Rev 8429 | 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 | #include |
||
| 5 | #include FT_FREETYPE_H |
||
| 6 | #include FT_XFREE86_H |
||
| 7 | |||
| 8 | static fz_error pdf_load_font_descriptor(pdf_font_desc *fontdesc, pdf_xref *xref, fz_obj *dict, char *collection, char *basefont); |
||
| 9 | |||
| 10 | static char *base_font_names[14][7] = |
||
| 11 | { |
||
| 12 | { "Courier", "CourierNew", "CourierNewPSMT", NULL }, |
||
| 13 | { "Courier-Bold", "CourierNew,Bold", "Courier,Bold", |
||
| 14 | "CourierNewPS-BoldMT", "CourierNew-Bold", NULL }, |
||
| 15 | { "Courier-Oblique", "CourierNew,Italic", "Courier,Italic", |
||
| 16 | "CourierNewPS-ItalicMT", "CourierNew-Italic", NULL }, |
||
| 17 | { "Courier-BoldOblique", "CourierNew,BoldItalic", "Courier,BoldItalic", |
||
| 18 | "CourierNewPS-BoldItalicMT", "CourierNew-BoldItalic", NULL }, |
||
| 19 | { "Helvetica", "ArialMT", "Arial", NULL }, |
||
| 20 | { "Helvetica-Bold", "Arial-BoldMT", "Arial,Bold", "Arial-Bold", |
||
| 21 | "Helvetica,Bold", NULL }, |
||
| 22 | { "Helvetica-Oblique", "Arial-ItalicMT", "Arial,Italic", "Arial-Italic", |
||
| 23 | "Helvetica,Italic", "Helvetica-Italic", NULL }, |
||
| 24 | { "Helvetica-BoldOblique", "Arial-BoldItalicMT", |
||
| 25 | "Arial,BoldItalic", "Arial-BoldItalic", |
||
| 26 | "Helvetica,BoldItalic", "Helvetica-BoldItalic", NULL }, |
||
| 27 | { "Times-Roman", "TimesNewRomanPSMT", "TimesNewRoman", |
||
| 28 | "TimesNewRomanPS", NULL }, |
||
| 29 | { "Times-Bold", "TimesNewRomanPS-BoldMT", "TimesNewRoman,Bold", |
||
| 30 | "TimesNewRomanPS-Bold", "TimesNewRoman-Bold", NULL }, |
||
| 31 | { "Times-Italic", "TimesNewRomanPS-ItalicMT", "TimesNewRoman,Italic", |
||
| 32 | "TimesNewRomanPS-Italic", "TimesNewRoman-Italic", NULL }, |
||
| 33 | { "Times-BoldItalic", "TimesNewRomanPS-BoldItalicMT", |
||
| 34 | "TimesNewRoman,BoldItalic", "TimesNewRomanPS-BoldItalic", |
||
| 35 | "TimesNewRoman-BoldItalic", NULL }, |
||
| 36 | { "Symbol", NULL }, |
||
| 37 | { "ZapfDingbats", NULL } |
||
| 38 | }; |
||
| 39 | |||
| 40 | static int is_dynalab(char *name) |
||
| 41 | { |
||
| 42 | if (strstr(name, "HuaTian")) |
||
| 43 | return 1; |
||
| 44 | if (strstr(name, "MingLi")) |
||
| 45 | return 1; |
||
| 46 | if ((strstr(name, "DF") == name) || strstr(name, "+DF")) |
||
| 47 | return 1; |
||
| 48 | if ((strstr(name, "DLC") == name) || strstr(name, "+DLC")) |
||
| 49 | return 1; |
||
| 50 | return 0; |
||
| 51 | } |
||
| 52 | |||
| 53 | static int strcmp_ignore_space(char *a, char *b) |
||
| 54 | { |
||
| 55 | while (1) |
||
| 56 | { |
||
| 57 | while (*a == ' ') |
||
| 58 | a++; |
||
| 59 | while (*b == ' ') |
||
| 60 | b++; |
||
| 61 | if (*a != *b) |
||
| 62 | return 1; |
||
| 63 | if (*a == 0) |
||
| 64 | return *a != *b; |
||
| 65 | if (*b == 0) |
||
| 66 | return *a != *b; |
||
| 67 | a++; |
||
| 68 | b++; |
||
| 69 | } |
||
| 70 | } |
||
| 71 | |||
| 72 | static char *clean_font_name(char *fontname) |
||
| 73 | { |
||
| 74 | int i, k; |
||
| 75 | for (i = 0; i < 14; i++) |
||
| 76 | for (k = 0; base_font_names[i][k]; k++) |
||
| 77 | if (!strcmp_ignore_space(base_font_names[i][k], fontname)) |
||
| 78 | return base_font_names[i][0]; |
||
| 79 | return fontname; |
||
| 80 | } |
||
| 81 | |||
| 82 | /* |
||
| 83 | * FreeType and Rendering glue |
||
| 84 | */ |
||
| 85 | |||
| 86 | enum { UNKNOWN, TYPE1, TRUETYPE }; |
||
| 87 | |||
| 88 | static int ft_kind(FT_Face face) |
||
| 89 | { |
||
| 90 | const char *kind = FT_Get_X11_Font_Format(face); |
||
| 91 | if (!strcmp(kind, "TrueType")) |
||
| 92 | return TRUETYPE; |
||
| 93 | if (!strcmp(kind, "Type 1")) |
||
| 94 | return TYPE1; |
||
| 95 | if (!strcmp(kind, "CFF")) |
||
| 96 | return TYPE1; |
||
| 97 | if (!strcmp(kind, "CID Type 1")) |
||
| 98 | return TYPE1; |
||
| 99 | return UNKNOWN; |
||
| 100 | } |
||
| 101 | |||
| 102 | static int ft_is_bold(FT_Face face) |
||
| 103 | { |
||
| 104 | return face->style_flags & FT_STYLE_FLAG_BOLD; |
