src/luatexja-xkanji.lua

   1 ------------------------------------------------------------------------
   2 -- MAIN PROCESS STEP 3: insert \xkanjiskip (prefix: none)
   3 ------------------------------------------------------------------------
   4 local ltjs = luatexja.stack
   5
   6 local node_type = node.type
   7 local node_new = node.new
   8 local node_prev = node.prev
   9 local node_next = node.next
  10 local node_copy = node.copy
  11 local has_attr = node.has_attribute
  12 local set_attr = node.set_attribute
  13 local node_insert_before = node.insert_before
  14 local node_insert_after = node.insert_after
  15 local node_hpack = node.hpack
  16 local round = tex.round
  17
  18 local id_penalty = node.id('penalty')
  19 local id_glyph = node.id('glyph')
  20 local id_glue = node.id('glue')
  21 local id_glue_spec = node.id('glue_spec')
  22 local id_kern = node.id('kern')
  23 local id_hlist = node.id('hlist')
  24 local id_ins = node.id('ins')
  25 local id_mark = node.id('mark')
  26 local id_adjust = node.id('adjust')
  27 local id_math = node.id('math')
  28 local id_whatsit = node.id('whatsit')
  29
  30 local attr_icflag = luatexbase.attributes['ltj@icflag']
  31 local attr_autospc = luatexbase.attributes['ltj@autospc']
  32 local attr_autoxspc = luatexbase.attributes['ltj@autoxspc']
  33 local attr_curjfnt = luatexbase.attributes['ltj@curjfnt']
  34 local max_dimen = 1073741823
  35
  36 local ITALIC = 1
  37 local TEMPORARY = 2
  38 local FROM_JFM = 3
  39 local KINSOKU = 4
  40 local LINE_END = 5
  41 local KANJI_SKIP = 6
  42 local XKANJI_SKIP = 7
  43 local PACKED = 8
  44
  45 local kanji_skip
  46 local xkanji_skip
  47
  48 -- (glyph_node nr) ... (node nq) <GLUE> ,,, (node np)
  49 local np, nq, nrc, nrf
  50 local np_spc, nr_spc
  51 local no_skip = 0
  52 local after_schar = 1
  53 local after_wchar = 2 -- nr is a Japanese glyph_node
  54 local insert_skip = no_skip
  55 local head
  56
  57 local function is_japanese_glyph_node(p)
  58    return p and (p.id==id_glyph)
  59    and (p.font==has_attr(p,attr_curjfnt))
  60 end
  61
  62 local function get_zero_glue()
  63    local g = node_new(id_glue_spec)
  64    g.width = 0; g.stretch_order = 0; g.stretch = 0
  65    g.shrink_order = 0; g.shrink = 0
  66    return g
  67 end
  68
  69 local function skip_table_to_spec(n)
  70    local g = node_new(id_glue_spec)
  71    local st = ltjs.get_skip_table(n, ltj.box_stack_level)
  72    g.width = st.width; g.stretch = st.stretch; g.shrink = st.shrink
  73    g.stretch_order = st.stretch_order; g.shrink_order = st.shrink_order
  74    return g
  75 end
  76
  77 local function add_glue_spec(g,h)
  78    -- g := g + h
  79    g.width = g.width + h.width
  80    if g.stretch_order<h.stretch_order then
  81       g.stretch_order = h.stretch_order
  82       g.stretch = h.stretch
  83    elseif g.stretch_order==h.stretch_order then
  84       g.stretch = g.stretch + h.stretch
  85    end
  86    if g.shrink_order<h.shrink_order then
  87       g.shrink_order = h.shrink_order
  88       g.shrink = h.shrink
  89    elseif g.shrink_order==h.shrink_order then
  90       g.shrink = g.shrink + h.shrink
  91    end
  92 end
  93
  94 -- lowest part of \xkanjiskip
  95 local function get_xkanji_skip_from_jfm(pf)
  96    if pf then
  97       local px = { ltj.font_metric_table[pf].size,
  98                    ltj.font_metric_table[pf].jfm }
  99       local i = ltj.metrics[px[2]].xkanjiskip
 100       if i then
 101          return { round(i[1]*px[1]), round(i[2]*px[1]), round(i[3]*px[1]) }
 102       else return nil
 103       end
