diff options
Diffstat (limited to 'tex/context/base/mkiv/char-utf.lua')
| -rw-r--r-- | tex/context/base/mkiv/char-utf.lua | 94 |
1 files changed, 66 insertions, 28 deletions
diff --git a/tex/context/base/mkiv/char-utf.lua b/tex/context/base/mkiv/char-utf.lua index fb47b5208..7d8ca219b 100644 --- a/tex/context/base/mkiv/char-utf.lua +++ b/tex/context/base/mkiv/char-utf.lua @@ -86,33 +86,40 @@ characters.decomposed = decomposed local graphemes = characters.graphemes local collapsed = characters.collapsed +local combined = characters.combined local mathlists = characters.mathlists if graphemes then mark(graphemes) mark(collapsed) + mark(combined) mark(mathlists) else graphemes = allocate() collapsed = allocate() + combined = allocate() mathlists = allocate() characters.graphemes = graphemes characters.collapsed = collapsed + characters.combined = combined characters.mathlists = mathlists local function backtrack(v,last,target) local vs = v.specials - if vs and #vs == 3 and vs[1] == "char" then - local one = vs[2] - local two = vs[3] - local first = utfchar(one) - local second = utfchar(two) .. last - collapsed[first..second] = target - backtrack(data[one],second,target) + if vs and #vs == 3 then + local kind = vs[1] + if kind == "char" or kind == "with" then + local one = vs[2] + local two = vs[3] + local first = utfchar(one) + local second = utfchar(two) .. last + collapsed[first..second] = target + backtrack(data[one],second,target) + end end end @@ -141,28 +148,39 @@ else if vs then local kind = vs[1] local size = #vs - if kind == "char" and size == 3 then -- what if more than 3 - -- - local one = vs[2] - local two = vs[3] - local first = utfchar(one) - local second = utfchar(two) - local combination = utfchar(unicode) - -- - collapsed[first..second] = combination - backtrack(data[one],second,combination) - -- sort of obsolete: - local cgf = graphemes[first] - if not cgf then - cgf = { [second] = combination } - graphemes[first] = cgf - else - cgf[second] = combination + if kind == "char" then + if size == 3 then + local one = vs[2] + local two = vs[3] + local first = utfchar(one) + local second = utfchar(two) + local combination = utfchar(unicode) + -- + collapsed[first..second] = combination + backtrack(data[one],second,combination) + -- sort of obsolete: + local cgf = graphemes[first] + if not cgf then + cgf = { [second] = combination } + graphemes[first] = cgf + else + cgf[second] = combination + end + end + if size > 2 and (v.mathclass or v.mathspec) then + setlist(unicode,vs,2,"specials") + end + elseif kind == "with" then + if size == 3 then + combined[utfchar(vs[2])..utfchar(vs[3])] = utfchar(unicode) + end + elseif kind == "compat" then + if size == 3 then + combined[utfchar(vs[2])..utfchar(vs[3])] = utfchar(unicode) + end + if size > 2 and (v.mathclass or v.mathspec) then + setlist(unicode,vs,2,"specials") end - -- - end - if (kind == "char" or kind == "compat") and (size > 2) and (v.mathclass or v.mathspec) then - setlist(unicode,vs,2,"specials") end end local ml = v.mathlist @@ -182,6 +200,7 @@ else if storage then storage.register("characters/graphemes", graphemes, "characters.graphemes") storage.register("characters/collapsed", collapsed, "characters.collapsed") + storage.register("characters/combined", combined, "characters.combined") storage.register("characters/mathlists", mathlists, "characters.mathlists") end @@ -226,6 +245,25 @@ function utffilters.collapse(str,filename) end end +local p_combine = nil -- only for tex + +local function prepare() + local tree = utfchartabletopattern(combined) + p_combine = Cs((tree/combined + p_utf8character)^0) +end + +function utffilters.combine(str) -- not in files + -- we could merge collapse into combine ... maybe + if not p_combine then + prepare() + end + if not str or str == "" or #str == 1 then + return str + else + return lpegmatch(p_combine,str) or str + end +end + local p_decompose = nil local function prepare() |