||
| 105 | } |
||
| 106 | |||
| 107 | static int ft_is_italic(FT_Face face) |
||
| 108 | { |
||
| 109 | return face->style_flags & FT_STYLE_FLAG_ITALIC; |
||
| 110 | } |
||
| 111 | |||
| 112 | static int ft_char_index(FT_Face face, int cid) |
||
| 113 | { |
||
| 114 | int gid = FT_Get_Char_Index(face, cid); |
||
| 115 | if (gid == 0) |
||
| 116 | gid = FT_Get_Char_Index(face, 0xf000 + cid); |
||
| 117 | |||
| 118 | /* some chinese fonts only ship the similarly looking 0x2026 */ |
||
| 119 | if (gid == 0 && cid == 0x22ef) |
||
| 120 | gid = FT_Get_Char_Index(face, 0x2026); |
||
| 121 | |||
| 122 | return gid; |
||
| 123 | } |
||
| 124 | |||
| 125 | static int ft_cid_to_gid(pdf_font_desc *fontdesc, int cid) |
||
| 126 | { |
||
| 127 | if (fontdesc->to_ttf_cmap) |
||
| 128 | { |
||
| 129 | cid = pdf_lookup_cmap(fontdesc->to_ttf_cmap, cid); |
||
| 130 | return ft_char_index(fontdesc->font->ft_face, cid); |
||
| 131 | } |
||
| 132 | |||
| 133 | if (fontdesc->cid_to_gid) |
||
| 134 | return fontdesc->cid_to_gid[cid]; |
||
| 135 | |||
| 136 | return cid; |
||
| 137 | } |
||
| 138 | |||
| 139 | int |
||
| 140 | pdf_font_cid_to_gid(pdf_font_desc *fontdesc, int cid) |
||
| 141 | { |
||
| 142 | if (fontdesc->font->ft_face) |
||
| 143 | return ft_cid_to_gid(fontdesc, cid); |
||
| 144 | return cid; |
||
| 145 | } |
||
| 146 | |||
| 147 | static int ft_width(pdf_font_desc *fontdesc, int cid) |
||
| 148 | { |
||
| 149 | int gid = ft_cid_to_gid(fontdesc, cid); |
||
| 150 | int fterr = FT_Load_Glyph(fontdesc->font->ft_face, gid, |
||
| 151 | FT_LOAD_NO_HINTING | FT_LOAD_NO_BITMAP | FT_LOAD_IGNORE_TRANSFORM); |
||
| 152 | if (fterr) |
||
| 153 | { |
||
| 154 | fz_warn("freetype load glyph (gid %d): %s", gid, ft_error_string(fterr)); |
||
| 155 | return 0; |
||
| 156 | } |
||
| 157 | return ((FT_Face)fontdesc->font->ft_face)->glyph->advance.x; |
||
| 158 | } |
||
| 159 | |||
| 160 | static int lookup_mre_code(char *name) |
||
| 161 | { |
||
| 162 | int i; |
||
| 163 | for (i = 0; i < 256; i++) |
||
| 164 | if (pdf_mac_roman[i] && !strcmp(name, pdf_mac_roman[i])) |
||
| 165 | return i; |
||
| 166 | return -1; |
||
| 167 | } |
||
| 168 | |||
| 169 | /* |
||
| 170 | * Load font files. |
||
| 171 | */ |
||
| 172 | |||
| 173 | static fz_error |
||
| 174 | pdf_load_builtin_font(pdf_font_desc *fontdesc, char *fontname) |
||
| 175 | { |
||
| 176 | fz_error error; |
||
| 177 | unsigned char *data; |
||
| 178 | unsigned int len; |
||
| 179 | |||
| 180 | data = pdf_find_builtin_font(fontname, &len); |
||
| 181 | if (!data) |
||
| 182 | return fz_throw("cannot find builtin font: '%s'", fontname); |
||
| 183 | |||
| 184 | error = fz_new_font_from_memory(&fontdesc->font, data, len, 0); |
||
| 185 | if (error) |
||
| 186 | return fz_rethrow(error, "cannot load freetype font from memory"); |
||
| 187 | |||
| 188 | if (!strcmp(fontname, "Symbol") || !strcmp(fontname, "ZapfDingbats")) |
||
| 189 | fontdesc->flags |= PDF_FD_SYMBOLIC; |
||
| 190 | |||
| 191 | return fz_okay; |
||
| 192 | } |
||
| 193 | |||
| 194 | static fz_error |
||
| 195 | pdf_load_substitute_font(pdf_font_desc *fontdesc, int mono, int serif, int bold, int italic) |
||
| 196 | { |
||
| 197 | fz_error error; |
||
| 198 | unsigned char *data; |
||
| 199 | unsigned int len; |
||
| 200 | |||
| 201 | data = pdf_find_substitute_font(mono, serif, bold, italic, &len); |
||
| 202 | if (!data) |
||
| 203 | return fz_throw("cannot find substitute font"); |
||
| 204 | |||
| 205 | error = fz_new_font_from_memory(&fontdesc->font, data, len, 0); |
||
| 206 | if (error) |
||
| 207 | return fz_rethrow(error, "cannot load freetype font from memory"); |
||
| 208 | |||
| 209 | fontdesc->font->ft_substitute = 1; |
||
| 210 | fontdesc->font->ft_bold = bold && !ft_is_bold(fontdesc->font->ft_face); |
||
| 211 | fontdesc->font->ft_italic = italic && !ft_is_italic(fontdesc->font->ft_face); |
||
| 212 | return fz_okay; |
||
| 213 | } |
