") local pattern = environment.arguments["pattern"] if not pattern or pattern == "" then pattern = filename end if not pattern or pattern == "" then logs.simple("provide name or --pattern=") else local t = resolvers.find_files(pattern) if not t or #t == 0 then t = resolvers.find_files("*/" .. pattern) end if not t or #t == 0 then t = resolvers.find_files("*/" .. pattern .. "*") end if t and #t > 0 then if environment.arguments["all"] then for _, v in pairs(t) do logs.simple("launching %s", v) resolvers.launch(v) end else logs.simple("launching %s", t[1]) resolvers.launch(t[1]) end else logs.simple("no match for %s", pattern) end end end function runners.find_mtx_script(filename) local function found(name) local path = file.dirname(name) if path and path ~= "" then return false else local fullname = own and own.path and file.join(own.path,name) return io.exists(fullname) and fullname end end filename = file.addsuffix(filename,"lua") local basename = file.removesuffix(file.basename(filename)) local suffix = file.extname(filename) -- qualified path, raw name local fullname = file.is_qualified_path(filename) and io.exists(filename) and filename if fullname and fullname ~= "" then return fullname end -- current path, raw name fullname = "./" .. filename fullname = io.exists(fullname) and fullname if fullname and fullname ~= "" then return fullname end -- mtx- prefix checking local mtxprefix = (filename:find("^mtx%-") and "") or "mtx-" -- context namespace, mtx- fullname = mtxprefix .. filename fullname = found(fullname) or resolvers.find_file(fullname) if fullname and fullname ~= "" then return fullname end -- context namespace, mtx-s fullname = mtxprefix .. basename .. "s" .. "." .. suffix fullname = found(fullname) or resolvers.find_file(fullname) if fullname and fullname ~= "" then return fullname end -- context namespace, mtx- fullname = mtxprefix .. basename:gsub("s$","") .. "." .. suffix fullname = found(fullname) or resolvers.find_file(fullname) if fullname and fullname ~= "" then return fullname end -- context namespace, just fullname = resolvers.find_file(filename) return fullname end function runners.register_arguments(...) local arguments = environment.arguments_after local passedon = { ... } for i=#passedon,1,-1 do local pi = passedon[i] if pi then table.insert(arguments,1,pi) end end end function runners.execute_ctx_script(filename,...) runners.register_arguments(...) local arguments = environment.arguments_after local fullname = runners.find_mtx_script(filename) or "" if file.extname(fullname) == "cld" then -- handy in editors where we force --autopdf logs.simple("running cld script: %s",filename) table.insert(arguments,1,fullname) table.insert(arguments,"--autopdf") fullname = runners.find_mtx_script("context") or "" end -- retry after generate but only if --autogenerate if fullname == "" and environment.argument("autogenerate") then -- might become the default instance.renewcache = true trackers.enable("resolvers.locating") resolvers.load() -- fullname = runners.find_mtx_script(filename) or "" end -- that should do it if fullname ~= "" then local state = runners.prepare() if state == 'error' then return false elseif state == 'skip' then return true elseif state == "run" then -- load and save ... kind of undocumented arg = { } for _,v in pairs(arguments) do arg[#arg+1] = resolvers.resolve(v) end environment.initialize_arguments(arg) local loadname = environment.arguments['load'] if loadname then if type(loadname) ~= "string" then loadname = file.basename(fullname) end loadname = file.replacesuffix(loadname,"cfg") runners.load_script_session(loadname) end filename = environment.files[1] if logs.verbose then logs.simple("using script: %s\n",fullname) end environment.ownscript = fullname dofile(fullname) local savename = environment.arguments['save'] if savename then local save_list = runners.save_list if save_list and next(save_list) then if type(savename) ~= "string" then savename = file.basename(fullname) end savename = file.replacesuffix(savename,"cfg") runners.save_script_session(savename,save_list) end end return true end else if filename == "" or filename == "help" then local context = resolvers.find_file("mtx-context.lua") trackers.enable("resolvers.locating") if context ~= "" then local result = dir.glob((string.gsub(context,"mtx%-context","mtx-*"))) -- () needed local valid = { } table.sort(result) for i=1,#result do local scriptname = result[i] local scriptbase = string.match(scriptname,".*mtx%-([^%-]-)%.lua") if scriptbase then local data = io.loaddata(scriptname) local banner, version = string.match(data,"[\n\r]logs%.extendbanner%s*%(%s*[\"\']([^\n\r]+)%s*(%d+%.