static void
pdf_parse_bf_range_array(fz_context *ctx, pdf_cmap *cmap, fz_stream *file, int lo, int hi)
{
pdf_lexbuf buf;
int tok;
int dst[256];
int i;
buf.size = PDF_LEXBUF_SMALL;
while (1)
{
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
if (tok == PDF_TOK_CLOSE_ARRAY)
return;
/* Note: does not handle [ /Name /Name ... ] */
else if (tok != PDF_TOK_STRING)
fz_throw(ctx, "expected string or ]");
if (buf.len / 2)
{
for (i = 0; i < buf.len / 2; i++)
dst[i] = pdf_code_from_string(&buf.scratch[i * 2], 2);
pdf_map_one_to_many(ctx, cmap, lo, dst, buf.len / 2);
}
lo ++;
}
}
static void
pdf_parse_bf_range_array(pdf_cmap *cmap, fz_stream *file, int lo, int hi)
{
char buf[256];
int tok;
int len;
int dst[256];
int i;
while (1)
{
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
if (tok == PDF_TOK_CLOSE_ARRAY)
return;
/* Note: does not handle [ /Name /Name ... ] */
else if (tok != PDF_TOK_STRING)
fz_throw(file->ctx, "expected string or ]");
if (len / 2)
{
for (i = 0; i < len / 2; i++)
dst[i] = pdf_code_from_string(buf + i * 2, 2);
pdf_map_one_to_many(file->ctx, cmap, lo, dst, len / 2);
}
lo ++;
}
}
static void
pdf_parse_bf_char(fz_context *ctx, pdf_cmap *cmap, fz_stream *file)
{
pdf_lexbuf buf;
int tok;
int dst[256];
int src;
int i;
buf.size = PDF_LEXBUF_SMALL;
while (1)
{
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
if (tok == TOK_END_BF_CHAR)
return;
else if (tok != PDF_TOK_STRING)
fz_throw(ctx, "expected string or endbfchar");
src = pdf_code_from_string(buf.scratch, buf.len);
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
/* Note: does not handle /dstName */
if (tok != PDF_TOK_STRING)
fz_throw(ctx, "expected string");
if (buf.len / 2)
{
for (i = 0; i < buf.len / 2; i++)
dst[i] = pdf_code_from_string(&buf.scratch[i * 2], 2);
pdf_map_one_to_many(ctx, cmap, src, dst, i);
}
}
}
static void
pdf_parse_bf_char(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int tok;
int len;
int dst[256];
int src;
int i;
while (1)
{
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
if (tok == TOK_END_BF_CHAR)
return;
else if (tok != PDF_TOK_STRING)
fz_throw(file->ctx, "expected string or endbfchar");
src = pdf_code_from_string(buf, len);
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
/* Note: does not handle /dstName */
if (tok != PDF_TOK_STRING)
fz_throw(file->ctx, "expected string");
if (len / 2)
{
for (i = 0; i < len / 2; i++)
dst[i] = pdf_code_from_string(buf + i * 2, 2);
pdf_map_one_to_many(file->ctx, cmap, src, dst, i);
}
}
}
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;
}
static void
pdf_parse_bf_range(fz_context *ctx, pdf_cmap *cmap, fz_stream *file)
{
pdf_lexbuf buf;
int tok;
int lo, hi, dst;
buf.size = PDF_LEXBUF_SMALL;
while (1)
{
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
if (tok == TOK_END_BF_RANGE)
return;
else if (tok != PDF_TOK_STRING)
fz_throw(ctx, "expected string or endbfrange");
lo = pdf_code_from_string(buf.scratch, buf.len);
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
if (tok != PDF_TOK_STRING)
fz_throw(ctx, "expected string");
hi = pdf_code_from_string(buf.scratch, buf.len);
tok = pdf_lex_cmap(file, &buf);
/* RJW: "syntaxerror in cmap" */
if (tok == PDF_TOK_STRING)
{
if (buf.len == 2)
{
dst = pdf_code_from_string(buf.scratch, buf.len);
pdf_map_range_to_range(ctx, cmap, lo, hi, dst);
}
else
{
int dststr[256];
int i;
if (buf.len / 2)
{
for (i = 0; i < buf.len / 2; i++)
dststr[i] = pdf_code_from_string(&buf.scratch[i * 2], 2);
while (lo <= hi)
{
dststr[i-1] ++;
pdf_map_one_to_many(ctx, cmap, lo, dststr, i);
lo ++;
}
}
}
}
else if (tok == PDF_TOK_OPEN_ARRAY)
{
pdf_parse_bf_range_array(ctx, cmap, file, lo, hi);
/* RJW: "cannot map bfrange" */
}
else
{
fz_throw(ctx, "expected string or array or endbfrange");
}
}
}
static void
pdf_parse_bf_range(pdf_cmap *cmap, fz_stream *file)
{
char buf[256];
int tok;
int len;
int lo, hi, dst;
while (1)
{
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
if (tok == TOK_END_BF_RANGE)
return;
else if (tok != PDF_TOK_STRING)
fz_throw(file->ctx, "expected string or endbfrange");
lo = pdf_code_from_string(buf, len);
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
if (tok != PDF_TOK_STRING)
fz_throw(file->ctx, "expected string");
hi = pdf_code_from_string(buf, len);
tok = pdf_lex_cmap(file, buf, sizeof buf, &len);
/* RJW: "syntaxerror in cmap" */
if (tok == PDF_TOK_STRING)
{
if (len == 2)
{
dst = pdf_code_from_string(buf, len);
pdf_map_range_to_range(file->ctx, cmap, lo, hi, dst);
}
else
{
int dststr[256];
int i;
if (len / 2)
{
for (i = 0; i < len / 2; i++)
dststr[i] = pdf_code_from_string(buf + i * 2, 2);
while (lo <= hi)
{
dststr[i-1] ++;
pdf_map_one_to_many(file->ctx, cmap, lo, dststr, i);
lo ++;
}
}
}
}
else if (tok == PDF_TOK_OPEN_ARRAY)
{
pdf_parse_bf_range_array(cmap, file, lo, hi);
/* RJW: "cannot map bfrange" */
}
else
{
fz_throw(file->ctx, "expected string or array or endbfrange");
}
}
}