ltj-jfont.lua etc.: use data of UTR#50

vert feature の適用範囲外にある文字について，
* 「和文フォントへの置換」時に，U+2025 -> U+FE30 のように，Unicode の
  - Vertical Forms (U+FE10--U+FE1F)
  -  CJK Compatibility Forms の一部 (U+FE30--U+FE44)
  へマッピングできるものは置き換える
* 「JAcharの位置補正」時に，UTF#50 で R or Tr と指定されており，
  かつ vert feature による置き換え先が定義されていないグリフは
  90度回転状態にする
* 例えば，小塚明朝では U=205A と U+FE30 が同一のグリフを用いることになっている．
  この場合 UTR#50 では U+2052 は R，U+FE30 は U となっているが，
  「回転しない」方を優先．

diff --git a/src/ltj-jfont.lua b/src/ltj-jfont.lua
index 2b88a68..baf413a 100644 (file)
--- a/src/ltj-jfont.lua
+++ b/src/ltj-jfont.lua
@@ -3,7 +3,7 @@
 --
 luatexbase.provides_module({
   name = 'luatexja.jfont',
-  date = '2017/05/05',
+  date = '2017/08/09',
   description = 'Loader for Japanese fonts',
 })
 module('luatexja.jfont', package.seeall)
@@ -718,7 +718,165 @@ font_extra_info = {}
 local font_extra_info = font_extra_info -- key: fontnumber
 local font_extra_basename = {} -- key: basename
 
