diff options
Diffstat (limited to 'tex/context/base/data-exp.lua')
| -rw-r--r-- | tex/context/base/data-exp.lua | 336 |
1 files changed, 336 insertions, 0 deletions
diff --git a/tex/context/base/data-exp.lua b/tex/context/base/data-exp.lua new file mode 100644 index 000000000..785679275 --- /dev/null +++ b/tex/context/base/data-exp.lua @@ -0,0 +1,336 @@ +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 = collapse_path(p) + 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 stripper = lpegCs( +--~ lpegpatterns.unspacer * lpegpatterns.unsingle +--~ + lpegpatterns.undouble * lpegpatterns.unspacer +--~ ) + +local expandhome = lpegP("~") / "$HOME" -- environment.homedir + +local dodouble = lpegP('"')/"" * (expandhome + (1 - lpegP('"')))^0 * lpegP('"')/"" +local dosingle = lpegP("'")/"" * (expandhome + (1 - lpegP("'")))^0 * lpegP("'")/"" +local dostring = (expandhome + 1 )^0 + +local stripper = lpegCs( + lpegpatterns.unspacer * (dosingle + dodouble + dostring) * lpegpatterns.unspacer +) + +function resolvers.checked_variable(str) -- assumes str is a string + return lpegmatch(stripper,str) or str +end + +-- The path splitter: + +-- A config (optionally) has the paths split in tables. Internally +-- we join them and split them after the expansion has taken place. This +-- is more convenient. + +--~ local checkedsplit = string.checkedsplit + +local cache = { } + +local splitter = lpegCt(lpeg.splitat(lpegS(ostype == "windows" and ";" or ":;"))) -- maybe add , + +local function split_configuration_path(str) -- beware, this can be either a path or a { specification } + if str then + local found = cache[str] + if not found then + if str == "" then + found = { } + else + str = gsub(str,"\\","/") + local split = lpegmatch(splitter,str) + found = { } + for i=1,#split do + local s = split[i] + if not find(s,"^{*unset}*") then + found[#found+1] = s + end + end + if trace_expansions then + report_resolvers("splitting path specification '%s'",str) + for k=1,#found do + report_resolvers("% 4i: %s",k,found[k]) + end + end + cache[str] = found + end + end + return found + end +end + +resolvers.split_configuration_path = split_configuration_path + +function resolvers.split_path(str) + if type(str) == 'table' then + return str + else + return split_configuration_path(str) + end +end + +function resolvers.join_path(str) + if type(str) == 'table' then + return file.join_path(str) + else + return str + end +end + +-- The next function scans directories and returns a hash where the +-- entries are either strings or tables. + +-- starting with . or .. etc or funny char + +--~ local l_forbidden = lpegS("~`!#$%^&*()={}[]:;\"\'||\\/<>,?\n\r\t") +--~ local l_confusing = lpegP(" ") +--~ local l_character = lpegpatterns.utf8 +--~ local l_dangerous = lpegP(".") + +--~ local l_normal = (l_character - l_forbidden - l_confusing - l_dangerous) * (l_character - l_forbidden - l_confusing^2)^0 * lpegP(-1) +--~ ----- l_normal = l_normal * lpegCc(true) + lpegCc(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 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 + +--~ print(table.serialize(resolvers.scan_files("t:/sources"))) |