||
| 214 | |||
| 215 | static fz_error |
||
| 216 | pdf_load_substitute_cjk_font(pdf_font_desc *fontdesc, int ros, int serif) |
||
| 217 | { |
||
| 218 | fz_error error; |
||
| 219 | unsigned char *data; |
||
| 220 | unsigned int len; |
||
| 221 | |||
| 222 | data = pdf_find_substitute_cjk_font(ros, serif, &len); |
||
| 223 | if (!data) |
||
| 224 | return fz_throw("cannot find builtin CJK font"); |
||
| 225 | |||
| 226 | error = fz_new_font_from_memory(&fontdesc->font, data, len, 0); |
||
| 227 | if (error) |
||
| 228 | return fz_rethrow(error, "cannot load builtin CJK font"); |
||
| 229 | |||
| 230 | fontdesc->font->ft_substitute = 1; |
||
| 231 | return fz_okay; |
||
| 232 | } |
||
| 233 | |||
| 234 | static fz_error |
||
| 235 | pdf_load_system_font(pdf_font_desc *fontdesc, char *fontname, char *collection) |
||
| 236 | { |
||
| 237 | fz_error error; |
||
| 238 | int bold = 0; |
||
| 239 | int italic = 0; |
||
| 240 | int serif = 0; |
||
| 241 | int mono = 0; |
||
| 242 | |||
| 243 | if (strstr(fontname, "Bold")) |
||
| 244 | bold = 1; |
||
| 245 | if (strstr(fontname, "Italic")) |
||
| 246 | italic = 1; |
||
| 247 | if (strstr(fontname, "Oblique")) |
||
| 248 | italic = 1; |
||
| 249 | |||
| 250 | if (fontdesc->flags & PDF_FD_FIXED_PITCH) |
||
| 251 | mono = 1; |
||
| 252 | if (fontdesc->flags & PDF_FD_SERIF) |
||
| 253 | serif = 1; |
||
| 254 | if (fontdesc->flags & PDF_FD_ITALIC) |
||
| 255 | italic = 1; |
||
| 256 | if (fontdesc->flags & PDF_FD_FORCE_BOLD) |
||
| 257 | bold = 1; |
||
| 258 | |||
| 259 | if (collection) |
||
| 260 | { |
||
| 261 | if (!strcmp(collection, "Adobe-CNS1")) |
||
| 262 | return pdf_load_substitute_cjk_font(fontdesc, PDF_ROS_CNS, serif); |
||
| 263 | else if (!strcmp(collection, "Adobe-GB1")) |
||
| 264 | return pdf_load_substitute_cjk_font(fontdesc, PDF_ROS_GB, serif); |
||
| 265 | else if (!strcmp(collection, "Adobe-Japan1")) |
||
| 266 | return pdf_load_substitute_cjk_font(fontdesc, PDF_ROS_JAPAN, serif); |
||
| 267 | else if (!strcmp(collection, "Adobe-Korea1")) |
||
| 268 | return pdf_load_substitute_cjk_font(fontdesc, PDF_ROS_KOREA, serif); |
||
| 269 | return fz_throw("unknown cid collection: %s", collection); |
||
| 270 | } |
||
| 271 | |||
| 272 | error = pdf_load_substitute_font(fontdesc, mono, serif, bold, italic); |
||
| 273 | if (error) |
||
| 274 | return fz_rethrow(error, "cannot load substitute font"); |
||
| 275 | |||
| 276 | return fz_okay; |
||
| 277 | } |
||
| 278 | |||
| 279 | static fz_error |
||
| 280 | pdf_load_embedded_font(pdf_font_desc *fontdesc, pdf_xref *xref, fz_obj *stmref) |
||
| 281 | { |
||
| 282 | fz_error error; |
||
| 283 | fz_buffer *buf; |
||
| 284 | |||
| 285 | error = pdf_load_stream(&buf, xref, fz_to_num(stmref), fz_to_gen(stmref)); |
||
| 286 | if (error) |
||
| 287 | return fz_rethrow(error, "cannot load font stream (%d %d R)", fz_to_num(stmref), fz_to_gen(stmref)); |
||
| 288 | |||
| 289 | error = fz_new_font_from_memory(&fontdesc->font, buf->data, buf->len, 0); |
||
| 290 | if (error) |
||
| 291 | { |
||
| 292 | fz_drop_buffer(buf); |
||
| 293 | return fz_rethrow(error, "cannot load embedded font (%d %d R)", fz_to_num(stmref), fz_to_gen(stmref)); |
||
| 294 | } |
||
| 295 | |||
| 296 | /* save the buffer so we can free it later */ |
||
| 297 | fontdesc->font->ft_data = buf->data; |
||
| 298 | fontdesc->font->ft_size = buf->len; |
||
| 299 | fz_free(buf); /* only free the fz_buffer struct, not the contained data */ |
||
| 300 | |||
| 301 | fontdesc->is_embedded = 1; |
||
| 302 | |||
| 303 | return fz_okay; |
||
| 304 | } |
||
| 305 | |||
| 306 | /* |
||
| 307 | * Create and destroy |
||
| 308 | */ |
||
| 309 | |||
| 310 | pdf_font_desc * |
||
| 311 | pdf_keep_font(pdf_font_desc *fontdesc) |
||
| 312 | { |
||
| 313 | fontdesc->refs ++; |
||
| 314 | return fontdesc; |
||
| 315 | } |
||
| 316 | |||
| 317 | void |
||
| 318 | pdf_drop_font(pdf_font_desc *fontdesc) |
||
| 319 | { |
||