--- IVS and vertical metrics
+local list_rotate_glyphs
+do
+  -- output of function_utr50.lua
+  local function rotate_in_utr50(i)
+  -- UTR#50 revision 17; 2016-10-20
+    if (0x0000<=i)and(i<0x00A7) then return true end
+    if (0x00A8<=i)and(i<0x00A9) then return true end
+    if (0x00AA<=i)and(i<0x00AE) then return true end
+    if (0x00AF<=i)and(i<0x00B1) then return true end
+    if (0x00B2<=i)and(i<0x00BC) then return true end
+    if (0x00BF<=i)and(i<0x00D7) then return true end
+    if (0x00D8<=i)and(i<0x00F7) then return true end
+    if (0x00F8<=i)and(i<0x02EA) then return true end
+    if (0x02EC<=i)and(i<0x0860) then return true end
+    if (0x08A0<=i)and(i<0x1100) then return true end
+    if (0x1200<=i)and(i<0x1401) then return true end
+    if (0x1680<=i)and(i<0x18B0) then return true end
+    if (0x1900<=i)and(i<0x1C90) then return true end
+    if (0x1CC0<=i)and(i<0x2016) then return true end
+    if (0x2017<=i)and(i<0x2020) then return true end
+    if (0x2022<=i)and(i<0x2030) then return true end
+    if (0x2032<=i)and(i<0x203B) then return true end
+    if (0x203D<=i)and(i<0x2042) then return true end
+    if (0x2043<=i)and(i<0x2047) then return true end
+    if (0x204A<=i)and(i<0x2051) then return true end
+    if (0x2052<=i)and(i<0x2065) then return true end
+    if (0x2066<=i)and(i<0x20DD) then return true end
+    if (0x20E1<=i)and(i<0x20E2) then return true end
+    if (0x20E5<=i)and(i<0x2100) then return true end
+    if (0x2102<=i)and(i<0x2103) then return true end
+    if (0x210A<=i)and(i<0x210F) then return true end
+    if (0x2110<=i)and(i<0x2113) then return true end
+    if (0x2115<=i)and(i<0x2116) then return true end
+    if (0x2118<=i)and(i<0x211E) then return true end
+    if (0x2124<=i)and(i<0x2125) then return true end
+    if (0x2126<=i)and(i<0x2127) then return true end
+    if (0x2128<=i)and(i<0x2129) then return true end
+    if (0x212A<=i)and(i<0x212E) then return true end
+    if (0x212F<=i)and(i<0x2135) then return true end
+    if (0x2140<=i)and(i<0x2145) then return true end
+    if (0x214B<=i)and(i<0x214C) then return true end
+    if (0x214E<=i)and(i<0x214F) then return true end
+    if (0x218A<=i)and(i<0x218C) then return true end
+    if (0x2190<=i)and(i<0x221E) then return true end
+    if (0x221F<=i)and(i<0x2234) then return true end
+    if (0x2236<=i)and(i<0x2300) then return true end
+    if (0x2308<=i)and(i<0x230C) then return true end
+    if (0x2320<=i)and(i<0x2324) then return true end
+    if (0x2329<=i)and(i<0x232B) then return true end
+    if (0x232C<=i)and(i<0x237D) then return true end
+    if (0x239B<=i)and(i<0x23BE) then return true end
+    if (0x23CE<=i)and(i<0x23CF) then return true end
+    if (0x23D0<=i)and(i<0x23D1) then return true end
+    if (0x23DC<=i)and(i<0x23E2) then return true end
+    if (0x2423<=i)and(i<0x2424) then return true end
+    if (0x2500<=i)and(i<0x25A0) then return true end
+    if (0x261A<=i)and(i<0x2620) then return true end
+    if (0x2768<=i)and(i<0x2776) then return true end
+    if (0x2794<=i)and(i<0x2B12) then return true end
+    if (0x2B30<=i)and(i<0x2B50) then return true end
+    if (0x2B5A<=i)and(i<0x2BB8) then return true end
+    if (0x2BEC<=i)and(i<0x2BF0) then return true end
+    if (0x2C00<=i)and(i<0x2E80) then return true end
+    if (0x3008<=i)and(i<0x3012) then return true end
+    if (0x3014<=i)and(i<0x3020) then return true end
+    if (0x3030<=i)and(i<0x3031) then return true end
+    if (0x30A0<=i)and(i<0x30A1) then return true end
+    if (0x30FC<=i)and(i<0x30FD) then return true end
+    if (0xA4D0<=i)and(i<0xA960) then return true end
+    if (0xA980<=i)and(i<0xAC00) then return true end
+    if (0xD800<=i)and(i<0xE000) then return true end
+    if (0xFB00<=i)and(i<0xFE10) then return true end
+    if (0xFE20<=i)and(i<0xFE30) then return true end
+    if (0xFE49<=i)and(i<0xFE50) then return true end