 104    else return nil
 105    end
 106 end
 107
 108 local function insert_xkanjiskip_node(q, f, p)
 109    if nr_spc[2] or np_spc[2] then
 110       local g = node_new(id_glue); g.subtype = 0
 111       if xkanji_skip.width==max_dimen then -- use xkanjiskip from JFM
 112          local gx = node_new(id_glue_spec)
 113          gx.stretch_order = 0; gx.shrink_order = 0
 114          local ak = get_xkanji_skip_from_jfm(f)
 115          if ak then
 116             gx.width = ak[1]; gx.stretch = ak[2]; gx.shrink = ak[3]
 117          else gx = get_zero_glue() -- fallback
 118          end
 119          g.spec = gx
 120       else g.spec=node_copy(xkanji_skip)
 121       end
 122       local h = node_prev(p)
 123       if h  and has_attr(h, attr_icflag)==TEMPORARY then
 124          if h.id==id_kern then
 125             g.spec.width = g.spec.width + h.kern
 126             set_attr(g,attr_icflag,XKANJI_SKIP)
 127             node_insert_after(head, q, g)
 128             head = node.remove(head, h)
 129             node.free(h)
 130          else
 131             add_glue_spec(h.spec, g.spec)
 132             node.free(g.spec); node.free(g)
 133          end
 134       else
 135          set_attr(g,attr_icflag,XKANJI_SKIP)
 136          node_insert_after(head, q, g)
 137       end
 138    end
 139 end
 140
 141 local function insert_ascii_kanji_xkskip(q, p)
 142    if ltjs.get_penalty_table('xsp', p.char, 3, ltj.box_stack_level)<=1 then return end
 143    insert_xkanjiskip_node(q, p.font, p)
 144 end
 145
 146 local function insert_kanji_ascii_xkskip(q, p)
 147    local g = true
 148    local c = p.char
 149    while p.components and p.subtype
 150       and math.floor(p.subtype/2)%2==1 do
 151       p = p.components; c = p.char
 152    end
 153    if ltjs.get_penalty_table('xsp', c, 3, ltj.box_stack_level)%2 == 1 then
 154       if ltjs.get_penalty_table('xsp', nrc, 3, ltj.box_stack_level)%2 == 0 then g = false end
 155    else g = false
 156    end
 157    if g then insert_xkanjiskip_node(q, nrf, p) end
 158 end
 159
 160 local function set_insert_skip_after_achar(p)
 161    local c = p.char
 162    while p.components and p.subtype
 163       and math.floor(p.subtype/2)%2 == 1 do
 164       p=node.tail(p.components); c = p.char
 165    end
 166   if ltjs.get_penalty_table('xsp', c, 3, ltj.box_stack_level)>=2 then
 167      insert_skip = after_schar
 168   else
 169      insert_skip = no_skip
 170   end
 171 end
 172
 173 -- lowest part of \kanjiskip
 174 local function get_kanji_skip_from_jfm(pf)
 175    if pf then
 176       local px = { ltj.font_metric_table[pf].size,
 177                    ltj.font_metric_table[pf].jfm }
 178       local i = ltj.metrics[px[2]].kanjiskip
 179       if i then
 180          return { round(i[1]*px[1]), round(i[2]*px[1]), round(i[3]*px[1]) }
 181       else return nil
 182       end
 183    else return nil
 184    end
 185 end
 186
 187 local function insert_kanji_skip(ope, p)
 188    local g = node_new(id_glue); g.subtype=0
 189    if nr_spc[1] or np_spc[1] then
 190       if kanji_skip.width==max_dimen then -- use kanjiskip from JFM
 191          local gx = node_new(id_glue_spec);
 192          gx.stretch_order = 0; gx.shrink_order = 0
 193          local bk = get_kanji_skip_from_jfm(nrf)
 194          local ak = get_kanji_skip_from_jfm(p.font)
 195          if bk then
 196             if ak then
 197                gx.width = round(ltj.ja_diffmet_rule(bk[1], ak[1]))
 198                gx.stretch = round(ltj.ja_diffmet_rule(bk[2], ak[2]))
 199                gx.shrink = -round(ltj.ja_diffmet_rule(-bk[3], -ak[3]))
