diff options
Diffstat (limited to 'tex/context/base/data-exp.lua')
| -rw-r--r-- | tex/context/base/data-exp.lua | 357 |
1 files changed, 200 insertions, 157 deletions
diff --git a/tex/context/base/data-exp.lua b/tex/context/base/data-exp.lua index c67e97bb1..0a7396171 100644 --- a/tex/context/base/data-exp.lua +++ b/tex/context/base/data-exp.lua @@ -11,16 +11,20 @@ local concat, sort = table.concat, table.sort local lpegmatch, lpegpatterns = lpeg.match, lpeg.patterns local Ct, Cs, Cc, Carg, P, C, S = lpeg.Ct, lpeg.Cs, lpeg.Cc, lpeg.Carg, lpeg.P, lpeg.C, lpeg.S local type, next = type, next +local isdir = lfs.isdir local ostype = os.type -local collapsepath = file.collapsepath +local collapsepath, joinpath, basename = file.collapsepath, file.join, file.basename 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 trace_globbing = true trackers.register("resolvers.globbing", function(v) trace_globbing = v end) local report_expansions = logs.reporter("resolvers","expansions") +local report_globbing = logs.reporter("resolvers","globbing") -local resolvers = resolvers +local resolvers = resolvers +local resolveprefix = resolvers.resolve -- 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 @@ -123,7 +127,7 @@ local function splitpathexpr(str, newlist, validate) -- I couldn't resist lpeggi local old = str str = lpegmatch(l_rest, str) until old == str - until old == str -- or not find(str,"{") + until old == str -- or not find(str,"{",1,true) str = lpegmatch(stripper_1,str) if validate then for s in gmatch(str,"[^,]+") do @@ -177,34 +181,28 @@ end -- {a,b,c/{p,q/{x,y,z},w}v,d/{p,q,r}} -- {$SELFAUTODIR,$SELFAUTOPARENT}{,{/share,}/texmf{-local,.local,}/web2c} -local cleanup = lpeg.replacer { - { "!" , "" }, - { "\\" , "/" }, -} +local usedhomedir = nil +local donegation = (P("!") /"" )^0 +local doslashes = (P("\\")/"/" + 1)^0 -function resolvers.cleanpath(str) -- tricky, maybe only simple paths - local doslashes = (P("\\")/"/" + 1)^0 - local donegation = (P("!") /"" )^0 - local homedir = lpegmatch(Cs(donegation * doslashes),environment.homedir or "") - if homedir == "~" or homedir == "" or not lfs.isdir(homedir) then - if trace_expansions then - report_expansions("no home dir set, ignoring dependent paths") - end - function resolvers.cleanpath(str) - if not str or find(str,"~") then - return "" -- special case - else - return lpegmatch(cleanup,str) +local function expandedhome() + if not usedhomedir then + usedhomedir = lpegmatch(Cs(donegation * doslashes),environment.homedir or "") + if usedhomedir == "~" or usedhomedir == "" or not isdir(usedhomedir) then + if trace_expansions then + report_expansions("no home dir set, ignoring dependent path using current path") end - end - else - local dohome = ((P("~")+P("$HOME"))/homedir)^0 - local cleanup = Cs(donegation * dohome * doslashes) - function resolvers.cleanpath(str) - return str and lpegmatch(cleanup,str) or "" + usedhomedir = "." end end - return resolvers.cleanpath(str) + return usedhomedir +end + +local dohome = ((P("~")+P("$HOME")+P("%HOME%"))/expandedhome)^0 +local cleanup = Cs(donegation * dohome * doslashes) + +resolvers.cleanpath = function(str) + return str and lpegmatch(cleanup,str) or "" end -- print(resolvers.cleanpath("")) @@ -216,11 +214,18 @@ end -- This one strips quotes and funny tokens. -local expandhome = P("~") / "$HOME" -- environment.homedir or "home:" +-- we have several options here: +-- +-- expandhome = P("~") / "$HOME" : relocateble +-- expandhome = P("~") / "home:" : relocateble +-- expandhome = P("~") / environment.homedir : frozen but unexpanded +-- expandhome = P("~") = dohome : frozen and expanded -local dodouble = P('"')/"" * (expandhome + (1 - P('"')))^0 * P('"')/"" -local dosingle = P("'")/"" * (expandhome + (1 - P("'")))^0 * P("'")/"" -local dostring = (expandhome + 1 )^0 +local expandhome = P("~") / "$HOME" + +local dodouble = P('"') / "" * (expandhome + (1 - P('"')))^0 * P('"') / "" +local dosingle = P("'") / "" * (expandhome + (1 - P("'")))^0 * P("'") / "" +local dostring = (expandhome + 1 )^0 local stripper = Cs( lpegpatterns.unspacer * (dosingle + dodouble + dostring) * lpegpatterns.unspacer @@ -285,7 +290,7 @@ end function resolvers.joinpath(str) if type(str) == 'table' then - return file.joinpath(str) + return joinpath(str) else return str end @@ -293,25 +298,25 @@ end -- The next function scans directories and returns a hash where the -- entries are either strings or tables. - +-- -- starting with . or .. etc or funny char - ---~ local l_forbidden = S("~`!