X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/4691e240b46160d0b8470f6d51f9f4740f5ac75a..refs/heads/master:/in_sfnt.c

diff --git a/in_sfnt.c b/in_sfnt.c
index 0d51ff9..218bcbf 100644
--- a/in_sfnt.c
+++ b/in_sfnt.c
@@ -132,6 +132,7 @@ sfnt_decode offsubdir_decode[] = {
 #define TAG_hmtx	0x686d7478
 #define TAG_kern	0x6b65726e
 #define TAG_loca	0x6c6f6361
+#define TAG_maxp	0x6d617870
 #define TAG_name	0x6e616d65
 #define TAG_post	0x706f7374
 #define sfnt_true	0x74727565
@@ -364,6 +365,18 @@ sfnt_decode kern_f0_pair_decode[] = {
     { d_end }
 };
 
+/* Maximum profile ('maxp') table */
+typedef struct t_maxp_Tag t_maxp;
+struct t_maxp_Tag {
+    unsigned version;
+    unsigned numGlyphs;
+};
+sfnt_decode t_maxp_decode[] = {
+    { d_uint32, offsetof(t_maxp, version) },
+    { d_uint16, offsetof(t_maxp, numGlyphs) },
+    { d_end }
+};
+
 /* Naming ('name') table  */
 typedef struct t_name_Tag t_name;
 typedef struct namerecord_Tag namerecord;
@@ -462,7 +475,7 @@ static char *sfnt_psname(font_info *fi) {
     namerecord *nr;
 
     if (!sfnt_findtable(sf, TAG_name, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "name");
+	err_sfntnotable(&sf->pos, "name");
 	return NULL;
     }
     ptr = decode(t_name_decode, ptr, end, &name);
@@ -483,7 +496,7 @@ static char *sfnt_psname(font_info *fi) {
 	    }
 	}
     }
-    error(err_sfntnopsname, &sf->pos);
+    err_sfntnopsname(&sf->pos);
     return NULL;
 }
 
@@ -493,6 +506,9 @@ static int glyphsbyname_cmp(void const *a, void const *b) {
     glyph gb = cmp_glyphsbyindex[*(unsigned short *)b];
     if (ga < gb) return -1;
     if (ga > gb) return 1;
+    /* For de-duping, we'd prefer to have the first glyph stay first */
+    if (*(unsigned short *)a < *(unsigned short *)b) return -1;
+    if (*(unsigned short *)a > *(unsigned short *)b) return 1;
     return 0;
 }
 static int glyphsbyname_cmp_search(void const *a, void const *b) {
@@ -503,6 +519,14 @@ static int glyphsbyname_cmp_search(void const *a, void const *b) {
     return 0;
 }
 
+/* Generate an name for a glyph that doesn't have one. */
+static glyph genglyph(unsigned idx) {
+    char buf[11];
+    if (idx == 0) return glyph_intern(".notdef");
+    sprintf(buf, "glyph%u", idx);
+    return glyph_intern(buf);
+}
+
 /*
  * Extract data from the 'post' table (mostly glyph mappings)
  *
@@ -515,63 +539,79 @@ static void sfnt_mapglyphs(font_info *fi) {
     void *ptr, *end;
     unsigned char *sptr;
     char tmp[256];
-    glyph *extraglyphs;
-    unsigned nextras, i, g;
+    glyph *extraglyphs, prev, this;
+    unsigned nextras, i, g, suflen;
 
     sf->glyphsbyname = sf->glyphsbyindex = NULL;
-    if (!sfnt_findtable(sf, TAG_post, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "post");
-	return;
-    }
-    ptr = decode(t_post_decode, ptr, end, &post);
-    if (ptr == NULL) {
-	error(err_sfntbadtable, &sf->pos, "post");
-	return;
-    }
+    if (sfnt_findtable(sf, TAG_post, &ptr, &end)) {
+	ptr = decode(t_post_decode, ptr, end, &post);
+	if (ptr == NULL) {
+	    err_sfntbadtable(&sf->pos, "post");
+	    goto noglyphs;
+	}
     
