dvipdfm-x: Simplify ToUnicode handling

Remove used_glyphs and its supporting code for ToUnicode handling,
because for CID fonts, glyph id can be looked up from cid which is
stored in used_chars already, it simplified a lot of code and reduced
memory usage.

Continue dedup create_ToUnicode_cmap{4,12}() with
add_to_cmap_if_used().

git-svn-id: svn://tug.org/texlive/trunk@34885 c570f23f-e606-0410-a88d-b1316a301751

7 files changed, 104 insertions, 177 deletions

diff --git a/Build/source/texk/dvipdfm-x/pdfdev.c b/Build/source/texk/dvipdfm-x/pdfdev.c
index 2819a3e79af..ec7605a6ab4 100644
--- a/Build/source/texk/dvipdfm-x/pdfdev.c
+++ b/Build/source/texk/dvipdfm-x/pdfdev.c
@@ -463,7 +463,6 @@ struct dev_font {
 
   pdf_obj *resource;
   char    *used_chars;
-  char    *used_glyphs;
 
   /* Font format:
    * simple, composite or bitmap.
@@ -877,8 +876,6 @@ dev_set_font (int font_id)
   if (!real_font->resource) {
     real_font->resource   = pdf_get_font_reference(real_font->font_id);
     real_font->used_chars = pdf_get_font_usedchars(real_font->font_id);
-    if (real_font->cff_charsets)
-      real_font->used_glyphs = pdf_get_font_usedglyphs(real_font->font_id);
   }
 
   if (!real_font->used_on_this_page) { 
@@ -1158,12 +1155,6 @@ pdf_dev_set_string (spt_t xpos, spt_t ypos,
   length  = instr_len;
 
   if (font->format == PDF_FONTTYPE_COMPOSITE) {
-    if (real_font->used_glyphs != NULL && ctype == -1) {
-      for (i = 0; i < length; i += 2) {
-        unsigned short gid = (str_ptr[i] << 8) | str_ptr[i + 1];
-        add_to_used_chars2(real_font->used_glyphs, gid);
-      }
-    }
     if (handle_multibyte_string(font, &str_ptr, &length, ctype) < 0) {
       ERROR("Error in converting input string...");
       return;
@@ -1171,10 +1162,6 @@ pdf_dev_set_string (spt_t xpos, spt_t ypos,
     if (real_font->used_chars != NULL) {
       for (i = 0; i < length; i += 2) {
         unsigned short cid = (str_ptr[i] << 8) | str_ptr[i + 1];
-        if (ctype == 2 && font->cff_charsets) {
-          unsigned short gid = cff_charsets_lookup_gid(font->cff_charsets, cid);
-          add_to_used_chars2(real_font->used_glyphs, gid);
-        }
         add_to_used_chars2(real_font->used_chars, cid);
       }
     }
@@ -1549,7 +1536,6 @@ pdf_dev_locate_font (const char *font_name, spt_t ptsize)
 
   font->resource   = NULL; /* Don't ref obj until font is actually used. */  
   font->used_chars = NULL;
-  font->used_glyphs = NULL;
 
   font->extend     = 1.0;
   font->slant      = 0.0;
diff --git a/Build/source/texk/dvipdfm-x/pdffont.c b/Build/source/texk/dvipdfm-x/pdffont.c
index 56a354a15db..9dc400700f3 100644
--- a/Build/source/texk/dvipdfm-x/pdffont.c
+++ b/Build/source/texk/dvipdfm-x/pdffont.c
@@ -334,24 +334,6 @@ pdf_get_font_usedchars (int font_id)
   }
 }
 