| 320 | if (fontdesc && --fontdesc->refs == 0) |
||
| 321 | { |
||
| 322 | if (fontdesc->font) |
||
| 323 | fz_drop_font(fontdesc->font); |
||
| 324 | if (fontdesc->encoding) |
||
| 325 | pdf_drop_cmap(fontdesc->encoding); |
||
| 326 | if (fontdesc->to_ttf_cmap) |
||
| 327 | pdf_drop_cmap(fontdesc->to_ttf_cmap); |
||
| 328 | if (fontdesc->to_unicode) |
||
| 329 | pdf_drop_cmap(fontdesc->to_unicode); |
||
| 330 | fz_free(fontdesc->cid_to_gid); |
||
| 331 | fz_free(fontdesc->cid_to_ucs); |
||
| 332 | fz_free(fontdesc->hmtx); |
||
| 333 | fz_free(fontdesc->vmtx); |
||
| 334 | fz_free(fontdesc); |
||
| 335 | } |
||
| 336 | } |
||
| 337 | |||
| 338 | pdf_font_desc * |
||
| 339 | pdf_new_font_desc(void) |
||
| 340 | { |
||
| 341 | pdf_font_desc *fontdesc; |
||
| 342 | |||
| 343 | fontdesc = fz_malloc(sizeof(pdf_font_desc)); |
||
| 344 | fontdesc->refs = 1; |
||
| 345 | |||
| 346 | fontdesc->font = NULL; |
||
| 347 | |||
| 348 | fontdesc->flags = 0; |
||
| 349 | fontdesc->italic_angle = 0; |
||
| 350 | fontdesc->ascent = 0; |
||
| 351 | fontdesc->descent = 0; |
||
| 352 | fontdesc->cap_height = 0; |
||
| 353 | fontdesc->x_height = 0; |
||
| 354 | fontdesc->missing_width = 0; |
||
| 355 | |||
| 356 | fontdesc->encoding = NULL; |
||
| 357 | fontdesc->to_ttf_cmap = NULL; |
||
| 358 | fontdesc->cid_to_gid_len = 0; |
||
| 359 | fontdesc->cid_to_gid = NULL; |
||
| 360 | |||
| 361 | fontdesc->to_unicode = NULL; |
||
| 362 | fontdesc->cid_to_ucs_len = 0; |
||
| 363 | fontdesc->cid_to_ucs = NULL; |
||
| 364 | |||
| 365 | fontdesc->wmode = 0; |
||
| 366 | |||
| 367 | fontdesc->hmtx_cap = 0; |
||
| 368 | fontdesc->vmtx_cap = 0; |
||
| 369 | fontdesc->hmtx_len = 0; |
||
| 370 | fontdesc->vmtx_len = 0; |
||
| 371 | fontdesc->hmtx = NULL; |
||
| 372 | fontdesc->vmtx = NULL; |
||
| 373 | |||
| 374 | fontdesc->dhmtx.lo = 0x0000; |
||
| 375 | fontdesc->dhmtx.hi = 0xFFFF; |
||
| 376 | fontdesc->dhmtx.w = 1000; |
||
| 377 | |||
| 378 | fontdesc->dvmtx.lo = 0x0000; |
||
| 379 | fontdesc->dvmtx.hi = 0xFFFF; |
||
| 380 | fontdesc->dvmtx.x = 0; |
||
| 381 | fontdesc->dvmtx.y = 880; |
||
| 382 | fontdesc->dvmtx.w = -1000; |
||
| 383 | |||
| 384 | fontdesc->is_embedded = 0; |
||
| 385 | |||
| 386 | return fontdesc; |
||
| 387 | } |
||
| 388 | |||
| 389 | /* |
||
| 390 | * Simple fonts (Type1 and TrueType) |
||
| 391 | */ |
||
| 392 | |||
| 393 | static fz_error |
||
| 394 | pdf_load_simple_font(pdf_font_desc **fontdescp, pdf_xref *xref, fz_obj *dict) |
||
| 395 | { |
||
| 396 | fz_error error; |
||
| 397 | fz_obj *descriptor; |
||
| 398 | fz_obj *encoding; |
||
| 399 | fz_obj *widths; |
||
| 400 | unsigned short *etable = NULL; |
||
| 401 | pdf_font_desc *fontdesc; |
||
| 402 | FT_Face face; |
||
| 403 | FT_CharMap cmap; |
||
| 404 | int symbolic; |
||
| 405 | int kind; |
||
| 406 | |||
| 407 | char *basefont; |
||
| 408 | char *fontname; |
||
| 409 | char *estrings[256]; |
||
| 410 | char ebuffer[256][32]; |
||
| 411 | int i, k, n; |
||
| 412 | int fterr; |
||
| 413 | |||
| 414 | basefont = fz_to_name(fz_dict_gets(dict, "BaseFont")); |
||
| 415 | fontname = clean_font_name(basefont); |
||
| 416 | |||
| 417 | /* Load font file */ |
||
| 418 | |||
| 419 | fontdesc = pdf_new_font_desc(); |
||
| 420 | |||
| 421 | descriptor = fz_dict_gets(dict, "FontDescriptor"); |
||
| 422 | if (descriptor) |
||
| 423 | error = pdf_load_font_descriptor(fontdesc, xref, descriptor, NULL, basefont); |
||
| 424 | else |
||
| 425 | error = pdf_load_builtin_font(fontdesc, fontname); |
||
| 426 | if (error) |
||
| 427 | goto cleanup; |
||
| 428 | |||
| 429 | /* Some chinese documents mistakenly consider WinAnsiEncoding to be codepage 936 */ |
||
| 430 | if (!*fontdesc->font->name && |
||
| 431 | !fz_dict_gets(dict, "ToUnicode") && |
||
| 432 | !strcmp(fz_to_name(fz_dict_gets(dict, "Encoding")), "WinAnsiEncoding") && |
||
| 433 | fz_to_int(fz_dict_gets(descriptor, "Flags")) == 4) |
||
| 434 | { |
||
