src/ltj-inputbuf.lua

   1 --
   2 -- luatexja/ltj-inputbuf.lua
   3 --
   4
   5 luatexja.load_module('base');      local ltjb = luatexja.base
   6 luatexja.load_module('charrange'); local ltjc = luatexja.charrange
   7
   8 require("unicode")
   9 local utflen = utf.len
  10 local utfbyte = utf.byte
  11 local utfchar = utf.char
  12 local node_new = node.new
  13 local node_free = node.free
  14 local id_glyph = node.id('glyph')
  15 local getcatcode, getcount = tex.getcatcode, tex.getcount
  16 local ltjc_is_japanese_char_curlist = ltjc.is_japanese_char_curlist
  17
  18 --- the following function is modified from jafontspec.lua (by K. Maeda).
  19 --- Instead of "%", we use U+FFFFF for suppressing spaces.
  20 --DEBUG require"socket"
  21 local time_line = 0
  22 local start_time_measure, stop_time_measure
  23    = ltjb.start_time_measure, ltjb.stop_time_measure
  24 local function add_comment(buffer)
  25    start_time_measure('inputbuf')
  26    local i = utflen(buffer)
  27    while (i>0) and (getcatcode(utfbyte(buffer, i))==1
  28                  or getcatcode(utfbyte(buffer, i))==2) do
  29       i=i-1
  30    end
  31    if i>0 then
  32       local c = utfbyte(buffer, i)
  33       if c>=0x80 then
  34          local ct = getcatcode(c)
  35          local te = tex.endlinechar
  36          local ctl = (te ~= -1) and (getcatcode(te)==5) and (getcatcode(getcount('ltjlineendcomment'))==14)
  37          -- Is the catcode of endline character is 5 (end-of-line)?
  38          -- Is the catcode of \ltjlineendcomment (new comment char) is 14 (comment)?
  39          if ((ct==11) or (ct==12)) and ctl then
  40             if ltjc_is_japanese_char_curlist(c) then
  41                buffer = buffer .. utfchar(getcount('ltjlineendcomment'))
  42             end
  43          end
  44       end
  45    end
  46    stop_time_measure('inputbuf')
  47    return buffer
  48 end
  49
  50 luatexbase.add_to_callback('process_input_buffer',
  51    add_comment,'ltj.process_input_buffer')
  52
  53 --EOF