-    sf->minmem = post.minMemType42;
-    sf->maxmem = post.maxMemType42;
-    fi->italicangle = post.italicAngle / 65536.0;
-    switch (post.format) {
-      case 0x00010000:
-	sf->nglyphs = 258;
-	sf->glyphsbyindex = (glyph *)tt_std_glyphs;
-	break;
-      case 0x00020000:
-	if ((char *)ptr + 2 > (char *)end) return;
-	decode_uint16(ptr, &sf->nglyphs);
-	ptr = (char *)ptr + 2;
-	if ((char *)ptr + 2*sf->nglyphs > (char *)end) return;
-	nextras = 0;
-	for (sptr = (unsigned char *)ptr + 2*sf->nglyphs;
-	     sptr < (unsigned char *)end;
-	     sptr += *sptr+1)
-	    nextras++;
-	extraglyphs = snewn(nextras, glyph);
-	i = 0;
-	for (sptr = (unsigned char *)ptr + 2*sf->nglyphs;
-	     sptr < (unsigned char *)end;
-	     sptr += *sptr+1) {
-	    memcpy(tmp, sptr + 1, *sptr);
-	    tmp[*sptr] = 0;
-	    assert(i < nextras);
-	    extraglyphs[i++] = glyph_intern(tmp);
+	sf->minmem = post.minMemType42;
+	sf->maxmem = post.maxMemType42;
+	fi->italicangle = post.italicAngle / 65536.0;
+	switch (post.format) {
+	  case 0x00010000:
+	    if (sf->nglyphs != 258) {
+		err_sfntbadtable(&sf->pos, "post");
+		break;
+	    }
+	    sf->glyphsbyindex = (glyph *)tt_std_glyphs;
+	    break;
+	  case 0x00020000:
+	    if ((char *)ptr + 2 > (char *)end) {
+		err_sfntbadtable(&sf->pos, "post");
+		break;
+	    }
+	    ptr = (char *)ptr + 2;
+	    if ((char *)ptr + 2*sf->nglyphs > (char *)end) {
+		err_sfntbadtable(&sf->pos, "post");
+		break;
+	    }
+	    nextras = 0;
+	    for (sptr = (unsigned char *)ptr + 2*sf->nglyphs;
+		 sptr < (unsigned char *)end;
+		 sptr += *sptr+1)
+		nextras++;
+	    extraglyphs = snewn(nextras, glyph);
+	    i = 0;
+	    for (sptr = (unsigned char *)ptr + 2*sf->nglyphs;
+		 sptr < (unsigned char *)end;
+		 sptr += *sptr+1) {
+		memcpy(tmp, sptr + 1, *sptr);
+		tmp[*sptr] = 0;
+		assert(i < nextras);
+		extraglyphs[i++] = glyph_intern(tmp);
+	    }
+	    sf->glyphsbyindex = snewn(sf->nglyphs, glyph);
+	    for (i = 0; i < sf->nglyphs; i++) {
+		decode_uint16((char *)ptr + 2*i, &g);
+		if (g <= 257)
+		    sf->glyphsbyindex[i] = tt_std_glyphs[g];
+		else if (g < 258 + nextras)
+		    sf->glyphsbyindex[i] = extraglyphs[g - 258];
+		else {
+		    err_sfntbadtable(&sf->pos, "post");
+		    sf->glyphsbyindex[i] = genglyph(i);
+		}
+	    }
+	    sfree(extraglyphs);
+	    break;
+	  case 0x00030000:
+	    break;
+	  default:
+	    err_sfnttablevers(&sf->pos, "post");
+	    break;
 	}
+    }
+  noglyphs:
+    if (!sf->glyphsbyindex) {
 	sf->glyphsbyindex = snewn(sf->nglyphs, glyph);
-	for (i = 0; i < sf->nglyphs; i++) {
-	    decode_uint16((char *)ptr + 2*i, &g);
-	    if (g <= 257)
-		sf->glyphsbyindex[i] = tt_std_glyphs[g];
-	    else if (g < 258 + nextras)
-		sf->glyphsbyindex[i] = extraglyphs[g - 258];
-	    else
-		sf->glyphsbyindex[i] = NOGLYPH;
-	}
-	sfree(extraglyphs);
-	break;
-      default:
-	error(err_sfnttablevers, &sf->pos, "post");
-	return;
+	for (i = 0; i < sf->nglyphs; i++)
+	    sf->glyphsbyindex[i] = genglyph(i);
     }
     /* Construct glyphsbyname */
     sf->glyphsbyname = snewn(sf->nglyphs, unsigned short);
@@ -580,6 +620,42 @@ static void sfnt_mapglyphs(font_info *fi) {
     cmp_glyphsbyindex = sf->glyphsbyindex;
     qsort(sf->glyphsbyname, sf->nglyphs, sizeof(*sf->glyphsbyname),
 	  glyphsbyname_cmp);
+    /*
+     * It's possible for fonts to specify the same name for multiple
+     * glyphs, which would make one of them inaccessible.  Check for
+     * that, and rename all but one of each set.
+     *
+     * To ensure that we don't clash with any existing glyph names,
+     * our renaming involves appending the glyph number formatted with
+     * enough leading zeroes to make it longer than any all-digit
+     * suffix that already exists in the font.
+     */
+    suflen = 4;
+    for (i = 0; i < sf->nglyphs; i++) {
+	char const *p;
+	p = strrchr(glyph_extern(sfnt_indextoglyph(sf, i)), '.');
+	if (p && !(p+1)[strspn(p+1, "0123456789")] && strlen(p+1) > suflen)
+	    suflen = strlen(p+1);
+    }
+    suflen++;
+    prev = sfnt_indextoglyph(sf, sf->glyphsbyname[0]);
+    for (i = 1; i < sf->nglyphs; i++) {
+	if (prev == (this = sfnt_indextoglyph(sf, sf->glyphsbyname[i]))) {
+	    char const *basename;
+	    char *buf;
+	    basename = glyph_extern(this);
+	    buf = snewn(strlen(basename) + 2 + suflen, char);
+	    strcpy(buf, basename);
+	    sprintf(buf + strlen(basename), ".%0*hu", suflen,
+		    sf->glyphsbyname[i]);
+	    sf->glyphsbyindex[sf->glyphsbyname[i]] = glyph_intern(buf);
+	    sfree(buf);
+	}
+	prev = this;
+    }
+    /* We may have renamed some glyphs, so re-sort the array. */
+    qsort(sf->glyphsbyname, sf->nglyphs, sizeof(*sf->glyphsbyname),
+	  glyphsbyname_cmp);
 }
 