+    if (0xFE58<=i)and(i<0xFE5F) then return true end
+    if (0xFE63<=i)and(i<0xFE67) then return true end
+    if (0xFE70<=i)and(i<0xFF01) then return true end
+    if (0xFF08<=i)and(i<0xFF0A) then return true end
+    if (0xFF0D<=i)and(i<0xFF0E) then return true end
+    if (0xFF1A<=i)and(i<0xFF1F) then return true end
+    if (0xFF3B<=i)and(i<0xFF3C) then return true end
+    if (0xFF3D<=i)and(i<0xFF3E) then return true end
+    if (0xFF3F<=i)and(i<0xFF40) then return true end
+    if (0xFF5B<=i)and(i<0xFFE0) then return true end
+    if (0xFFE3<=i)and(i<0xFFE4) then return true end
+    if (0xFFE8<=i)and(i<0xFFF0) then return true end
+    if (0xFFF9<=i)and(i<0xFFFC) then return true end
+    if (0xFFFE<=i)and(i<0x10200) then return true end
+    if (0x10280<=i)and(i<0x103E0) then return true end
+    if (0x10400<=i)and(i<0x10570) then return true end
+    if (0x10600<=i)and(i<0x10780) then return true end
+    if (0x10800<=i)and(i<0x108B0) then return true end
+    if (0x108E0<=i)and(i<0x10940) then return true end
+    if (0x109A0<=i)and(i<0x10AA0) then return true end
+    if (0x10AC0<=i)and(i<0x10BB0) then return true end
+    if (0x10C00<=i)and(i<0x10C50) then return true end
+    if (0x10C80<=i)and(i<0x10D00) then return true end
+    if (0x10E60<=i)and(i<0x10E80) then return true end
+    if (0x11000<=i)and(i<0x11250) then return true end
+    if (0x11280<=i)and(i<0x11380) then return true end
+    if (0x11400<=i)and(i<0x114E0) then return true end
+    if (0x11600<=i)and(i<0x116D0) then return true end
+    if (0x11700<=i)and(i<0x11740) then return true end
+    if (0x118A0<=i)and(i<0x11900) then return true end
+    if (0x11AC0<=i)and(i<0x11B00) then return true end
+    if (0x11C00<=i)and(i<0x11CC0) then return true end
+    if (0x12000<=i)and(i<0x12550) then return true end
+    if (0x16800<=i)and(i<0x16A70) then return true end
+    if (0x16AD0<=i)and(i<0x16B90) then return true end
+    if (0x16F00<=i)and(i<0x16FA0) then return true end
+    if (0x1BC00<=i)and(i<0x1BCB0) then return true end
+    if (0x1D200<=i)and(i<0x1D250) then return true end
+    if (0x1D400<=i)and(i<0x1D800) then return true end
+    if (0x1E000<=i)and(i<0x1E030) then return true end
+    if (0x1E800<=i)and(i<0x1E8E0) then return true end
+    if (0x1E900<=i)and(i<0x1E960) then return true end
+    if (0x1EE00<=i)and(i<0x1EF00) then return true end
+    if (0x1F800<=i)and(i<0x1F900) then return true end
+    if (0xE0000<=i)and(i<0xE0080) then return true end
+    if (0xE0100<=i)and(i<0xE01F0) then return true end
+  end
+   list_rotate_glyphs = function (dest, id)
+      if id.specification and id.resources then
+         local rot = {}
+         for i,_ in pairs(id.characters) do
+           if rotate_in_utr50(i) then rot[i] = true end
+         end
+         if id.resources.sequences then
+        for _,i in pairs(id.resources.sequences) do
+           if i.order[1]== 'vert' and i.type == 'gsub_single' and i.steps then
+              for _,j in pairs(i.steps) do
+                 if type(j)=='table' then 
+                    if type(j,coverage)=='table' then
+                       for i,_ in pairs(j.coverage) do rot[i]=nil end
+                    end
+                 end
+              end
+           end
+        end; end
+        -- コードポイントが共有されているグリフについて
+         if id.resources.duplicates then
+        for i,v in pairs(id.resources.duplicates) do
+           local f = rot[i]
+            for j,_ in pairs(v) do f = f and rot[j] end
+            rot[i]=f
+            for j,_ in pairs(v) do rot[j] = f end
+         end; end
+        
+        for i,_ in pairs(rot) do
+           dest = dest or {}
+           dest[i] = dest[i] or {}
+           dest[i].rotation = true
+         end
+      end
+      return dest
+   end
+end
+
+-- vertical metrics
 local prepare_fl_data
 local supply_vkern_table
 do
@@ -855,7 +1013,7 @@ end
 