-char *
-pdf_get_font_usedglyphs (int font_id)
-{
-  pdf_font *font;
-
-  CHECK_ID(font_id);
-
-  font = GET_FONT(font_id);
-  if (font->subtype == PDF_FONT_FONTTYPE_TYPE0) {
-    Type0Font *t0font;
-
-    t0font = Type0Font_cache_get(font->font_id);
-    return Type0Font_get_usedglyphs(t0font);
-  }
-
-  return NULL;
-}
-
 int
 pdf_get_font_wmode (int font_id)
 {
diff --git a/Build/source/texk/dvipdfm-x/pdffont.h b/Build/source/texk/dvipdfm-x/pdffont.h
index 38ca3a590e2..a833fe89dc1 100644
--- a/Build/source/texk/dvipdfm-x/pdffont.h
+++ b/Build/source/texk/dvipdfm-x/pdffont.h
@@ -63,7 +63,6 @@ extern int      pdf_font_findresource  (const char *font_name,
 extern int      pdf_get_font_subtype   (int font_id);
 extern pdf_obj *pdf_get_font_reference (int font_id);
 extern char    *pdf_get_font_usedchars (int font_id);
-extern char    *pdf_get_font_usedglyphs (int font_id);
 
 #if 0
 extern char    *pdf_get_font_fontname  (int font_id); /* without unique tag */
diff --git a/Build/source/texk/dvipdfm-x/tt_cmap.c b/Build/source/texk/dvipdfm-x/tt_cmap.c
index f288891a195..d0e8af6db68 100644
--- a/Build/source/texk/dvipdfm-x/tt_cmap.c
+++ b/Build/source/texk/dvipdfm-x/tt_cmap.c
@@ -862,11 +862,13 @@ sfnt_get_glyphname(struct tt_post_table *post, cff_font *cffont, USHORT gid)
 
 static USHORT
 handle_subst_glyphs (CMap *cmap,
-		     CMap *cmap_add, const char *used_glyphs,
-		     sfnt *sfont, cff_font *cffont)
+                     CMap *cmap_add,
+                     const char *used_glyphs,
+                     sfnt *sfont,
+                     cff_font *cffont)
 {
   USHORT count;
-  USHORT i, gid;
+  USHORT i;
   struct tt_post_table *post = NULL;
 
   if (!cmap_add)
@@ -882,10 +884,10 @@ handle_subst_glyphs (CMap *cmap,
       continue;
 
     for (j = 0; j < 8; j++) {
-      gid = 8 * i + j;
+      USHORT gid = 8 * i + j;
 
       if (!is_used_char2(used_glyphs, gid))
-	continue;
+        continue;
 
       if (!cmap_add) {
 #define MAX_UNICODES	16
@@ -918,35 +920,33 @@ handle_subst_glyphs (CMap *cmap,
         }
         RELEASE(name);
       } else {
-	wbuf[0] = (gid >> 8) & 0xff;
-	wbuf[1] =  gid & 0xff;
-	inbuf        = wbuf;
-	inbytesleft  = 2;
-	outbuf       = wbuf + 2;
-	outbytesleft = WBUF_SIZE - 2;
-	CMap_decode(cmap_add,
-		    &inbuf , &inbytesleft,
-		    &outbuf, &outbytesleft);
-	if (inbytesleft != 0) {
-	  WARN("CMap conversion failed...");
-	} else {
-	  len = WBUF_SIZE - 2 - outbytesleft;
-	  CMap_add_bfchar(cmap, wbuf, 2, wbuf + 2, len);
-	  count++;
-
-	  if (verbose > VERBOSE_LEVEL_MIN) {
-	    long _i;
-
-	    MESG("otf_cmap>> Additional ToUnicode mapping: <%04X> <", gid);
-	    for (_i = 0; _i < len; _i++) {
-	      MESG("%02X", wbuf[2 + _i]);
-	    }
-	    MESG(">\n");
-	  }
+        wbuf[0] = (gid >> 8) & 0xff;
+        wbuf[1] =  gid & 0xff;
 
-	}
-      }
+        inbuf        = wbuf;
+        inbytesleft  = 2;
+        outbuf       = wbuf + 2;
+        outbytesleft = WBUF_SIZE - 2;
+        CMap_decode(cmap_add, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+
+        if (inbytesleft != 0) {
+          WARN("CMap conversion failed...");
+        } else {
+          len = WBUF_SIZE - 2 - outbytesleft;
+          CMap_add_bfchar(cmap, wbuf, 2, wbuf + 2, len);
+          count++;
+
+          if (verbose > VERBOSE_LEVEL_MIN) {
+            long _i;
 