| 435 | /* note: without the comma, pdf_load_font_descriptor would prefer /FontName over /BaseFont */ |
||
| 436 | char *cp936fonts[] = { |
||
| 437 | "\xCB\xCE\xCC\xE5", "SimSun,Regular", |
||
| 438 | "\xBA\xDA\xCC\xE5", "SimHei,Regular", |
||
| 439 | "\xBF\xAC\xCC\xE5_GB2312", "SimKai,Regular", |
||
| 440 | "\xB7\xC2\xCB\xCE_GB2312", "SimFang,Regular", |
||
| 441 | "\xC1\xA5\xCA\xE9", "SimLi,Regular", |
||
| 442 | NULL |
||
| 443 | }; |
||
| 444 | for (i = 0; cp936fonts[i]; i += 2) |
||
| 445 | if (!strcmp(basefont, cp936fonts[i])) |
||
| 446 | break; |
||
| 447 | if (cp936fonts[i]) |
||
| 448 | { |
||
| 449 | fz_warn("workaround for S22PDF lying about chinese font encodings"); |
||
| 450 | pdf_drop_font(fontdesc); |
||
| 451 | fontdesc = pdf_new_font_desc(); |
||
| 452 | error = pdf_load_font_descriptor(fontdesc, xref, descriptor, "Adobe-GB1", cp936fonts[i+1]); |
||
| 453 | error |= pdf_load_system_cmap(&fontdesc->encoding, "GBK-EUC-H"); |
||
| 454 | error |= pdf_load_system_cmap(&fontdesc->to_unicode, "Adobe-GB1-UCS2"); |
||
| 455 | error |= pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-GB1-UCS2"); |
||
| 456 | if (error) |
||
| 457 | return fz_rethrow(error, "cannot load font"); |
||
| 458 | |||
| 459 | face = fontdesc->font->ft_face; |
||
| 460 | kind = ft_kind(face); |
||
| 461 | goto skip_encoding; |
||
| 462 | } |
||
| 463 | } |
||
| 464 | |||
| 465 | face = fontdesc->font->ft_face; |
||
| 466 | kind = ft_kind(face); |
||
| 467 | |||
| 468 | /* Encoding */ |
||
| 469 | |||
| 470 | symbolic = fontdesc->flags & 4; |
||
| 471 | |||
| 472 | if (face->num_charmaps > 0) |
||
| 473 | cmap = face->charmaps[0]; |
||
| 474 | else |
||
| 475 | cmap = NULL; |
||
| 476 | |||
| 477 | for (i = 0; i < face->num_charmaps; i++) |
||
| 478 | { |
||
| 479 | FT_CharMap test = face->charmaps[i]; |
||
| 480 | |||
| 481 | if (kind == TYPE1) |
||
| 482 | { |
||
| 483 | if (test->platform_id == 7) |
||
| 484 | cmap = test; |
||
| 485 | } |
||
| 486 | |||
| 487 | if (kind == TRUETYPE) |
||
| 488 | { |
||
| 489 | if (test->platform_id == 1 && test->encoding_id == 0) |
||
| 490 | cmap = test; |
||
| 491 | if (test->platform_id == 3 && test->encoding_id == 1) |
||
| 492 | cmap = test; |
||
| 493 | } |
||
| 494 | } |
||
| 495 | |||
| 496 | if (cmap) |
||
| 497 | { |
||
| 498 | fterr = FT_Set_Charmap(face, cmap); |
||
| 499 | if (fterr) |
||
| 500 | fz_warn("freetype could not set cmap: %s", ft_error_string(fterr)); |
||
| 501 | } |
||
| 502 | else |
||
| 503 | fz_warn("freetype could not find any cmaps"); |
||
| 504 | |||
| 505 | etable = fz_calloc(256, sizeof(unsigned short)); |
||
| 506 | for (i = 0; i < 256; i++) |
||
| 507 | { |
||
| 508 | estrings[i] = NULL; |
||
| 509 | etable[i] = 0; |
||
| 510 | } |
||
| 511 | |||
| 512 | encoding = fz_dict_gets(dict, "Encoding"); |
||
| 513 | if (encoding) |
||
| 514 | { |
||
| 515 | if (fz_is_name(encoding)) |
||
| 516 | pdf_load_encoding(estrings, fz_to_name(encoding)); |
||
| 517 | |||
| 518 | if (fz_is_dict(encoding)) |
||
| 519 | { |
||
| 520 | fz_obj *base, *diff, *item; |
||
| 521 | |||
| 522 | base = fz_dict_gets(encoding, "BaseEncoding"); |
||
| 523 | if (fz_is_name(base)) |
||
| 524 | pdf_load_encoding(estrings, fz_to_name(base)); |
||
| 525 | else if (!fontdesc->is_embedded && !symbolic) |
||
| 526 | pdf_load_encoding(estrings, "StandardEncoding"); |
||
| 527 | |||
| 528 | diff = fz_dict_gets(encoding, "Differences"); |
||
| 529 | if (fz_is_array(diff)) |
||
| 530 | { |
||
| 531 | n = fz_array_len(diff); |
||
| 532 | k = 0; |
||
| 533 | for (i = 0; i < n; i++) |
||
| 534 | { |
||
| 535 | item = fz_array_get(diff, i); |
||
| 536 | if (fz_is_int(item)) |
||
| 537 | k = fz_to_int(item); |
||
| 538 | if (fz_is_name(item)) |
||
| 539 | estrings[k++] = fz_to_name(item); |
||
| 540 | if (k < 0) k = 0; |
||
| 541 | if (k > 255) k = 255; |
||
| 542 | } |
||
| 543 | } |
||
| 544 | } |
||
| 545 | } |
||
| 546 | |||
| 547 | /* start with the builtin encoding */ |