 --
 do
-   local cache_ver = 13
+   local cache_ver = 14
    local checksum = file.checksum
 
    local function prepare_extra_data_base(id)
@@ -875,6 +1033,7 @@ do
         else
            local dat = nil
            dat = prepare_fl_data(dat, id)
+           dat = list_rotate_glyphs(dat, id)
            font_extra_basename[bname] = dat or {}
            ltjb.save_cache( v,
                             {
@@ -978,20 +1137,21 @@ end
 
 ------------------------------------------------------------------------
 -- make table of vertical glyphs which does not covered by vert feature
+-- nor UTR#50
 ------------------------------------------------------------------------
 do
 ------------------------------------------------------------------------
 -- VERT VARIANT TABLE
 ------------------------------------------------------------------------
   local vert_form_table = {
-     [0x2013]=0xFE32, [0x2014]=0xFE31, [0x2025]=0xFE30, [0x2026]=0xFE19,
+     [0x3001]=0xFE11, [0x3002]=0xFE12, [0x3016]=0xFE17, [0x3017]=0xFE18,
+     [0x2026]=0xFE19,
+     [0x2025]=0xFE30, [0x2014]=0xFE31, [0x2013]=0xFE32, [0xFF3F]=0xFE33,
      [0xFF08]=0xFE35, [0xFF09]=0xFE36, [0xFF5B]=0xFE37, [0xFF5D]=0xFE38,
      [0x3014]=0xFE39, [0x3015]=0xFE3A, [0x3010]=0xFE3B, [0x3011]=0xFE3C,
      [0x300A]=0xFE3D, [0x300B]=0xFE3E, [0x3008]=0xFE3F, [0x3009]=0xFE40,
      [0x300C]=0xFE41, [0x300D]=0xFE42, [0x300E]=0xFE43, [0x300F]=0xFE44,
-     [0xFF3B]=0xFE47, [0xFF3D]=0xFE48, [0xFF3F]=0xFE33,
-     [0x2190]=0x2191, [0x2191]=0x2192, [0x2192]=0x2193, [0x2193]=0x2190,
-     [0x21E6]=0x21E7, [0x21E7]=0x21E8, [0x21E8]=0x21E9, [0x21E9]=0x21E6,
+     [0xFF3B]=0xFE47, [0xFF3D]=0xFE48, 
   }
   local function add_vform(coverage, vform, ft)
      if type(coverage)~='table' then return end

diff --git a/src/ltj-pretreat.lua b/src/ltj-pretreat.lua
index c0757cf..985914d 100644 (file)
--- a/src/ltj-pretreat.lua
+++ b/src/ltj-pretreat.lua
@@ -149,7 +149,6 @@ local function set_box_stack_level(head, mode)
            if ltjf_font_metric_table[nf].vert_activated then
               local pc = getchar(p)
               pc = ltjf_font_metric_table[nf].vform[pc]
-              if pc then setchar(p,  pc) end
            end
         end
       end

diff --git a/src/ltj-setwidth.lua b/src/ltj-setwidth.lua
index 007ccb0..83a4644 100644 (file)
--- a/src/ltj-setwidth.lua
+++ b/src/ltj-setwidth.lua
@@ -57,7 +57,6 @@ local attr_ykblshift = luatexbase.attributes['ltj@ykblshift']
 local attr_tkblshift = luatexbase.attributes['ltj@tkblshift']
 local attr_icflag = luatexbase.attributes['ltj@icflag']
 
-local ltjf_font_metric_table = ltjf.font_metric_table
 local ltjf_font_extra_info = ltjf.font_extra_info
 
 local PACKED       = luatexja.icflag_table.PACKED
@@ -144,6 +143,33 @@ end
 luatexja.setwidth.capsule_glyph_yoko = capsule_glyph_yoko
 
 -- 和文文字の位置補正（縦）
+-- UTR#50 で R もしくは Tr と指定されているが，縦組用グリフがないもの
+local function capsule_glyph_tate_rot(p, met, char_data, head, dir, asc)
+   fshift.down = char_data.down; fshift.left = char_data.left
+   fshift = call_callback("luatexja.set_width", fshift, met, char_data)
+   local kbl = has_attr(p, attr_tkblshift) or 0
+   -- f*: whd specified in JFM
+   local pwidth, pheight,pdepth = getwhd(p)
+   local fwidth = char_data.width or pwidth
+   local fheight= char_data.height or pheight
+   local fdepth = char_data.depth or pdepth
+   local q
+   head, q = node_remove(head, p)
+   local xo, yo = getoffsets(p)
+   setoffsets(p, xo + char_data.align*(fwidth-pwidth) - fshift.left,
+              yo - fshift.down - asc);
+   setnext(p, nil)
+   local box = node_new(id_hlist)
+   setwhd(box, fwidth, fheight, fdepth)
+   setfield(box, 'head', p)
+   setfield(box, 'shift', kbl)
+   setdir(box, dir)
+   set_attr(box, attr_icflag, PACKED)
+   head = q and node_insert_before(head, q, box)
+      or node_insert_after(head, node_tail(head), box)
+   return q, head, box
+end
+
 local function capsule_glyph_tate(p, met, char_data, head, dir)
    if not char_data then return node_next(p), head end
    local ascent, descent = met.ascent, met.descent
@@ -151,10 +177,13 @@ local function capsule_glyph_tate(p, met, char_data, head, dir)
    do
       local pf = getfont(p)
       local pc = getchar(p)
-      setchar(p, pc)
-      pwidth = ltjf_font_extra_info[pf] and  ltjf_font_extra_info[pf][pc]
-        and ltjf_font_extra_info[pf][pc].vwidth
-        and ltjf_font_extra_info[pf][pc].vwidth * met.size or (ascent+descent)
+      local cei = ltjf_font_extra_info[pf] and  ltjf_font_extra_info[pf][pc]
+      if cei and met.vert_activated then
+        if cei.rotation then
+           return capsule_glyph_tate_rot(p, met, char_data, head, dir, 0.5*(ascent-descent))
+        end
+      end
+      pwidth = (cei and cei.vwidth) and cei.vwidth * met.size or (ascent+descent)
       pwidth = pwidth + (met.v_advance[pc] or 0)
       ascent = met.v_origin[pc] and ascent - met.v_origin[pc] or ascent
    end