+            MESG("otf_cmap>> Additional ToUnicode mapping: <%04X> <", gid);
+            for (_i = 0; _i < len; _i++) {
+              MESG("%02X", wbuf[2 + _i]);
+            }
+            MESG(">\n");
+          }
+        }
+      }
     }
   }
 
@@ -977,56 +977,69 @@ prepare_CIDFont_from_sfnt(sfnt* sfont)
 }
 
 static USHORT
+add_to_cmap_if_used (CMap *cmap,
+                     cff_font *cffont,
+                     char *used_chars,
+                     USHORT gid,
+                     ULONG ch)
+{
+  USHORT count = 0;
+  USHORT cid = cffont ? cff_charsets_lookup_inverse(cffont, gid) : gid;
+  if (is_used_char2(used_chars, cid)) {
+    int len;
+    unsigned char *p = wbuf + 2;
+
+    count++;
+
+    wbuf[0] = (cid >> 8) & 0xff;
+    wbuf[1] = (cid & 0xff);
+    len = UC_sput_UTF16BE((long) ch, &p, wbuf + WBUF_SIZE);
+    CMap_add_bfchar(cmap, wbuf, 2, wbuf + 2, len);
+
+    /* Skip PUA characters and alphabetic presentation forms, allowing
+     * handle_subst_glyphs() as it might find better mapping. Fixes the
+     * mapping of ligatures encoded in PUA in fonts like Linux Libertine
+     * and old Adobe fonts.
+     */
+    if (!is_PUA_or_presentation(ch)) {
+      /* Avoid duplicate entry
+       * There are problem when two Unicode code is mapped to
+       * single glyph...
+       */
+      used_chars[gid / 8] &= ~(1 << (7 - (gid % 8)));
+    }
+  }
+
+  return count;
+}
+
+static USHORT
 create_ToUnicode_cmap4 (CMap *cmap,
                         struct cmap4 *map,
-                        char *used_glyphs,
+                        char *used_chars,
                         cff_font *cffont)
 {
-  USHORT    c0, c1, gid, count, ch;
-  USHORT    i, j, d, segCount;
+  USHORT count = 0, segCount = map->segCountX2 / 2;
+  USHORT i, j;
 
-  segCount = map->segCountX2 / 2;
-  for (count = 0, i = 0; i < segCount; i++) {
-    c0 = map->startCount[i];
-    c1 = map->endCount[i];
-    d  = map->idRangeOffset[i] / 2 - (segCount - i);
+  for (i = 0; i < segCount; i++) {
+    USHORT c0 = map->startCount[i];
+    USHORT c1 = map->endCount[i];
+    USHORT d  = map->idRangeOffset[i] / 2 - (segCount - i);
     for (j = 0; j <= c1 - c0; j++) {
-      ch = c0 + j;
+      USHORT ch = c0 + j;
+      USHORT gid;
+
       if (map->idRangeOffset[i] == 0) {
-	gid = (ch + map->idDelta[i]) & 0xffff;
+        gid = (ch + map->idDelta[i]) & 0xffff;
       } else if (c0 == 0xffff && c1 == 0xffff && map->idRangeOffset[i] == 0xffff) {
-	/* this is for protection against some old broken fonts... */
-	gid = 0;
+        /* this is for protection against some old broken fonts... */
+        gid = 0;
       } else {
-	gid = (map->glyphIndexArray[j+d] +
-	       map->idDelta[i]) & 0xffff;
+        gid = (map->glyphIndexArray[j + d] + map->idDelta[i]) & 0xffff;
       }
-      if (is_used_char2(used_glyphs, gid)) {
-        unsigned int cid = cffont ? cff_charsets_lookup_inverse(cffont, gid) : gid;
-	count++;
-
-	wbuf[0] = (cid >> 8) & 0xff;
-	wbuf[1] = (cid & 0xff);
-
-	wbuf[2] = (ch >> 8) & 0xff;
-	wbuf[3] =  ch & 0xff;
 