||
| 548 | for (i = 0; i < 256; i++) |
||
| 549 | etable[i] = ft_char_index(face, i); |
||
| 550 | |||
| 551 | /* encode by glyph name where we can */ |
||
| 552 | if (kind == TYPE1) |
||
| 553 | { |
||
| 554 | for (i = 0; i < 256; i++) |
||
| 555 | { |
||
| 556 | if (estrings[i]) |
||
| 557 | { |
||
| 558 | etable[i] = FT_Get_Name_Index(face, estrings[i]); |
||
| 559 | if (etable[i] == 0) |
||
| 560 | { |
||
| 561 | int aglcode = pdf_lookup_agl(estrings[i]); |
||
| 562 | const char **dupnames = pdf_lookup_agl_duplicates(aglcode); |
||
| 563 | while (*dupnames) |
||
| 564 | { |
||
| 565 | etable[i] = FT_Get_Name_Index(face, (char*)*dupnames); |
||
| 566 | if (etable[i]) |
||
| 567 | break; |
||
| 568 | dupnames++; |
||
| 569 | } |
||
| 570 | } |
||
| 571 | } |
||
| 572 | } |
||
| 573 | } |
||
| 574 | |||
| 575 | /* encode by glyph name where we can */ |
||
| 576 | if (kind == TRUETYPE) |
||
| 577 | { |
||
| 578 | /* Unicode cmap */ |
||
| 579 | if (!symbolic && face->charmap && face->charmap->platform_id == 3) |
||
| 580 | { |
||
| 581 | for (i = 0; i < 256; i++) |
||
| 582 | { |
||
| 583 | if (estrings[i]) |
||
| 584 | { |
||
| 585 | int aglcode = pdf_lookup_agl(estrings[i]); |
||
| 586 | if (!aglcode) |
||
| 587 | etable[i] = FT_Get_Name_Index(face, estrings[i]); |
||
| 588 | else |
||
| 589 | etable[i] = ft_char_index(face, aglcode); |
||
| 590 | } |
||
| 591 | } |
||
| 592 | } |
||
| 593 | |||
| 594 | /* MacRoman cmap */ |
||
| 595 | else if (!symbolic && face->charmap && face->charmap->platform_id == 1) |
||
| 596 | { |
||
| 597 | for (i = 0; i < 256; i++) |
||
| 598 | { |
||
| 599 | if (estrings[i]) |
||
| 600 | { |
||
| 601 | k = lookup_mre_code(estrings[i]); |
||
| 602 | if (k <= 0) |
||
| 603 | etable[i] = FT_Get_Name_Index(face, estrings[i]); |
||
| 604 | else |
||
| 605 | etable[i] = ft_char_index(face, k); |
||
| 606 | } |
||
| 607 | } |
||
| 608 | } |
||
| 609 | |||
| 610 | /* Symbolic cmap */ |
||
| 611 | else |
||
| 612 | { |
||
| 613 | for (i = 0; i < 256; i++) |
||
| 614 | { |
||
| 615 | if (estrings[i]) |
||
| 616 | { |
||
| 617 | etable[i] = FT_Get_Name_Index(face, estrings[i]); |
||
| 618 | if (etable[i] == 0) |
||
| 619 | etable[i] = ft_char_index(face, i); |
||
| 620 | } |
||
| 621 | } |
||
| 622 | } |
||
| 623 | } |
||
| 624 | |||
| 625 | /* try to reverse the glyph names from the builtin encoding */ |
||
| 626 | for (i = 0; i < 256; i++) |
||
| 627 | { |
||
| 628 | if (etable[i] && !estrings[i]) |
||
| 629 | { |
||
| 630 | if (FT_HAS_GLYPH_NAMES(face)) |
||
| 631 | { |
||
| 632 | fterr = FT_Get_Glyph_Name(face, etable[i], ebuffer[i], 32); |
||
| 633 | if (fterr) |
||
| 634 | fz_warn("freetype get glyph name (gid %d): %s", etable[i], ft_error_string(fterr)); |
||
| 635 | if (ebuffer[i][0]) |
||
| 636 | estrings[i] = ebuffer[i]; |
||
| 637 | } |
||
| 638 | else |
||
| 639 | { |
||
| 640 | estrings[i] = (char*) pdf_win_ansi[i]; /* discard const */ |
||
| 641 | } |
||
| 642 | } |
||
| 643 | } |
||
| 644 | |||
| 645 | fontdesc->encoding = pdf_new_identity_cmap(0, 1); |
||
| 646 | fontdesc->cid_to_gid_len = 256; |
||
| 647 | fontdesc->cid_to_gid = etable; |
||
| 648 | |||
| 649 | error = pdf_load_to_unicode(fontdesc, xref, estrings, NULL, fz_dict_gets(dict, "ToUnicode")); |
||
| 650 | if (error) |
||
| 651 | fz_catch(error, "cannot load to_unicode"); |
||
| 652 | |||
| 653 | skip_encoding: |
||
| 654 | |||
| 655 | /* Widths */ |
||
| 656 | |||
| 657 | pdf_set_default_hmtx(fontdesc, fontdesc->missing_width); |
||
| 658 | |||
| 659 | widths = fz_dict_gets(dict, "Widths"); |
||
| 660 | if (widths) |
||
| 661 | { |
||
| 662 | int first, last; |
||
| 663 | |||
| 664 | first = fz_to_int(fz_dict_gets(dict, "FirstChar")); |
||
| 665 | last = fz_to_int(fz_dict_gets(dict, "LastChar")); |
||
| 666 | |||
| 667 | if (first < 0 || last > 255 || first > last) |
||
| 668 | first = last = 0; |
||
| 669 | |||
| 670 | for (i = 0; i < last - first + 1; i++) |
