/*
* Create an Identity-* CMap (for both 1 and 2-byte encodings)
*/
pdf_cmap *
pdf_new_identity_cmap(int wmode, int bytes)
{
pdf_cmap *cmap = pdf_new_cmap();
sprintf(cmap->cmap_name, "Identity-%c", wmode ? 'V' : 'H');
pdf_add_codespace(cmap, 0x0000, 0xffff, bytes);
pdf_map_range_to_range(cmap, 0x0000, 0xffff, 0);
pdf_sort_cmap(cmap);
pdf_set_wmode(cmap, wmode);
return cmap;
}
/*
* Create an Identity-* CMap (for both 1 and 2-byte encodings)
*/
pdf_cmap *
pdf_new_identity_cmap(fz_context *ctx, int wmode, int bytes)
{
pdf_cmap *cmap = pdf_new_cmap(ctx);
fz_try(ctx)
{
sprintf(cmap->cmap_name, "Identity-%c", wmode ? 'V' : 'H');
pdf_add_codespace(ctx, cmap, 0x0000, 0xffff, bytes);
pdf_map_range_to_range(ctx, cmap, 0x0000, 0xffff, 0);
pdf_sort_cmap(ctx, cmap);
pdf_set_cmap_wmode(ctx, cmap, wmode);
}
fz_catch(ctx)
{
pdf_drop_cmap(ctx, cmap);
fz_rethrow(ctx);
}
return cmap;
}
/*
* Create an Identity-* CMap (for both 1 and 2-byte encodings)
*/
pdf_cmap *
pdf_new_identity_cmap(fz_context *ctx, int wmode, int bytes)
{
pdf_cmap *cmap = pdf_new_cmap(ctx);
fz_try(ctx)
{
unsigned int high = (1 << (bytes * 8)) - 1;
sprintf(cmap->cmap_name, "Identity-%c", wmode ? 'V' : 'H');
pdf_add_codespace(ctx, cmap, 0, high, bytes);
pdf_map_range_to_range(ctx, cmap, 0, high, 0);
pdf_sort_cmap(ctx, cmap);
pdf_set_cmap_wmode(ctx, cmap, wmode);
}
fz_catch(ctx)
{
pdf_drop_cmap(ctx, cmap);
fz_rethrow(ctx);
}
return cmap;
}
void
pdf_load_to_unicode(pdf_document *doc, pdf_font_desc *font,
char **strings, char *collection, pdf_obj *cmapstm)
{
pdf_cmap *cmap;
int cid;
int ucsbuf[8];
int ucslen;
int i;
fz_context *ctx = doc->ctx;
if (pdf_is_stream(doc, pdf_to_num(cmapstm), pdf_to_gen(cmapstm)))
{
cmap = pdf_load_embedded_cmap(doc, cmapstm);
font->to_unicode = pdf_new_cmap(ctx);
for (i = 0; i < (strings ? 256 : 65536); i++)
{
cid = pdf_lookup_cmap(font->encoding, i);
if (cid >= 0)
{
ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf);
if (ucslen == 1)
pdf_map_range_to_range(ctx, font->to_unicode, cid, cid, ucsbuf[0]);
if (ucslen > 1)
pdf_map_one_to_many(ctx, font->to_unicode, cid, ucsbuf, ucslen);
}
}
pdf_sort_cmap(ctx, font->to_unicode);
pdf_drop_cmap(ctx, cmap);
font->size += pdf_cmap_size(ctx, font->to_unicode);
}
else if (collection)
{
if (!strcmp(collection, "Adobe-CNS1"))
font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Korea1"))
font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Korea1-UCS2");
return;
}
if (strings)
{
/* TODO one-to-many mappings */
font->cid_to_ucs_len = 256;
font->cid_to_ucs = fz_malloc_array(ctx, 256, sizeof(unsigned short));
font->size += 256 * sizeof(unsigned short);
for (i = 0; i < 256; i++)
{
if (strings[i])
font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]);
else
font->cid_to_ucs[i] = '?';
}
}
if (!font->to_unicode && !font->cid_to_ucs)
{
/* TODO: synthesize a ToUnicode if it's a freetype font with
* cmap and/or post tables or if it has glyph names. */
}
}
fz_error
pdf_load_to_unicode(pdf_font_desc *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_is_stream(xref, fz_to_num(cmapstm), fz_to_gen(cmapstm)))