diff --git a/test/test51a-vert_vrt2.pdf b/test/test51a-vert_vrt2.pdf
index 5ab6b18..db00840 100644 (file)
Binary files a/test/test51a-vert_vrt2.pdf and b/test/test51a-vert_vrt2.pdf differ

diff --git a/test/test51a-vert_vrt2.tex b/test/test51a-vert_vrt2.tex
index d4348a2..3465bb0 100644 (file)
--- a/test/test51a-vert_vrt2.tex
+++ b/test/test51a-vert_vrt2.tex
@@ -1,7 +1,7 @@
 \documentclass{ltjarticle}
 \usepackage[margin=15mm,centering]{geometry}
 \usepackage{lltjext,cancel,luacode}
-
+%\usepackage{lua-visual-debug}
 \begin{luacode}
 local provides_feature = luaotfload.aux.provides_feature
 local function myfeat_sub(fn,t,a)
@@ -52,7 +52,7 @@ end
 \def\TEST{
   左←↑↓→☜☝☟☞⇦⇧⇩⇨⬅⬆⬇➡右¶\ltjjachar"253B\ltjjachar"0165\relax
   あ：いーう，え、お．か。き；け\ltjjachar"201Cこ\ltjjachar"301D%
-  さ\ltjjachar"301Fし\ltjjachar"201Dす
+  さ\ltjjachar"301Fし\ltjjachar"201Dす\ltjjachar"2015せ\ltjjachar"2014そ
   ‥…〜ー＝―‐゠㍉㌔㌢㍍㌘㌧㌃㌶㍑㍗㌍㌦㌣㌫㍊㌻㍾㍽㍼㍻＝≒≠≡}
 
 

diff --git a/tool/function_utr50.lua b/tool/function_utr50.lua
new file mode 100644 (file)
index 0000000..bb85379
--- /dev/null
+++ b/tool/function_utr50.lua
@@ -0,0 +1,32 @@
+local utr_revision = 17
+print('  local function rotate_in_utr50(i)')
+print('  -- UTR#50 revision ' ..  utr_revision)
+local fh = io.open('VerticalOrientation-' .. utr_revision .. '.txt')
+local t = {}
+
+for c in fh:lines() do
+    if c:match('(%x+)%.%.(%x+)%s+;%sT-[rR]') then
+       local b, e = c:match('(%x+)%.%.(%x+)%s+;')
+       b, e = tonumber(b,16), tonumber(e,16)
+       for i=b,e do t[i]=true end
+    elseif c:match('(%x+)%s+;%sT-[rR]') then
+       local b = c:match('(%x+)%s+;')
+       t[tonumber(b,16)]=true
+    end
+end
+
+fh:close()
+local b, v
+for i=0,0x10ffff do
+    if t[i] then
+       if not v then 
+           b, v = i, true
+       else
+           e=i
+       end
+    else
+       if v then print(string.format('    if (0x%04X<=i)and(i<0x%04X) then return true end', b,i)); v = false; end
+    end
+end
+print('  end')
+