 200             else
 201                gx.width = bk[1]; gx.stretch = bk[2]; gx.shrink = bk[3]
 202             end
 203          elseif ak then
 204             gx.width = ak[1]; gx.stretch = ak[2]; gx.shrink = ak[3]
 205          else gx = get_zero_glue() -- fallback
 206          end
 207          g.spec = gx
 208       else g.spec=node_copy(kanji_skip)
 209       end
 210    else g.spec = get_zero_glue()
 211    end
 212    local h = node_prev(p)
 213    if h  and has_attr(h, attr_icflag)==TEMPORARY then
 214       if h.id==id_kern then
 215          g.spec.width = g.spec.width + h.kern
 216          head = node.remove(head, h)
 217          node.free(h)
 218          set_attr(g,attr_icflag,KANJI_SKIP)
 219          ope(head, p, g)
 220       else
 221          add_glue_spec(h.spec, g.spec)
 222          node.free(g.spec); node.free(g)
 223       end
 224    else
 225       set_attr(g,attr_icflag,KANJI_SKIP)
 226       ope(head, p, g)
 227    end
 228 end
 229
 230 -- When p is a glyph_node ...
 231 local function insks_around_char()
 232    if is_japanese_glyph_node(np) then
 233       if insert_skip==after_wchar then
 234          insert_kanji_skip(node_insert_before, np)
 235       elseif insert_skip==after_schar then
 236          insert_ascii_kanji_xkskip(nq, np)
 237       end
 238       insert_skip=after_wchar
 239       nrc = np.char; nrf = np.font; nr_spc = np_spc
 240    else
 241       if insert_skip==after_wchar then
 242          insert_kanji_ascii_xkskip(nq, np)
 243       end
 244       set_insert_skip_after_achar(np); nr_spc = np_spc
 245    end
 246    nq = np
 247 end
 248
 249 -- Return first and last glyph nodes in a hbox
 250 local first_char = nil
 251 local last_char = nil
 252 local find_first_char = nil
 253 local function check_box(box_ptr)
 254    local p = box_ptr; local found_visible_node = false
 255    if not p then
 256       find_first_char = false; first_char = nil; last_char = nil
 257       return true
 258    end
 259    while p do
 260       if p.id==id_glyph then
 261          repeat
 262             if find_first_char then
 263                first_char = p; find_first_char = false
 264             end
 265             last_char = p; found_visible_node = true; p=node_next(p)
 266             if not p then return found_visible_node end
 267          until p.id~=id_glyph
 268       end
 269       if p.id==id_hlist then
 270          if has_attr(p, attr_icflag)==PACKED then
 271             for q in node.traverse_id(id_glyph, p.head) do
 272                if find_first_char then
 273                   first_char = q; find_first_char = false
 274                end
 275                last_char = q; found_visible_node = true
 276             end
 277          else
 278             if p.shift==0 then
 279                if check_box(p.head) then found_visible_node = true end
 280             else if find_first_char then
 281                   find_first_char = false
 282                else
 283                   last_char = nil
 284                end
 285             end
 286          end
 287       elseif p.id==id_ins    or p.id==id_mark
 288           or p.id==id_adjust or p.id==id_whatsit
 289           or p.id==id_penalty then
 290          p=p
 291       else
 292          found_visible_node = true
 293          if find_first_char then
 294             find_first_char = false
 295          else
 296             last_char = nil
 297          end
 298       end
 299       p = node_next(p)
 300    end
 301    return found_visible_node
 302 end
 303
 304 -- When np is a hlist_node ...
