/*
* Create an Identity-* CMap (for both 1 and 2-byte encodings)
*/
fz_error
pdf_newidentitycmap(pdf_cmap **cmapp, int wmode, int bytes)
{
fz_error error;
pdf_cmap *cmap;
error = pdf_newcmap(&cmap);
if (error)
return fz_rethrow(error, "cannot create cmap");
sprintf(cmap->cmapname, "Identity-%c", wmode ? 'V' : 'H');
error = pdf_addcodespace(cmap, 0x0000, 0xffff, bytes);
if (error) {
pdf_dropcmap(cmap);
return fz_rethrow(error, "cannot add code space");
}
error = pdf_maprangetorange(cmap, 0x0000, 0xffff, 0);
if (error) {
pdf_dropcmap(cmap);
return fz_rethrow(error, "cannot map <0000> to <ffff>");
}
error = pdf_sortcmap(cmap);
if (error) {
pdf_dropcmap(cmap);
return fz_rethrow(error, "cannot sort cmap");
}
pdf_setwmode(cmap, wmode);
*cmapp = cmap;
return fz_okay;
}
static void ftdropfont(fz_font *font)
{
pdf_font *pfont = (pdf_font*)font;
if (pfont->encoding)
pdf_dropcmap(pfont->encoding);
if (pfont->tottfcmap)
pdf_dropcmap(pfont->tottfcmap);
if (pfont->tounicode)
pdf_dropcmap(pfont->tounicode);
fz_free(pfont->cidtogid);
fz_free(pfont->cidtoucs);
if (pfont->ftface)
FT_Done_Face((FT_Face)pfont->ftface);
if (pfont->fontdata)
fz_dropbuffer(pfont->fontdata);
}
static void
t3dropfont(fz_font *font)
{
int i;
pdf_font *pfont = (pdf_font*)font;
if (pfont->encoding)
pdf_dropcmap(pfont->encoding);
for (i = 0; i < 256; i++)
if (pfont->charprocs[i])
fz_droptree(pfont->charprocs[i]);
}
void
pdf_dropfont(pdf_fontdesc *fontdesc)
{
if (fontdesc && --fontdesc->refs == 0)
{
if (fontdesc->font)
fz_dropfont(fontdesc->font);
if (fontdesc->encoding)
pdf_dropcmap(fontdesc->encoding);
if (fontdesc->tottfcmap)
pdf_dropcmap(fontdesc->tottfcmap);
if (fontdesc->tounicode)
pdf_dropcmap(fontdesc->tounicode);
fz_free(fontdesc->cidtogid);
fz_free(fontdesc->cidtoucs);
fz_free(fontdesc->hmtx);
fz_free(fontdesc->vmtx);
fz_free(fontdesc);
}
}
static void dropitem(pdf_itemkind kind, void *val)
{
switch (kind)
{
case PDF_KCOLORSPACE: fz_dropcolorspace(val); break;
case PDF_KFUNCTION: pdf_dropfunction(val); break;
case PDF_KXOBJECT: pdf_dropxobject(val); break;
case PDF_KIMAGE: fz_dropimage(val); break;
case PDF_KPATTERN: pdf_droppattern(val); break;
case PDF_KSHADE: fz_dropshade(val); break;
case PDF_KCMAP: pdf_dropcmap(val); break;
case PDF_KFONT: pdf_dropfont(val); break;
}
}
fz_error
pdf_loadtounicode(pdf_fontdesc *font, pdf_xref *xref,
char **strings, char *collection, fz_obj *cmapstm)
{
fz_error error = fz_okay;
pdf_cmap *cmap;
int cid;
int ucsbuf[8];
int ucslen;
int i;
if (pdf_isstream(xref, fz_tonum(cmapstm), fz_togen(cmapstm)))
{
pdf_logfont("tounicode embedded cmap\n");
error = pdf_loadembeddedcmap(&cmap, xref, cmapstm);
if (error)
return fz_rethrow(error, "cannot load embedded cmap (%d %d R)", fz_tonum(cmapstm), fz_togen(cmapstm));
font->tounicode = pdf_newcmap();
for (i = 0; i < (strings ? 256 : 65536); i++)
{
cid = pdf_lookupcmap(font->encoding, i);
if (cid >= 0)
{
ucslen = pdf_lookupcmapfull(cmap, i, ucsbuf);
if (ucslen == 1)
pdf_maprangetorange(font->tounicode, cid, cid, ucsbuf[0]);
if (ucslen > 1)
pdf_maponetomany(font->tounicode, cid, ucsbuf, ucslen);
}
}
pdf_sortcmap(font->tounicode);
pdf_dropcmap(cmap);
}
else if (collection)
{
pdf_logfont("tounicode cid collection (%s)\n", collection);
error = fz_okay;
if (!strcmp(collection, "Adobe-CNS1"))
error = pdf_loadsystemcmap(&font->tounicode, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
error = pdf_loadsystemcmap(&font->tounicode, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
error = pdf_loadsystemcmap(&font->tounicode, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Japan2"))
error = pdf_loadsystemcmap(&font->tounicode, "Adobe-Japan2-UCS2"); /* where's this? */
else if (!strcmp(collection, "Adobe-Korea1"))
error = pdf_loadsystemcmap(&font->tounicode, "Adobe-Korea1-UCS2");
if (error)
return fz_rethrow(error, "cannot load tounicode system cmap %s-UCS2", collection);
}
if (strings)
{
pdf_logfont("tounicode strings\n");
/* TODO one-to-many mappings */
font->ncidtoucs = 256;
