diff options
Diffstat (limited to 'tex/context/base/mkxl/lang-tra.lmt')
| -rw-r--r-- | tex/context/base/mkxl/lang-tra.lmt | 160 |
1 files changed, 105 insertions, 55 deletions
diff --git a/tex/context/base/mkxl/lang-tra.lmt b/tex/context/base/mkxl/lang-tra.lmt index b3fedc7c1..259165928 100644 --- a/tex/context/base/mkxl/lang-tra.lmt +++ b/tex/context/base/mkxl/lang-tra.lmt @@ -9,7 +9,11 @@ if not modules then modules = { } end modules ['lang-tra'] = { -- The indic transliterations was researched by kauĊika and after some experiments -- we settled on the current approach (mappings and a more specific lpeg). -local concat = table.concat +-- Todo: initial and final in addition to mapping so that we can do hebrew and +-- such. + +local concat, setmetatableindex = table.concat, table.setmetatableindex +local nospaces = string.nospaces local utfbyte, utfchar, utfsplit, utfvalues = utf.byte, utf.char, utf.split, utf.values local C, Cc, Cs, lpegmatch = lpeg.C, lpeg.Cc, lpeg.Cs, lpeg.match local utfchartabletopattern = lpeg.utfchartabletopattern @@ -30,6 +34,9 @@ local removenode = nuts.remove local texsetattribute = tex.setattribute +local registervalue = attributes.registervalue +local getvalue = attributes.getvalue + local transliteration = { } languages.transliteration = transliteration @@ -41,9 +48,10 @@ local context = context local zwj = utf.char(0x200D) -local lastmapping = 0 +-- local lastmapping = 0 local loadedmappings = { } local loadedlibraries = { } +local exceptions = { } local report = logs.reporter("transliteration") local trace = false trackers.register("languages.transliteration", function(v) trace = v end) @@ -80,21 +88,23 @@ local converters = { ["mapping"] = function(m) local t_mapping = m.mapping if t_mapping then + local t_exceptions = m.exceptions local p = Cs ( ( utfchartabletopattern(t_mapping) / t_mapping + C(utfcharacterpattern) )^0 ) -- lpeg.print(p) - return function(s) - return lpegmatch(p,s) or s + return function(s,e) + return (e and e[s]) or t_exceptions[s] or lpegmatch(p,s) or s end else return false end end, ["default"] = function(m) - return function(s) - return s + local t_exceptions = m.exceptions + return function(s,e) + return (e and e[s]) or t_exceptions[s] or s end end, } @@ -109,19 +119,21 @@ function transliteration.use(library) if transliterations then for name, d in table.sortedhash(transliterations) do local vector = d.vector + if not d.exceptions then + d.exceptions = { } + end if vector then - report("vector %a in %a is %sloaded with index %i",name,library," already",d.attribute) + report("vector %a in %a is %sloaded",name,library," already") else - lastmapping = lastmapping + 1 d.vector = (converters[name] or converters.mapping or converters.default)(d) or (converters.default)(d) - report("vector %a in set %a is %sloaded with index %i",name,library,"",lastmapping) + report("vector %a in set %a is %sloaded",name,library,"") end - d.attribute = lastmapping d.library = library d.name = name + d.mapping = nil -- for now, saves memory loadedmappings[name] = d - loadedmappings[lastmapping] = d + loadedmappings[nospaces(name)] = d end else report("library %a has no transliterations",library) @@ -140,28 +152,47 @@ enable = function() enable = false end -function transliteration.set(vector) +function transliteration.register(parent,name) + local p = exceptions[parent] + if p then + if trace then + report("%a has exceptions that default to %a",name,parent) + end + exceptions[name] = setmetatableindex({ },p) + else + if trace then + report("%a has independent exceptions",name) + end + exceptions[name] = { } + end +end + +function transliteration.set(name,vector) if enable then enable() end - local m = loadedmappings[vector] - local a = m and m.attribute or unsetvalue + local a = registervalue(a_transliteration, { + m = loadedmappings[vector], + e = exceptions[name], + }) if trace then - report("setting transliteration %s",vector) + report("setting transliteration %i, name %a, vector %a",a,name,vector) end texsetattribute(a_transliteration,a) end -function transliteration.register(vector) - if enable then - enable() - end - local m = loadedmappings[vector] - local a = m and m.attribute or unsetvalue - if trace then - report("registering transliteration %s",vector) +function transliteration.exception(name,old,new) + local m = loadedmappings[name] + if m then + m.exceptions[old] = new + else + local e = exceptions[name] + if not e then + e = { } + exceptions[name] = e + end + e[old] = new end - return a end -- When there is need I will improve the performance of the next handler. @@ -169,6 +200,7 @@ end function transliteration.handler(head) local aprev = nil local vector = nil + local except = nil local current = head local first = nil local last = nil @@ -180,34 +212,36 @@ function transliteration.handler(head) -- actually we can generalize the replacer elsewhere local function flush(nxt) - -- we can do some optimization here by having the split as replacement - local old = concat(list,"",1,size) - local new = vector(old) - if old ~= new then - if trace then - report("old: %s",old) - report("new: %s",new) - end - local c = first - local x = false - for s in utfvalues(new) do - if x then - head, c = insertafter(head,c,copynode(first)) - setchar(c,s) - else - setchar(c,s) - if c == last then - x = true + if vector then + -- we can do some optimization here by having the split as replacement + local old = concat(list,"",1,size) + local new = vector(old,except) + if old ~= new then + if trace then + report("old: %s",old) + report("new: %s",new) + end + local c = first + local x = false + for s in utfvalues(new) do + if x then + head, c = insertafter(head,c,copynode(first)) + setchar(c,s) else - c = getnext(c) + setchar(c,s) + if c == last then + x = true + else + c = getnext(c) + end end end - end - if not x then - while c do - head, c = removenode(head,c,true) - if c == nxt then - break + if not x then + while c do + head, c = removenode(head,c,true) + if c == nxt then + break + end end end end @@ -226,9 +260,19 @@ function transliteration.handler(head) size = 0 end aprev = a - vector = loadedmappings[a] - if vector then - vector = vector.vector + local data = getvalue(a_transliteration,a) + if data then + local m = data.m + if m then + vector = m.vector + except = data.e + else + vector = nil + except = nil + end + else + vector = nil + except = nil end end if not first then @@ -269,14 +313,20 @@ interfaces.implement { implement { name = "settransliteration", - arguments = "string", + arguments = "2 strings", actions = transliteration.set, } implement { name = "registertransliteration", - arguments = "string", - actions = { transliteration.register, context }, + arguments = "3 strings", + actions = transliteration.register, +} + +implement { + name = "transliterationexception", + arguments = "3 strings", + actions = transliteration.exception, } nodes.tasks.prependaction("processors", "normalizers", "languages.transliteration.handler", nil, "nut", "disabled" ) |