-	CMap_add_bfchar(cmap, wbuf, 2, wbuf+2, 2);
-
-        /* Skip PUA characters and alphabetic presentation forms, allowing
-         * handle_subst_glyphs() as it might find better mapping. Fixes the
-         * mapping of ligatures encoded in PUA in fonts like Linux Libertine
-         * and old Adobe fonts.
-         */
-        if (!is_PUA_or_presentation(ch)) {
-          /* Avoid duplicate entry
-           * There are problem when two Unicode code is mapped to
-           * single glyph...
-           */
-          used_glyphs[gid/8] &= ~(1 << (7 - (gid % 8)));
-        }
-        count++;
-      }
+      count += add_to_cmap_if_used(cmap, cffont, used_chars, gid, ch);
     }
   }
 
@@ -1036,45 +1049,17 @@ create_ToUnicode_cmap4 (CMap *cmap,
 static USHORT
 create_ToUnicode_cmap12 (CMap *cmap,
                          struct cmap12 *map,
-                         char *used_glyphs,
+                         char *used_chars,
                          cff_font *cffont)
 {
-  ULONG  i, ch;
-  USHORT gid, count = 0;
+  ULONG i, ch, count = 0;
 
   for (i = 0; i < map->nGroups; i++) {
     for (ch  = map->groups[i].startCharCode;
          ch <= map->groups[i].endCharCode; ch++) {
-      unsigned char *p;
-      int      len;
-      long     d;
-
-      p   = wbuf + 2;
-      d   = ch - map->groups[i].startCharCode;
-      gid = (USHORT) ((map->groups[i].startGlyphID + d) & 0xffff);
-      if (is_used_char2(used_glyphs, gid)) {
-        unsigned int cid = cffont ? cff_charsets_lookup_inverse(cffont, gid) : gid;
-        count++;
-        wbuf[0] = (cid >> 8) & 0xff;
-        wbuf[1] = (cid & 0xff);
-        len = UC_sput_UTF16BE((long)ch, &p, wbuf+WBUF_SIZE);
-
-        CMap_add_bfchar(cmap, wbuf, 2, wbuf+2, len);
-
-        /* Skip PUA characters and alphabetic presentation forms, allowing
-         * handle_subst_glyphs() as it might find better mapping. Fixes the
-         * mapping of ligatures encoded in PUA in fonts like Linux Libertine
-         * and old Adobe fonts.
-         */
-        if (!is_PUA_or_presentation(ch)) {
-          /* Avoid duplicate entry
-           * There are problem when two Unicode code is mapped to
-           * single glyph...
-           */
-          used_glyphs[gid/8] &= ~(1 << (7 - (gid % 8)));
-        }
-        count++;
-      }
+      long d = ch - map->groups[i].startCharCode;
+      USHORT gid = (USHORT) ((map->groups[i].startGlyphID + d) & 0xffff);
+      count += add_to_cmap_if_used(cmap, cffont, used_chars, gid, ch);
     }
   }
 
@@ -1085,14 +1070,13 @@ static pdf_obj *
 create_ToUnicode_cmap (tt_cmap *ttcmap,
                        const char *cmap_name,
                        CMap *cmap_add,
-                       const char *used_glyphs,
+                       const char *used_chars,
                        sfnt *sfont,
                        CMap *code_to_cid_cmap)
 {
   pdf_obj  *stream = NULL;
   CMap     *cmap;
   USHORT    i, gid, count = 0;
-  char      used_glyphs_copy[8192];
   cff_font *cffont = prepare_CIDFont_from_sfnt(sfont);
   char      is_cidfont = cffont && (cffont->flag & FONTTYPE_CIDFONT);
 
@@ -1105,23 +1089,21 @@ create_ToUnicode_cmap (tt_cmap *ttcmap,
 
   if (code_to_cid_cmap && cffont && is_cidfont) {
     for (i = 0; i < 8192; i++) {
-      int   j;
-      long  len;
+      int j;
 
-      if (used_glyphs[i] == 0)
+      if (used_chars[i] == 0)
         continue;
 
       for (j = 0; j < 8; j++) {
-        unsigned int cid;
+        USHORT cid = 8 * i + j;
         int ch;
-        gid = 8 * i + j;
 
-        if (!is_used_char2(used_glyphs, gid))
+        if (!is_used_char2(used_chars, cid))
           continue;
 