 glyph sfnt_indextoglyph(sfnt *sf, unsigned idx) {
@@ -614,31 +690,31 @@ void sfnt_getmetrics(font_info *fi) {
     fi->fontbbox[2] = sf->head.xMax * FUNITS_PER_PT /  sf->head.unitsPerEm;
     fi->fontbbox[3] = sf->head.yMax * FUNITS_PER_PT /  sf->head.unitsPerEm;
     if (!sfnt_findtable(sf, TAG_hhea, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "hhea");
+	err_sfntnotable(&sf->pos, "hhea");
 	return;
     }
     if (decode(t_hhea_decode, ptr, end, &hhea) == NULL) {
-	error(err_sfntbadtable, &sf->pos, "hhea");
+	err_sfntbadtable(&sf->pos, "hhea");
 	return;
     }
     if ((hhea.version & 0xffff0000) != 0x00010000) {
-	error(err_sfnttablevers, &sf->pos, "hhea");
+	err_sfnttablevers(&sf->pos, "hhea");
 	return;
     }
     fi->ascent = hhea.ascent;
     fi->descent = hhea.descent;
     if (hhea.metricDataFormat != 0) {
-	error(err_sfnttablevers, &sf->pos, "hmtx");
+	err_sfnttablevers(&sf->pos, "hmtx");
 	return;
     }
     if (!sfnt_findtable(sf, TAG_hmtx, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "hmtx");
+	err_sfntnotable(&sf->pos, "hmtx");
 	return;
     }
     hmtx = snewn(hhea.numOfLongHorMetrics, unsigned);
     if (decoden(longhormetric_decode, ptr, end, hmtx, sizeof(*hmtx),
 		hhea.numOfLongHorMetrics) == NULL) {
-	error(err_sfntbadtable, &sf->pos, "hmtx");
+	err_sfntbadtable(&sf->pos, "hmtx");
 	return;
     }
     for (i = 0; i < sf->nglyphs; i++) {
@@ -667,7 +743,7 @@ void sfnt_getmetrics(font_info *fi) {
     fi->descent = OS_2.sTypoDescender * FUNITS_PER_PT / sf->head.unitsPerEm;
     return;
   bados2:
-    error(err_sfntbadtable, &sf->pos, "OS/2");
+    err_sfntbadtable(&sf->pos, "OS/2");
 }
 