%d+)") if banner then valid[#valid+1] = { scriptbase, version, banner } end end end if #valid > 0 then logs.reportbanner() logs.reportline() logs.simple("no script name given, known scripts:") logs.simple() for k=1,#valid do local v = valid[k] logs.simple("%-12s %4s %s",v[1],v[2],v[3]) end end else logs.simple("no script name given") end else filename = file.addsuffix(filename,"lua") if file.is_qualified_path(filename) then logs.simple("unknown script '%s'",filename) else logs.simple("unknown script '%s' or 'mtx-%s'",filename,filename) end end return false end end function runners.prefixes() logs.reportbanner() logs.reportline() logs.simple(table.concat(resolvers.allprefixes(true)," ")) end function runners.timedrun(filename) -- just for me if filename and filename ~= "" then runners.timed(function() os.execute(filename) end) end end function runners.timed(action) statistics.timed(action) end -- this is a bit dirty ... first we store the first filename and next we -- split the arguments so that we only see the ones meant for this script -- ... later we will use the second half local filename = environment.files[1] or "" local ok = true local before, after = environment.split_arguments(filename) environment.arguments_before, environment.arguments_after = before, after environment.initialize_arguments(before) instance.engine = environment.argument("engine") or 'luatex' instance.progname = environment.argument("progname") or 'context' instance.lsrmode = environment.argument("lsr") or false -- maybe the unset has to go to this level local is_mkii_stub = runners.registered[file.removesuffix(file.basename(filename))] if environment.argument("usekpse") or environment.argument("forcekpse") or is_mkii_stub then os.setenv("engine","") os.setenv("progname","") local remapper = { otf = "opentype fonts", ttf = "truetype fonts", ttc = "truetype fonts", pfb = "type1 fonts", other = "other text files", } local function kpse_initialized() texconfig.kpse_init = true local t = os.clock() local k = kpse.original.new("luatex",instance.progname) local dummy = k:find_file("mtxrun.lua") -- so that we're initialized logs.simple("kpse fallback with progname '%s' initialized in %s seconds",instance.progname,os.clock()-t) kpse_initialized = function() return k end return k end local find_file = resolvers.find_file local show_path = resolvers.show_path if environment.argument("forcekpse") then function resolvers.find_file(name,kind) return (kpse_initialized():find_file(resolvers.clean_path(name),(kind ~= "" and (remapper[kind] or kind)) or "tex") or "") or "" end function resolvers.show_path(name) return (kpse_initialized():show_path(name)) or "" end elseif environment.argument("usekpse") or is_mkii_stub then resolvers.load() function resolvers.find_file(name,kind) local found = find_file(name,kind) or "" if found ~= "" then return found else return (kpse_initialized():find_file(resolvers.clean_path(name),(kind ~= "" and (remapper[kind] or kind)) or "tex") or "") or "" end end function resolvers.show_path(name) local found = show_path(name) or "" if found ~= "" then return found else return (kpse_initialized():show_path(name)) or "" end end end function runners.loadbase() end else function runners.loadbase(...) if not resolvers.load(...) then logs.simple("forcing cache reload") instance.renewcache = true trackers.enable("resolvers.locating") if not resolvers.load(...) then logs.simple("the resolver databases are not present or outdated") end end end end resolvers.load_tree(environment.argument('tree')) if environment.argument("selfmerge") then -- embed used libraries runners.loadbase() local found = locate_libs() if found then utils.merger.selfmerge(own.name,own.libs,{ found }) end elseif environment.argument("selfclean") then -- remove embedded libraries runners.loadbase() utils.merger.selfclean(own.name) elseif environment.argument("selfupdate") then runners.loadbase() trackers.enable("resolvers.locating") resolvers.update_script(own.name,"mtxrun") elseif environment.argument("ctxlua") or environment.argument("internal") then -- run a script by loading it (using libs) runners.loadbase() ok = runners.execute_script(filename,true) elseif environment.argument("script") or environment.argument("scripts") then -- run a script by loading it (using libs), pass args runners.loadbase() if is_mkii_stub then ok = runners.execute_script(filename,false,true) else ok = runners.execute_ctx_script(filename) end elseif environment.argument("execute") then -- execute script runners.loadbase() ok = runners.execute_script(filename) elseif environment.argument("direct") then -- equals bin: runners.loadbase() ok = runners.execute_program(filename) elseif environment.argument("edit") then -- edit file runners.loadbase() runners.edit_script(filename) elseif environment.argument("launch") then runners.loadbase() runners.launch_file(filename) elseif environment.argument("makestubs") then -- make stubs (depricated) runners.handle_stubs(true) elseif environment.argument("removestubs") then -- remove stub (depricated) runners.loadbase() runners.handle_stubs(false) elseif environment.argument("resolve") then -- resolve string runners.loadbase() runners.resolve_string(filename) elseif environment.argument("locate") then -- locate file runners.loadbase() runners.locate_file(filename) elseif environment.argument("platform") or