||
| 671 | { |
||
| 672 | int wid = fz_to_int(fz_array_get(widths, i)); |
||
| 673 | pdf_add_hmtx(fontdesc, i + first, i + first, wid); |
||
| 674 | } |
||
| 675 | } |
||
| 676 | else |
||
| 677 | { |
||
| 678 | fterr = FT_Set_Char_Size(face, 1000, 1000, 72, 72); |
||
| 679 | if (fterr) |
||
| 680 | fz_warn("freetype set character size: %s", ft_error_string(fterr)); |
||
| 681 | for (i = 0; i < 256; i++) |
||
| 682 | { |
||
| 683 | pdf_add_hmtx(fontdesc, i, i, ft_width(fontdesc, i)); |
||
| 684 | } |
||
| 685 | } |
||
| 686 | |||
| 687 | pdf_end_hmtx(fontdesc); |
||
| 688 | |||
| 689 | *fontdescp = fontdesc; |
||
| 690 | return fz_okay; |
||
| 691 | |||
| 692 | cleanup: |
||
| 693 | if (etable != fontdesc->cid_to_gid) |
||
| 694 | fz_free(etable); |
||
| 695 | pdf_drop_font(fontdesc); |
||
| 696 | return fz_rethrow(error, "cannot load simple font (%d %d R)", fz_to_num(dict), fz_to_gen(dict)); |
||
| 697 | } |
||
| 698 | |||
| 699 | /* |
||
| 700 | * CID Fonts |
||
| 701 | */ |
||
| 702 | |||
| 703 | static fz_error |
||
| 704 | load_cid_font(pdf_font_desc **fontdescp, pdf_xref *xref, fz_obj *dict, fz_obj *encoding, fz_obj *to_unicode) |
||
| 705 | { |
||
| 706 | fz_error error; |
||
| 707 | fz_obj *widths; |
||
| 708 | fz_obj *descriptor; |
||
| 709 | pdf_font_desc *fontdesc; |
||
| 710 | FT_Face face; |
||
| 711 | int kind; |
||
| 712 | char collection[256]; |
||
| 713 | char *basefont; |
||
| 714 | int i, k, fterr; |
||
| 715 | fz_obj *obj; |
||
| 716 | int dw; |
||
| 717 | |||
| 718 | /* Get font name and CID collection */ |
||
| 719 | |||
| 720 | basefont = fz_to_name(fz_dict_gets(dict, "BaseFont")); |
||
| 721 | |||
| 722 | { |
||
| 723 | fz_obj *cidinfo; |
||
| 724 | char tmpstr[64]; |
||
| 725 | int tmplen; |
||
| 726 | |||
| 727 | cidinfo = fz_dict_gets(dict, "CIDSystemInfo"); |
||
| 728 | if (!cidinfo) |
||
| 729 | return fz_throw("cid font is missing info"); |
||
| 730 | |||
| 731 | obj = fz_dict_gets(cidinfo, "Registry"); |
||
| 732 | tmplen = MIN(sizeof tmpstr - 1, fz_to_str_len(obj)); |
||
| 733 | memcpy(tmpstr, fz_to_str_buf(obj), tmplen); |
||
| 734 | tmpstr[tmplen] = '\0'; |
||
| 735 | fz_strlcpy(collection, tmpstr, sizeof collection); |
||
| 736 | |||
| 737 | fz_strlcat(collection, "-", sizeof collection); |
||
| 738 | |||
| 739 | obj = fz_dict_gets(cidinfo, "Ordering"); |
||
| 740 | tmplen = MIN(sizeof tmpstr - 1, fz_to_str_len(obj)); |
||
| 741 | memcpy(tmpstr, fz_to_str_buf(obj), tmplen); |
||
| 742 | tmpstr[tmplen] = '\0'; |
||
| 743 | fz_strlcat(collection, tmpstr, sizeof collection); |
||
| 744 | } |
||
| 745 | |||
| 746 | /* Load font file */ |
||
| 747 | |||
| 748 | fontdesc = pdf_new_font_desc(); |
||
| 749 | |||
| 750 | descriptor = fz_dict_gets(dict, "FontDescriptor"); |
||
| 751 | if (descriptor) |
||
| 752 | error = pdf_load_font_descriptor(fontdesc, xref, descriptor, collection, basefont); |
||
| 753 | else |
||
| 754 | error = fz_throw("syntaxerror: missing font descriptor"); |
||
| 755 | if (error) |
||
| 756 | goto cleanup; |
||
| 757 | |||
| 758 | face = fontdesc->font->ft_face; |
||
| 759 | kind = ft_kind(face); |
||
| 760 | |||
| 761 | /* Encoding */ |
||
| 762 | |||
| 763 | error = fz_okay; |
||
| 764 | if (fz_is_name(encoding)) |
||
| 765 | { |
||
| 766 | if (!strcmp(fz_to_name(encoding), "Identity-H")) |
||
| 767 | fontdesc->encoding = pdf_new_identity_cmap(0, 2); |
||
| 768 | else if (!strcmp(fz_to_name(encoding), "Identity-V")) |
||
| 769 | fontdesc->encoding = pdf_new_identity_cmap(1, 2); |
||
| 770 | else |
||
| 771 | error = pdf_load_system_cmap(&fontdesc->encoding, fz_to_name(encoding)); |
||
| 772 | } |
||
| 773 | else if (fz_is_indirect(encoding)) |
||
| 774 | { |
||
| 775 | error = pdf_load_embedded_cmap(&fontdesc->encoding, xref, encoding); |
||
| 776 | } |
||
| 777 | else |
||
| 778 | { |
||
| 779 | error = fz_throw("syntaxerror: font missing encoding"); |
||
| 780 | } |
||
| 781 | if (error) |