font->cidtoucs = fz_calloc(256, sizeof(unsigned short));
for (i = 0; i < 256; i++)
{
if (strings[i])
font->cidtoucs[i] = pdf_lookupagl(strings[i]);
else
font->cidtoucs[i] = '?';
}
}
if (!font->tounicode && !font->cidtoucs)
{
pdf_logfont("tounicode could not be loaded\n");
/* TODO: synthesize a ToUnicode if it's a freetype font with
* cmap and/or post tables or if it has glyph names. */
}
return fz_okay;
}
/*
* Load CMap stream in PDF file
*/
fz_error
pdf_loadembeddedcmap(pdf_cmap **cmapp, pdf_xref *xref, fz_obj *stmref)
{
fz_error error = fz_okay;
fz_obj *stmobj;
fz_stream *file = nil;
pdf_cmap *cmap = nil;
pdf_cmap *usecmap;
fz_obj *wmode;
fz_obj *obj;
if ((*cmapp = pdf_finditem(xref->store, PDF_KCMAP, stmref)))
{
pdf_keepcmap(*cmapp);
return fz_okay;
}
pdf_logfont("load embedded cmap (%d %d R) {\n", fz_tonum(stmref), fz_togen(stmref));
stmobj = fz_resolveindirect(stmref);
error = pdf_openstream(&file, xref, fz_tonum(stmref), fz_togen(stmref));
if (error)
{
error = fz_rethrow(error, "cannot open cmap stream");
goto cleanup;
}
error = pdf_parsecmap(&cmap, file);
if (error)
{
error = fz_rethrow(error, "cannot parse cmap stream");
goto cleanup;
}
fz_dropstream(file);
wmode = fz_dictgets(stmobj, "WMode");
if (fz_isint(wmode))
{
pdf_logfont("wmode %d\n", wmode);
pdf_setwmode(cmap, fz_toint(wmode));
}
obj = fz_dictgets(stmobj, "UseCMap");
if (fz_isname(obj))
{
pdf_logfont("usecmap /%s\n", fz_toname(obj));
error = pdf_loadsystemcmap(&usecmap, fz_toname(obj));
if (error)
{
error = fz_rethrow(error, "cannot load system usecmap '%s'", fz_toname(obj));
goto cleanup;
}
pdf_setusecmap(cmap, usecmap);
pdf_dropcmap(usecmap);
}
else if (fz_isindirect(obj))
{
pdf_logfont("usecmap (%d %d R)\n", fz_tonum(obj), fz_togen(obj));
error = pdf_loadembeddedcmap(&usecmap, xref, obj);
if (error)
{
error = fz_rethrow(error, "cannot load embedded usecmap");
goto cleanup;
}
pdf_setusecmap(cmap, usecmap);
pdf_dropcmap(usecmap);
}
pdf_logfont("}\n");
error = pdf_storeitem(xref->store, PDF_KCMAP, stmref, cmap);
if (error)
{
error = fz_rethrow(error, "cannot store cmap resource");
goto cleanup;
}
*cmapp = cmap;
return fz_okay;
cleanup:
if (file)
fz_dropstream(file);
if (cmap)
pdf_dropcmap(cmap);
return error; /* already rethrown */
}
fz_error
pdf_parsecmap(pdf_cmap **cmapp, fz_stream *file)
{
fz_error error;
pdf_cmap *cmap;
char key[64];
char buf[256];
pdf_token_e tok;
int len;
cmap = pdf_newcmap();
strcpy(key, ".notdef");
while (1)
{
error = lexcmap(&tok, file, buf, sizeof buf, &len);
if (error)
{
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=642 */
fz_warn("syntaxerror in cmap");
tok = PDF_TEOF;
}
if (tok == PDF_TEOF)
break;
else if (tok == PDF_TNAME)
{
if (!strcmp(buf, "CMapName"))
{
error = parsecmapname(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap after /CMapName");
goto cleanup;
}
}
else if (!strcmp(buf, "WMode"))
{
error = parsewmode(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap after /WMode");
goto cleanup;
}
}
else
strlcpy(key, buf, sizeof key);
}
else if (tok == TUSECMAP)
{
strlcpy(cmap->usecmapname, key, sizeof(cmap->usecmapname));
}
else if (tok == TBEGINCODESPACERANGE)
{
error = parsecodespacerange(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap codespacerange");
goto cleanup;
}
}
else if (tok == TBEGINBFCHAR)
{
error = parsebfchar(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap bfchar");
goto cleanup;
}
}
else if (tok == TBEGINCIDCHAR)
{
error = parsecidchar(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap cidchar");
goto cleanup;
}
}
else if (tok == TBEGINBFRANGE)
{
error = parsebfrange(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap bfrange");
goto cleanup;
}
}
else if (tok == TBEGINCIDRANGE)
{
error = parsecidrange(cmap, file);
if (error)
{
error = fz_rethrow(error, "syntaxerror in cmap cidrange");
goto cleanup;
}
}
/* ignore everything else */
}
pdf_sortcmap(cmap);
*cmapp = cmap;
return fz_okay;
cleanup:
pdf_dropcmap(cmap);
return error; /* already rethrown */
}