#$%^&*()={}[]:;\"\'||\\/<>,?\n\r\t") ---~ local l_confusing = P(" ") ---~ local l_character = lpegpatterns.utf8 ---~ local l_dangerous = P(".") - ---~ local l_normal = (l_character - l_forbidden - l_confusing - l_dangerous) * (l_character - l_forbidden - l_confusing^2)^0 * P(-1) ---~ ----- l_normal = l_normal * Cc(true) + Cc(false) - ---~ local function test(str) ---~ print(str,lpegmatch(l_normal,str)) ---~ end ---~ test("ヒラギノ明朝 Pro W3") ---~ test("..ヒラギノ明朝 Pro W3") ---~ test(":ヒラギノ明朝 Pro W3;") ---~ test("ヒラギノ明朝 /Pro W3;") ---~ test("ヒラギノ明朝 Pro W3") +-- +-- local l_forbidden = S("~`!#$%^&*()={}[]:;\"\'||\\/<>,?\n\r\t") +-- local l_confusing = P(" ") +-- local l_character = lpegpatterns.utf8 +-- local l_dangerous = P(".") +-- +-- local l_normal = (l_character - l_forbidden - l_confusing - l_dangerous) * (l_character - l_forbidden - l_confusing^2)^0 * P(-1) +-- ----- l_normal = l_normal * Cc(true) + Cc(false) +-- +-- local function test(str) +-- print(str,lpegmatch(l_normal,str)) +-- end +-- test("ヒラギノ明朝 Pro W3") +-- test("..ヒラギノ明朝 Pro W3") +-- test(":ヒラギノ明朝 Pro W3;") +-- test("ヒラギノ明朝 /Pro W3;") +-- test("ヒラギノ明朝 Pro W3") -- a lot of this caching can be stripped away when we have ssd's everywhere -- @@ -319,41 +324,67 @@ end local attributes, directory = lfs.attributes, lfs.dir -local weird = P(".")^1 + lpeg.anywhere(S("~`!#$%^&*()={}[]:;\"\'||<>,?\n\r\t")) -local timer = { } -local scanned = { } -local nofscans = 0 -local scancache = { } +local weird = P(".")^1 + lpeg.anywhere(S("~`!#$%^&*()={}[]:;\"\'||<>,?\n\r\t")) +local lessweird = P(".")^1 + lpeg.anywhere(S("~`#$%^&*:;\"\'||<>,?\n\r\t")) +local timer = { } +local scanned = { } +local nofscans = 0 +local scancache = { } +local fullcache = { } +----- simplecache = { } +local nofsharedscans = 0 + +-- So, we assume either a lowercase name or a mixed case one but only one such case +-- as having Foo fOo foo FoO FOo etc on the system is braindead in any sane project. -local function scan(files,spec,path,n,m,r) - local full = (path == "" and spec) or (spec .. path .. '/') +local function scan(files,remap,spec,path,n,m,r,onlyone,tolerant) + local full = path == "" and spec or (spec .. path .. '/') local dirs = { } local nofdirs = 0 + local pattern = tolerant and lessweird or weird for name in directory(full) do - if not lpegmatch(weird,name) then - local mode = attributes(full..name,'mode') - if mode == 'file' then + if not lpegmatch(pattern,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 } + local lower = lower(name) + local paths = files[lower] + if paths then + if onlyone then + -- forget about it else - f[#f+1] = path + if type(paths) == "string" then + files[lower] = { paths, path } + else + paths[#paths+1] = path + end + if name ~= lower then + local rl = remap[lower] + if not rl then + remap[lower] = name + r = r + 1 + elseif trace_globbing and rl ~= name then + report_globbing("confusing filename, name: %a, lower: %a, already: %a",name,lower,rl) + end + end end else -- probably unique anyway - files[name] = path - local lower = lower(name) + files[lower] = path if name ~= lower then - files["remap:"..lower] = name - r = r + 1 + local rl = remap[lower] + if not rl then + remap[lower] = name + r = r + 1 + elseif trace_globbing and rl ~= name then + report_globbing("confusing filename, name: %a, lower: %a, already: %a",name,lower,rl) + end end end - elseif mode == 'directory' then + elseif mode == "directory" then m = m + 1 nofdirs = nofdirs + 1 if path ~= "" then - dirs[nofdirs] = path..'/'..name + dirs[nofdirs] = path .. "/" .. name else dirs[nofdirs] = name end @@ -363,113 +394,72 @@ local function scan(files,spec,path,n,m,r) if nofdirs > 0 then sort(dirs) for i=1,nofdirs do - files, n, m, r = scan(files,spec,dirs[i],n,m,r) + files, remap, n, m, r = scan(files,remap,spec,dirs[i],n,m,r,onlyonce,tolerant) end end scancache[sub(full,1,-2)] = files - return files, n, m, r + return files, remap, n, m, r end -local fullcache = { } - -function resolvers.scanfiles(path,branch,usecache) - statistics.starttiming(timer) - local realpath = resolvers.resolve(path) -- no shortcut +function resolvers.scanfiles(path,branch,usecache,onlyonce,tolerant) + local realpath = resolveprefix(path) if usecache then - local files = fullcache[realpath] - if files then + local content = fullcache[realpath] + if content then if trace_locating then - report_expansions("using caches scan of path %a, branch %a",path,branch or path) + report_expansions("using cached scan of path %a, branch %a",path,branch or path) end - return files + nofsharedscans = nofsharedscans + 1 + return content end end + -- + statistics.starttiming(timer) if trace_locating then report_expansions("scanning path %a, branch %a",path,branch or path) end - local files, n, m, r = scan({ },realpath .. '/',"",0,0,0) - files.