 /*
@@ -737,16 +813,17 @@ static void sfnt_getkern(font_info *fi) {
     }
     return;
   bad:
-    error(err_sfntbadtable, &sf->pos, "kern");
+    err_sfntbadtable(&sf->pos, "kern");
     return;
 }
 
 /*
  * Get mapping data from 'cmap' table
  *
- * We look for either a (0, 3), or (3, 1) table, both of these being
- * versions of UCS-2.  We only handle format 4 of this table, since
- * that seems to be the only one in use.
+ * We look for either a (0, 0), (0, 2), (0, 3), or (3, 1) table, all
+ * of these being versions of UCS-2.  We ignore (0, 1), since it's
+ * Unicode 1.1 with precomposed Hangul syllables.  We only handle
+ * format 4 of this table, since that seems to be the only one in use.
  */
 void sfnt_getmap(font_info *fi) {
     sfnt *sf = fi->fontfile;
@@ -760,7 +837,7 @@ void sfnt_getmap(font_info *fi) {
     for (i = 0; i < lenof(fi->bmp); i++)
 	    fi->bmp[i] = 0xFFFF;
     if (!sfnt_findtable(sf, TAG_cmap, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "cmap");
+	err_sfntnotable(&sf->pos, "cmap");
     }
     base = ptr;
     ptr = decode(t_cmap_decode, ptr, end, &cmap);
@@ -772,7 +849,9 @@ void sfnt_getmap(font_info *fi) {
     for (i = 0; i < cmap.numTables; i++) {
 	if (!decode(uint16_decode, (char *)base + esd[i].offset, end, &format))
 	    goto bad;
-	if ((esd[i].platformID == 0 && esd[i].encodingID == 3) ||
+	if ((esd[i].platformID == 0 && esd[i].encodingID == 0) ||
+	    (esd[i].platformID == 0 && esd[i].encodingID == 2) ||
+	    (esd[i].platformID == 0 && esd[i].encodingID == 3) ||
 	    (esd[i].platformID == 3 && esd[i].encodingID == 1)) {
 	    /* UCS-2 encoding */
 	    if (!decode(uint16_decode, (char *)base + esd[i].offset, end,
@@ -806,17 +885,28 @@ void sfnt_getmap(font_info *fi) {
 			for (k = startCode[j]; k <= endCode[j]; k++) {
 			    idx = (k + idDelta[j]) & 0xffff;
 			    if (idx != 0) {
-				if (idx > sf->nglyphs) goto bad;
+				if (idx > sf->nglyphs)  {
+				    err_sfntbadglyph(&sf->pos, k);
+				    continue;
+				}
 				fi->bmp[k] = sfnt_indextoglyph(sf, idx);
 			    }
 			}
 		    } else {
 			unsigned startidx = idRangeOffset[j]/2 - segcount + j;
 			for (k = startCode[j]; k <= endCode[j]; k++) {
+			    if (startidx + k - startCode[j] >=
+				nglyphindex) {
+				err_sfntbadglyph(&sf->pos, k);
+				continue;
+			    }
 			    idx = glyphIndexArray[startidx + k - startCode[j]];
 			    if (idx != 0) {
 				idx = (idx + idDelta[j]) & 0xffff;
-				if (idx > sf->nglyphs) goto bad;
+				if (idx > sf->nglyphs) {
+				    err_sfntbadglyph(&sf->pos, k);
+				    continue;
+				}
 				fi->bmp[k] = sfnt_indextoglyph(sf, idx);
 			    }
 			}
@@ -827,10 +917,10 @@ void sfnt_getmap(font_info *fi) {
 	    }
 	}
     }
-    error(err_sfntnounicmap, &sf->pos);
+    err_sfntnounicmap(&sf->pos);
     return;
   bad:
-    error(err_sfntbadtable, &sf->pos, "cmap");
+    err_sfntbadtable(&sf->pos, "cmap");
 }
 