{
error = pdf_load_embedded_cmap(&cmap, xref, cmapstm);
if (error)
return fz_rethrow(error, "cannot load embedded cmap (%d %d R)", fz_to_num(cmapstm), fz_to_gen(cmapstm));
font->to_unicode = pdf_new_cmap();
for (i = 0; i < (strings ? 256 : 65536); i++)
{
cid = pdf_lookup_cmap(font->encoding, i);
if (cid >= 0)
{
ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf);
if (ucslen == 1)
pdf_map_range_to_range(font->to_unicode, cid, cid, ucsbuf[0]);
if (ucslen > 1)
pdf_map_one_to_many(font->to_unicode, cid, ucsbuf, ucslen);
}
}
pdf_sort_cmap(font->to_unicode);
pdf_drop_cmap(cmap);
}
else if (collection)
{
error = fz_okay;
if (!strcmp(collection, "Adobe-CNS1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-CNS1-UCS2");
else if (!strcmp(collection, "Adobe-GB1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-GB1-UCS2");
else if (!strcmp(collection, "Adobe-Japan1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Japan1-UCS2");
else if (!strcmp(collection, "Adobe-Korea1"))
error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Korea1-UCS2");
if (error)
return fz_rethrow(error, "cannot load ToUnicode system cmap %s-UCS2", collection);
}
if (strings)
{
/* TODO one-to-many mappings */
font->cid_to_ucs_len = 256;
font->cid_to_ucs = fz_calloc(256, sizeof(unsigned short));
for (i = 0; i < 256; i++)
{
if (strings[i])
font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]);
else
font->cid_to_ucs[i] = '?';
}
}
if (!font->to_unicode && !font->cid_to_ucs)
{
/* 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;
}
pdf_cmap *
pdf_load_cmap(fz_context *ctx, fz_stream *file)
{
pdf_cmap *cmap;
char key[64];
pdf_lexbuf buf;
int tok;
const char *where;
buf.size = PDF_LEXBUF_SMALL;
cmap = pdf_new_cmap(ctx);
strcpy(key, ".notdef");
fz_var(where);
fz_try(ctx)
{
while (1)
{
where = "";
tok = pdf_lex_cmap(file, &buf);
if (tok == PDF_TOK_EOF || tok == TOK_END_CMAP)
break;
else if (tok == PDF_TOK_NAME)
{
if (!strcmp(buf.scratch, "CMapName"))
{
where = " after CMapName";
pdf_parse_cmap_name(ctx, cmap, file);
}
else if (!strcmp(buf.scratch, "WMode"))
{
where = " after WMode";
pdf_parse_wmode(ctx, cmap, file);
}
else
fz_strlcpy(key, buf.scratch, sizeof key);
}
else if (tok == TOK_USECMAP)
{
fz_strlcpy(cmap->usecmap_name, key, sizeof(cmap->usecmap_name));
}
else if (tok == TOK_BEGIN_CODESPACE_RANGE)
{
where = " codespacerange";
pdf_parse_codespace_range(ctx, cmap, file);
}
else if (tok == TOK_BEGIN_BF_CHAR)
{
where = " bfchar";
pdf_parse_bf_char(ctx, cmap, file);
}
else if (tok == TOK_BEGIN_CID_CHAR)
{
where = " cidchar";
pdf_parse_cid_char(ctx, cmap, file);
}
else if (tok == TOK_BEGIN_BF_RANGE)
{
where = " bfrange";
pdf_parse_bf_range(ctx, cmap, file);
}
else if (tok == TOK_BEGIN_CID_RANGE)
{
where = "cidrange";
pdf_parse_cid_range(ctx, cmap, file);
}
/* ignore everything else */
}
pdf_sort_cmap(ctx, cmap);
}
fz_catch(ctx)
{
pdf_drop_cmap(ctx, cmap);
fz_throw(ctx, "syntaxerror in cmap%s", where);
}
return cmap;
}