 305 local function insks_around_hbox()
 306    if np.shift==0 then
 307       find_first_char = true; first_char = nil; last_char = nil
 308       if check_box(np.head) then
 309          -- first char
 310          if is_japanese_glyph_node(first_char) then
 311             nrc = first_char.char; nrf = first_char.font
 312             if insert_skip==after_schar then
 313                insert_ascii_kanji_xkskip(nq, first_char)
 314             elseif insert_skip==after_wchar then
 315                np_spc = { has_attr(first_char, attr_autospc)==1,
 316                           has_attr(first_char, attr_autoxspc)==1 }
 317                insert_kanji_skip(node_insert_before, np)
 318             end
 319             insert_skip = after_wchar
 320          elseif first_char then
 321             if insert_skip==after_wchar then
 322                insert_kanji_ascii_xkskip(nq, first_char)
 323             end
 324             set_insert_skip_after_achar(first_char)
 325          end
 326          -- last char
 327          if is_japanese_glyph_node(last_char) then
 328             insert_skip = after_wchar
 329             nrc = last_char.char; nrf = last_char.font
 330             nr_spc = { has_attr(last_char, attr_autospc)==1,
 331                        has_attr(last_char, attr_autoxspc)==1 }
 332             if is_japanese_glyph_node(node_next(np)) then
 333                insert_kanji_skip(node_insert_after, np)
 334             end
 335          elseif last_char then
 336             set_insert_skip_after_achar(last_char)
 337             nr_spc = { has_attr(last_char, attr_autospc)==1,
 338                        has_attr(last_char, attr_autoxspc)==1 }
 339          else insert_skip = no_skip
 340          end
 341       else insert_skip = no_skip
 342       end
 343    else insert_skip = no_skip
 344    end
 345    nq = np
 346 end
 347
 348 -- When np is a penalty ...
 349 local function insks_around_penalty()
 350    nq = np
 351 end
 352
 353 -- When np is a kern ...
 354 --
 355 local function insks_around_kern()
 356    if np.subtype==1 then -- \kern or \/
 357       local i = has_attr(np, attr_icflag)
 358       if not i or i==FROM_JFM then -- \kern
 359          insert_skip = no_skip
 360       elseif i==ITALIC or i==LINE_END or i==TEMPORARY then
 361          nq = np
 362       end
 363    elseif np.subtype==2 then
 364       -- (np = kern from \accent) .. (accent char) .. (kern from \accent) .. (glyph)
 365       np = node_next(node_next(np))
 366    else  -- kern from TFM
 367       nq = np
 368    end
 369 end
 370
 371 -- When np is a math_node ...
 372 local function insks_around_math()
 373    local g = { char = -1 }
 374    if (np.subtype==0) and (insert_skip==after_wchar) then
 375       insert_kanji_ascii_xkskip(nq, g)
 376       insert_skip = no_skip
 377    else
 378       nq = np; set_insert_skip_after_achar(g); nr_spc = np_spc
 379    end
 380 end
 381
 382 function ltj.int_insert_kanji_skip(ahead)
 383    kanji_skip=skip_table_to_spec('kanjiskip')
 384    xkanji_skip=skip_table_to_spec('xkanjiskip')
 385    head = ahead
 386    np = head; nq = nil; insert_skip = no_skip
 387    while np do
 388       np_spc = { (has_attr(np, attr_autospc)==1),
 389                  (has_attr(np, attr_autoxspc)==1) }
 390       if np.id==id_glyph then
 391          repeat
 392             np_spc = { has_attr(np, attr_autospc)==1,
 393                        has_attr(np, attr_autoxspc)==1 }
 394             insks_around_char(); np=node_next(np)
 395          until (not np) or np.id~=id_glyph
 396       else
 397          if np.id==id_hlist then
 398             insks_around_hbox()
 399          elseif np.id==id_penalty then
 400             insks_around_penalty()
 401          elseif np.id==id_kern then
 402             insks_around_kern()
 403          elseif np.id==id_math then
 404             insks_around_math()
 405          elseif np.id==id_ins    or np.id==id_mark
 406              or np.id==id_adjust or np.id==id_whatsit then
 407             -- do nothing
 408             np = np
 409          else
 410             -- rule, disc, glue, margin_kern
 411             insert_skip = no_skip
 412          end
 413          np = node_next(np)
 414       end
 415    end
 416    return head
 417 end