environment.argument("show-platform") then -- locate platform runners.loadbase() runners.locate_platform() elseif environment.argument("prefixes") then runners.loadbase() runners.prefixes() elseif environment.argument("timedrun") then -- locate platform runners.loadbase() runners.timedrun(filename) elseif environment.argument("variables") or environment.argument("show-variables") then -- luatools: runners.execute_ctx_script("mtx-base","--variables",filename) resolvers.load("nofiles") resolvers.listers.variables(false,environment.argument("pattern")) elseif environment.argument("expansions") or environment.argument("show-expansions") then -- luatools: runners.execute_ctx_script("mtx-base","--expansions",filename) resolvers.load("nofiles") resolvers.listers.expansions(false,environment.argument("pattern")) elseif environment.argument("configurations") or environment.argument("show-configurations") then -- luatools: runners.execute_ctx_script("mtx-base","--configurations",filename) resolvers.load("nofiles") resolvers.listers.configurations() elseif environment.argument("find-file") then -- luatools: runners.execute_ctx_script("mtx-base","--find-file",filename) resolvers.load() local pattern = environment.argument("pattern") local format = environment.arguments["format"] or instance.format if not pattern then runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.find_files,environment.files,format) elseif type(pattern) == "string" then instance.allresults = true -- brrrr resolvers.for_files(resolvers.find_files,{ pattern }, format) end elseif environment.argument("find-path") then -- luatools: runners.execute_ctx_script("mtx-base","--find-path",filename) resolvers.load() local path = resolvers.find_path(filename, instance.my_format) if logs.verbose then logs.simple(path) else print(path) end elseif environment.argument("expand-braces") then -- luatools: runners.execute_ctx_script("mtx-base","--expand-braces",filename) resolvers.load("nofiles") runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.expand_braces, environment.files) elseif environment.argument("expand-path") then -- luatools: runners.execute_ctx_script("mtx-base","--expand-path",filename) resolvers.load("nofiles") runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.expand_path, environment.files) elseif environment.argument("expand-var") or environment.argument("expand-variable") then -- luatools: runners.execute_ctx_script("mtx-base","--expand-var",filename) resolvers.load("nofiles") runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.expand_var, environment.files) elseif environment.argument("show-path") or environment.argument("path-value") then -- luatools: runners.execute_ctx_script("mtx-base","--show-path",filename) resolvers.load("nofiles") runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.show_path, environment.files) elseif environment.argument("var-value") or environment.argument("show-value") then -- luatools: runners.execute_ctx_script("mtx-base","--show-value",filename) resolvers.load("nofiles") runners.register_arguments(filename) environment.initialize_arguments(environment.arguments_after) resolvers.for_files(resolvers.var_value,environment.files) elseif environment.argument("format-path") then -- luatools: runners.execute_ctx_script("mtx-base","--format-path",filename) resolvers.load() logs.simple(caches.getwritablepath("format")) elseif environment.argument("pattern") then -- luatools runners.execute_ctx_script("mtx-base","--pattern='" .. environment.argument("pattern") .. "'",filename) elseif environment.argument("generate") then -- luatools instance.renewcache = true trackers.enable("resolvers.locating") resolvers.load() elseif environment.argument("make") or environment.argument("ini") or environment.argument("compile") then -- luatools: runners.execute_ctx_script("mtx-base","--make",filename) resolvers.load() trackers.enable("resolvers.locating") environment.make_format(filename) elseif environment.argument("run") then -- luatools runners.execute_ctx_script("mtx-base","--run",filename) elseif environment.argument("fmt") then -- luatools runners.execute_ctx_script("mtx-base","--fmt",filename) elseif environment.argument("help") and filename=='base' then -- luatools runners.execute_ctx_script("mtx-base","--help") elseif environment.argument("help") or filename=='help' or filename == "" then logs.help(messages.help) elseif filename:find("^bin:") then runners.loadbase() ok = runners.execute_program(filename) elseif is_mkii_stub then -- execute mkii script runners.loadbase() ok = runners.execute_script(filename,false,true) elseif false then runners.loadbase() ok = runners.execute_ctx_script(filename) if not ok then ok = runners.execute_script(filename) end else runners.execute_ctx_script("mtx-base",filename) end if logs.verbose then logs.simpleline() logs.simple("runtime: %0.3f seconds",os.runtime()) end if os.type ~= "windows" then texio.write("\n") end if ok == false then ok = 1 elseif ok == true then ok = 0 end os.exit(ok)