int
pdf_lookup_cmap_full(pdf_cmap *cmap, int cpt, int *out)
{
int i, k, n;
int l = 0;
int r = cmap->rlen - 1;
int m;
while (l <= r)
{
m = (l + r) >> 1;
if (cpt < cmap->ranges[m].low)
r = m - 1;
else if (cpt > pdf_range_high(&cmap->ranges[m]))
l = m + 1;
else
{
k = cpt - cmap->ranges[m].low + cmap->ranges[m].offset;
if (pdf_range_flags(&cmap->ranges[m]) == PDF_CMAP_TABLE)
{
out[0] = cmap->table[k];
return 1;
}
else if (pdf_range_flags(&cmap->ranges[m]) == PDF_CMAP_MULTI)
{
n = cmap->ranges[m].offset;
for (i = 0; i < cmap->table[n]; i++)
out[i] = cmap->table[n + i + 1];
return cmap->table[n];
}
else
{
out[0] = k;
return 1;
}
}
}
if (cmap->usecmap)
return pdf_lookup_cmap_full(cmap->usecmap, cpt, out);
return 0;
}
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;
}