 void read_sfnt_file(input *in) {
@@ -839,6 +929,7 @@ void read_sfnt_file(input *in) {
     FILE *fp = in->currfp;
     font_info *fi = snew(font_info);
     void *ptr, *end;
+    t_maxp maxp;
 
     fi->name = NULL;
     fi->widths = newtree234(width_cmp);
@@ -868,28 +959,41 @@ void read_sfnt_file(input *in) {
     sf->nglyphs = 0;
     ptr = decode(offsubdir_decode, sf->data, sf->end, &sf->osd);
     if (ptr == NULL) {
-	error(err_sfntbadhdr, &sf->pos);
+	err_sfntbadhdr(&sf->pos);
 	return;
     }
     sf->td = snewn(sf->osd.numTables, tabledir);
     ptr = decoden(tabledir_decode, ptr, sf->end, sf->td, sizeof(*sf->td),
 		  sf->osd.numTables);
     if (ptr == NULL) {
-	error(err_sfntbadhdr, &sf->pos);
+	err_sfntbadhdr(&sf->pos);
 	return;
     }	
     if (!sfnt_findtable(sf, TAG_head, &ptr, &end)) {
-	error(err_sfntnotable, &sf->pos, "head");
+	err_sfntnotable(&sf->pos, "head");
 	return;
     }
     if (decode(t_head_decode, ptr, end, &sf->head) == NULL) {
-	error(err_sfntbadtable, &sf->pos, "head");
+	err_sfntbadtable(&sf->pos, "head");
 	return;
     }
     if ((sf->head.version & 0xffff0000) != 0x00010000) {
-	error(err_sfnttablevers, &sf->pos, "head");
+	err_sfnttablevers(&sf->pos, "head");
+	return;
+    }
+    if (!sfnt_findtable(sf, TAG_maxp, &ptr, &end)) {
+	err_sfntnotable(&sf->pos, "maxp");
+	return;
+    }
+    if (decode(t_maxp_decode, ptr, end, &maxp) == NULL) {
+	err_sfntbadtable(&sf->pos, "maxp");
+	return;
+    }
+    if (maxp.version < 0x00005000 || maxp.version > 0x0001ffff) {
+	err_sfnttablevers(&sf->pos, "maxp");
 	return;
     }
+    sf->nglyphs = maxp.numGlyphs;
     fi->name = sfnt_psname(fi);
     if (fi->name == NULL) return;
     sfnt_mapglyphs(fi);
@@ -957,13 +1061,13 @@ void sfnt_writeps(font_info const *fi, FILE *ofp) {
 	breaks[i] = sf->td[i].offset;
     }
     if (!sfnt_findtable(sf, TAG_glyf, &glyfptr, &glyfend)) {
-	error(err_sfntnotable, &sf->pos, "glyf");
+	err_sfntnotable(&sf->pos, "glyf");
 	return;
     }
     glyfoff = (char *)glyfptr - (char *)sf->data;
     glyflen = (char *)glyfend - (char *)glyfptr;
     if (!sfnt_findtable(sf, TAG_loca, &locaptr, &locaend)) {
-	error(err_sfntnotable, &sf->pos, "loca");
+	err_sfntnotable(&sf->pos, "loca");
 	return;
     }
     loca = snewn(sf->nglyphs, unsigned);
@@ -996,7 +1100,7 @@ void sfnt_writeps(font_info const *fi, FILE *ofp) {
     fprintf(ofp, "end /%s exch definefont\n", fi->name);
     return;
   badloca:
-    error(err_sfntbadtable, &sf->pos, "loca");
+    err_sfntbadtable(&sf->pos, "loca");
 }
 
 void sfnt_data(font_info *fi, char **bufp, size_t *lenp) {