Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 4680 | right-hear | 1 | #include "fitz.h" |
| 2 | #include "mupdf.h" |
||
| 3 | |||
| 4 | /* Load or synthesize ToUnicode map for fonts */ |
||
| 5 | |||
| 6 | fz_error |
||
| 7 | pdf_load_to_unicode(pdf_font_desc *font, pdf_xref *xref, |
||
| 8 | char **strings, char *collection, fz_obj *cmapstm) |
||
| 9 | { |
||
| 10 | fz_error error = fz_okay; |
||
| 11 | pdf_cmap *cmap; |
||
| 12 | int cid; |
||
| 13 | int ucsbuf[8]; |
||
| 14 | int ucslen; |
||
| 15 | int i; |
||
| 16 | |||
| 17 | if (pdf_is_stream(xref, fz_to_num(cmapstm), fz_to_gen(cmapstm))) |
||
| 18 | { |
||
| 19 | error = pdf_load_embedded_cmap(&cmap, xref, cmapstm); |
||
| 20 | if (error) |
||
| 21 | return fz_rethrow(error, "cannot load embedded cmap (%d %d R)", fz_to_num(cmapstm), fz_to_gen(cmapstm)); |
||
| 22 | |||
| 23 | font->to_unicode = pdf_new_cmap(); |
||
| 24 | |||
| 25 | for (i = 0; i < (strings ? 256 : 65536); i++) |
||
| 26 | { |
||
| 27 | cid = pdf_lookup_cmap(font->encoding, i); |
||
| 28 | if (cid >= 0) |
||
| 29 | { |
||
| 30 | ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf); |
||
| 31 | if (ucslen == 1) |
||
| 32 | pdf_map_range_to_range(font->to_unicode, cid, cid, ucsbuf[0]); |
||
| 33 | if (ucslen > 1) |
||
| 34 | pdf_map_one_to_many(font->to_unicode, cid, ucsbuf, ucslen); |
||
| 35 | } |
||
| 36 | } |
||
| 37 | |||
| 38 | pdf_sort_cmap(font->to_unicode); |
||
| 39 | |||
| 40 | pdf_drop_cmap(cmap); |
||
| 41 | } |
||
| 42 | |||
| 43 | else if (collection) |
||
| 44 | { |
||
| 45 | error = fz_okay; |
||
| 46 | |||
| 47 | if (!strcmp(collection, "Adobe-CNS1")) |
||
| 48 | error = pdf_load_system_cmap(&font->to_unicode, "Adobe-CNS1-UCS2"); |
||
| 49 | else if (!strcmp(collection, "Adobe-GB1")) |
||
| 50 | error = pdf_load_system_cmap(&font->to_unicode, "Adobe-GB1-UCS2"); |
||
| 51 | else if (!strcmp(collection, "Adobe-Japan1")) |
||
| 52 | error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Japan1-UCS2"); |
||
| 53 | else if (!strcmp(collection, "Adobe-Korea1")) |
||
| 54 | error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Korea1-UCS2"); |
||
| 55 | |||
| 56 | if (error) |
||
| 57 | return fz_rethrow(error, "cannot load ToUnicode system cmap %s-UCS2", collection); |
||
| 58 | } |
||
| 59 | |||
| 60 | if (strings) |
||
| 61 | { |
||
| 62 | /* TODO one-to-many mappings */ |
||
| 63 | |||
| 64 | font->cid_to_ucs_len = 256; |
||
| 65 | font->cid_to_ucs = fz_calloc(256, sizeof(unsigned short)); |
||
| 66 | |||
| 67 | for (i = 0; i < 256; i++) |
||
| 68 | { |
||
| 69 | if (strings[i]) |
||
| 70 | font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]); |
||
| 71 | else |
||
| 72 | font->cid_to_ucs[i] = '?'; |
||
| 73 | } |
||
| 74 | } |
||
| 75 | |||
| 76 | if (!font->to_unicode && !font->cid_to_ucs) |
||
| 77 | { |
||
| 78 | /* TODO: synthesize a ToUnicode if it's a freetype font with |
||
| 79 | * cmap and/or post tables or if it has glyph names. */ |
||
| 80 | } |
||
| 81 | |||
| 82 | return fz_okay; |
||
| 83 | } |