__path__ = path -- can be selfautoparent:texmf-whatever - files.__files__ = n - files.__directories__ = m - files.__remappings__ = r - if trace_locating then - report_expansions("%s files found on %s directories with %s uppercase remappings",n,m,r) + local content + if isdir(realpath) then + local files, remap, n, m, r = scan({ },{ },realpath .. '/',"",0,0,0,onlyonce,tolerant) + content = { + metadata = { + path = path, -- can be selfautoparent:texmf-whatever + files = n, + directories = m, + remappings = r, + }, + files = files, + remap = remap, + } + if trace_locating then + report_expansions("%s files found on %s directories with %s uppercase remappings",n,m,r) + end + else + content = { + metadata = { + path = path, -- can be selfautoparent:texmf-whatever + files = 0, + directories = 0, + remappings = 0, + }, + files = { }, + remap = { }, + } + if trace_locating then + report_expansions("invalid path %a",realpath) + end end if usecache then scanned[#scanned+1] = realpath - fullcache[realpath] = files + fullcache[realpath] = content end nofscans = nofscans + 1 statistics.stoptiming(timer) - return files + return content end -local function simplescan(files,spec,path) -- first match only, no map and such - local full = (path == "" and spec) or (spec .. path .. '/') - local dirs = { } - local nofdirs = 0 - for name in directory(full) do - if not lpegmatch(weird,name) then - local mode = attributes(full..name,'mode') - if mode == 'file' then - if not files[name] then - -- only first match - files[name] = path - end - elseif mode == 'directory' then - nofdirs = nofdirs + 1 - if path ~= "" then - dirs[nofdirs] = path..'/'..name - else - dirs[nofdirs] = name - end - end - end - end - if nofdirs > 0 then - sort(dirs) - for i=1,nofdirs do - files = simplescan(files,spec,dirs[i]) - end - end - return files -end - -local simplecache = { } -local nofsharedscans = 0 - function resolvers.simplescanfiles(path,branch,usecache) - statistics.starttiming(timer) - local realpath = resolvers.resolve(path) -- no shortcut - if usecache then - local files = simplecache[realpath] - if not files then - files = scancache[realpath] - if files then - nofsharedscans = nofsharedscans + 1 - end - end - if files then - if trace_locating then - report_expansions("using caches scan of path %a, branch %a",path,branch or path) - end - return files - end - end - if trace_locating then - report_expansions("scanning path %a, branch %a",path,branch or path) - end - local files = simplescan({ },realpath .. '/',"") - if trace_locating then - report_expansions("%s files found",table.count(files)) - end - if usecache then - scanned[#scanned+1] = realpath - simplecache[realpath] = files - end - nofscans = nofscans + 1 - statistics.stoptiming(timer) - return files + return resolvers.scanfiles(path,branch,usecache,true,true) -- onlyonce end function resolvers.scandata() @@ -482,4 +472,57 @@ function resolvers.scandata() } end ---~ print(table.serialize(resolvers.scanfiles("t:/sources"))) +function resolvers.get_from_content(content,path,name) -- or (content,name) + if not content then + return + end + local files = content.files + if not files then + return + end + local remap = content.remap + if not remap then + return + end + if name then + -- this one resolves a remapped name + local used = lower(name) + return path, remap[used] or used + else + -- this one does a lookup and resolves a remapped name + local name = path + local used = lower(name) + local path = files[used] + if path then + return path, remap[used] or used + end + end +end + +local nothing = function() end + +function resolvers.filtered_from_content(content,pattern) + if content and type(pattern) == "string" then + local pattern = lower(pattern) + local files = content.files + local remap = content.remap + if files and remap then + local n = next(files) + local function iterator() + while n do + local k = n + n = next(files,k) + if find(k,pattern) then + return files[k], remap and remap[k] or k + end + end + end + return iterator + end + end + return nothing +end + + +-- inspect(resolvers.simplescanfiles("e:/temporary/mb-mp")) +-- inspect(resolvers.scanfiles("e:/temporary/mb-mp")) |