-        cid = cff_charsets_lookup_inverse(cffont, gid);
         ch = CMap_reverse_decode(code_to_cid_cmap, cid);
         if (ch >= 0) {
+          long len;
           unsigned char *p = wbuf + 2;
           wbuf[0] = (cid >> 8) & 0xff;
           wbuf[1] =  cid & 0xff;
@@ -1132,24 +1114,25 @@ create_ToUnicode_cmap (tt_cmap *ttcmap,
       }
     }
   } else {
-    memcpy(used_glyphs_copy, used_glyphs, 8192);
+    char used_chars_copy[8192];
+    memcpy(used_chars_copy, used_chars, 8192);
 
     /* For create_ToUnicode_cmap{4,12}(), cffont is for GID -> CID lookup,
      * so it is only needed for CID fonts. */
     switch (ttcmap->format) {
       case 4:
-        count = create_ToUnicode_cmap4(cmap, ttcmap->map, used_glyphs_copy,
+        count = create_ToUnicode_cmap4(cmap, ttcmap->map, used_chars_copy,
                                        is_cidfont ? cffont : NULL);
         break;
       case 12:
-        count = create_ToUnicode_cmap12(cmap, ttcmap->map, used_glyphs_copy,
+        count = create_ToUnicode_cmap12(cmap, ttcmap->map, used_chars_copy,
                                         is_cidfont ? cffont : NULL);
         break;
     }
 
     /* For handle_subst_glyphs(), cffont is for GID -> glyph name lookup, so
      * it is only needed for non-CID fonts. */
-    count += handle_subst_glyphs(cmap, cmap_add, used_glyphs_copy, sfont,
+    count += handle_subst_glyphs(cmap, cmap_add, used_chars_copy, sfont,
                                  is_cidfont ? NULL : cffont);
   }
 
@@ -1182,7 +1165,7 @@ static cmap_plat_enc_rec cmap_plat_encs[] = {
 pdf_obj *
 otf_create_ToUnicode_stream (const char *font_name,
                              int ttc_index, /* 0 for non-TTC */
-                             const char *used_glyphs,
+                             const char *used_chars,
                              int cmap_id)
 {
   pdf_obj    *cmap_ref = NULL;
@@ -1275,7 +1258,7 @@ otf_create_ToUnicode_stream (const char *font_name,
       continue;
 
     if (ttcmap->format == 4 || ttcmap->format == 12) {
-      cmap_obj = create_ToUnicode_cmap(ttcmap, cmap_name, cmap_add, used_glyphs,
+      cmap_obj = create_ToUnicode_cmap(ttcmap, cmap_name, cmap_add, used_chars,
                                        sfont, code_to_cid_cmap);
       break;
     }
diff --git a/Build/source/texk/dvipdfm-x/tt_cmap.h b/Build/source/texk/dvipdfm-x/tt_cmap.h
index afc492cb127..2e9bbbbbc17 100644
--- a/Build/source/texk/dvipdfm-x/tt_cmap.h
+++ b/Build/source/texk/dvipdfm-x/tt_cmap.h
@@ -68,7 +68,7 @@ extern void     tt_cmap_release (tt_cmap *cmap);
 /* Indirect reference */
 extern pdf_obj *otf_create_ToUnicode_stream (const char *map_name,
                                              int ttc_index,
-                                             const char *used_glyphs,
+                                             const char *used_chars,
                                              int cmap_id);
 /* CMap ID */
 extern int      otf_load_Unicode_CMap       (const char *map_name,
diff --git a/Build/source/texk/dvipdfm-x/type0.c b/Build/source/texk/dvipdfm-x/type0.c
index 060c9b9cfe8..5fe77a7f86d 100644
--- a/Build/source/texk/dvipdfm-x/type0.c
+++ b/Build/source/texk/dvipdfm-x/type0.c
@@ -85,7 +85,6 @@ struct Type0Font {
   char    *fontname;   /* BaseFont */
   char    *encoding;   /* "Identity-H" or "Identity-V" (not ID) */
   char    *used_chars; /* Used chars (CIDs) */
-  char    *used_glyphs; /* Used glyphs (GIDs) */
 
   /*
    * Type0 only
@@ -114,7 +113,6 @@ Type0Font_init_font_struct (Type0Font *font)
   font->descriptor = NULL;
   font->encoding   = NULL;
   font->used_chars = NULL;
-  font->used_glyphs = NULL;
   font->descendant = NULL;
   font->wmode      = -1;
   font->cmap_id    = -1;
@@ -135,8 +133,6 @@ Type0Font_clean (Type0Font *font)
       ERROR("%s: FontDescriptor unexpected for Type0 font.", TYPE0FONT_DEBUG_STR);
     if (!(font->flags & FLAG_USED_CHARS_SHARED) && font->used_chars)
       RELEASE(font->used_chars);
-    if (!(font->flags & FLAG_USED_CHARS_SHARED) && font->used_glyphs)
-      RELEASE(font->used_glyphs);
     if (font->encoding)
       RELEASE(font->encoding);
     if (font->fontname)
@@ -145,7 +141,6 @@ Type0Font_clean (Type0Font *font)
     font->indirect   = NULL;
     font->descriptor = NULL;
     font->used_chars = NULL;
-    font->used_glyphs = NULL;
     font->encoding   = NULL;
     font->fontname   = NULL;
   }
@@ -157,13 +152,9 @@ Type0Font_clean (Type0Font *font)
 static pdf_obj *
 Type0Font_create_ToUnicode_stream(Type0Font *font) {
   CIDFont *cidfont = font->descendant;
-  char *used = Type0Font_get_usedglyphs(font);
-  if (!used)
-    used = Type0Font_get_usedchars(font);
-
   return otf_create_ToUnicode_stream(CIDFont_get_ident(cidfont),
                                      CIDFont_get_opt_index(cidfont),
-                                     used,
+                                     Type0Font_get_usedchars(font),
                                      font->cmap_id);
 }
 
@@ -331,14 +322,6 @@ Type0Font_get_usedchars (Type0Font *font)
   return font->used_chars;
 }
 
-char *
-Type0Font_get_usedglyphs (Type0Font *font)
-{
-  ASSERT(font);
-
-  return font->used_glyphs;
-}
-
 pdf_obj *
 Type0Font_get_resource (Type0Font *font)
 {
@@ -517,7 +500,6 @@ Type0Font_cache_find (const char *map_name, int cmap_id, fontmap_opt *fmap_opt)
    */
 
   font->used_chars = NULL;
-  font->used_glyphs = NULL;
   font->flags      = FLAG_NONE;
 
   switch (CIDFont_get_subtype(cidfont)) {
@@ -531,12 +513,9 @@ Type0Font_cache_find (const char *map_name, int cmap_id, fontmap_opt *fmap_opt)
      */
     if ((parent_id = CIDFont_get_parent_id(cidfont, wmode ? 0 : 1)) < 0) {
       font->used_chars = new_used_chars2();
-      if (fmap_opt->cff_charsets)
-        font->used_glyphs = new_used_chars2();
     } else {
       /* Don't allocate new one. */
       font->used_chars = Type0Font_get_usedchars(Type0Font_cache_get(parent_id));
-      font->used_glyphs = Type0Font_get_usedglyphs(Type0Font_cache_get(parent_id));
       font->flags     |= FLAG_USED_CHARS_SHARED;
     }
     break;
diff --git a/Build/source/texk/dvipdfm-x/type0.h b/Build/source/texk/dvipdfm-x/type0.h
index c22d41e3da0..553de1e760f 100644
--- a/Build/source/texk/dvipdfm-x/type0.h
+++ b/Build/source/texk/dvipdfm-x/type0.h
@@ -36,8 +36,6 @@ extern char      *Type0Font_get_encoding  (Type0Font *font);
 #endif
 extern char      *Type0Font_get_usedchars (Type0Font *font);
 
-extern char      *Type0Font_get_usedglyphs (Type0Font *font);
-
 extern pdf_obj   *Type0Font_get_resource  (Type0Font *font);
 
 extern void       Type0Font_set_ToUnicode (Type0Font *font, pdf_obj *cmap_ref);