||
| 782 | goto cleanup; |
||
| 783 | |||
| 784 | pdf_set_font_wmode(fontdesc, pdf_get_wmode(fontdesc->encoding)); |
||
| 785 | |||
| 786 | if (kind == TRUETYPE) |
||
| 787 | { |
||
| 788 | fz_obj *cidtogidmap; |
||
| 789 | |||
| 790 | cidtogidmap = fz_dict_gets(dict, "CIDToGIDMap"); |
||
| 791 | if (fz_is_indirect(cidtogidmap)) |
||
| 792 | { |
||
| 793 | fz_buffer *buf; |
||
| 794 | |||
| 795 | error = pdf_load_stream(&buf, xref, fz_to_num(cidtogidmap), fz_to_gen(cidtogidmap)); |
||
| 796 | if (error) |
||
| 797 | goto cleanup; |
||
| 798 | |||
| 799 | fontdesc->cid_to_gid_len = (buf->len) / 2; |
||
| 800 | fontdesc->cid_to_gid = fz_calloc(fontdesc->cid_to_gid_len, sizeof(unsigned short)); |
||
| 801 | for (i = 0; i < fontdesc->cid_to_gid_len; i++) |
||
| 802 | fontdesc->cid_to_gid[i] = (buf->data[i * 2] << 8) + buf->data[i * 2 + 1]; |
||
| 803 | |||
| 804 | fz_drop_buffer(buf); |
||
| 805 | } |
||
| 806 | |||
| 807 | /* if truetype font is external, cidtogidmap should not be identity */ |
||
| 808 | /* so we map from cid to unicode and then map that through the (3 1) */ |
||
| 809 | /* unicode cmap to get a glyph id */ |
||
| 810 | else if (fontdesc->font->ft_substitute) |
||
| 811 | { |
||
| 812 | fterr = FT_Select_Charmap(face, ft_encoding_unicode); |
||
| 813 | if (fterr) |
||
| 814 | { |
||
| 815 | error = fz_throw("fonterror: no unicode cmap when emulating CID font: %s", ft_error_string(fterr)); |
||
| 816 | goto cleanup; |
||
| 817 | } |
||
| 818 | |||
| 819 | if (!strcmp(collection, "Adobe-CNS1")) |
||
| 820 | error = pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-CNS1-UCS2"); |
||
| 821 | else if (!strcmp(collection, "Adobe-GB1")) |
||
| 822 | error = pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-GB1-UCS2"); |
||
| 823 | else if (!strcmp(collection, "Adobe-Japan1")) |
||
| 824 | error = pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-Japan1-UCS2"); |
||
| 825 | else if (!strcmp(collection, "Adobe-Japan2")) |
||
| 826 | error = pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-Japan2-UCS2"); |
||
| 827 | else if (!strcmp(collection, "Adobe-Korea1")) |
||
| 828 | error = pdf_load_system_cmap(&fontdesc->to_ttf_cmap, "Adobe-Korea1-UCS2"); |
||
| 829 | else |
||
| 830 | error = fz_okay; |
||
| 831 | |||
| 832 | if (error) |
||
| 833 | { |
||
| 834 | error = fz_rethrow(error, "cannot load system cmap %s", collection); |
||
| 835 | goto cleanup; |
||
| 836 | } |
||
| 837 | } |
||
| 838 | } |
||
| 839 | |||
| 840 | error = pdf_load_to_unicode(fontdesc, xref, NULL, collection, to_unicode); |
||
| 841 | if (error) |
||
| 842 | fz_catch(error, "cannot load to_unicode"); |
||
| 843 | |||
| 844 | /* Horizontal */ |
||
| 845 | |||
| 846 | dw = 1000; |
||
| 847 | obj = fz_dict_gets(dict, "DW"); |
||
| 848 | if (obj) |
||
| 849 | dw = fz_to_int(obj); |
||
| 850 | pdf_set_default_hmtx(fontdesc, dw); |
||
| 851 | |||
| 852 | widths = fz_dict_gets(dict, "W"); |
||
| 853 | if (widths) |
||
| 854 | { |
||
| 855 | int c0, c1, w; |
||
| 856 | |||
| 857 | for (i = 0; i < fz_array_len(widths); ) |
||
| 858 | { |
||
| 859 | c0 = fz_to_int(fz_array_get(widths, i)); |
||
| 860 | obj = fz_array_get(widths, i + 1); |
||
| 861 | if (fz_is_array(obj)) |
||
| 862 | { |
||
| 863 | for (k = 0; k < fz_array_len(obj); k++) |
||
| 864 | { |
||
| 865 | w = fz_to_int(fz_array_get(obj, k)); |
||
| 866 | pdf_add_hmtx(fontdesc, c0 + k, c0 + k, w); |
||
| 867 | } |
||
| 868 | i += 2; |
||
| 869 | } |
||
| 870 | else |
||
| 871 | { |
||
| 872 | c1 = fz_to_int(obj); |
||
| 873 | w = fz_to_int(fz_array_get(widths, i + 2)); |
||
| 874 | pdf_add_hmtx(fontdesc, c0, c1, w); |
||
| 875 | i += 3; |
||
| 876 | } |
||
| 877 | } |
||
| 878 | } |
||
| 879 | |||
| 880 | pdf_end_hmtx(fontdesc); |
||
| 881 | |||
| 882 | /* Vertical */ |
||
| 883 | |||
| 884 | if (pdf_get_wmode(fontdesc->encoding) == 1) |
||
| 885 | { |
||
| 886 | int dw2y = 880; |
||
| 887 | int dw2w = -1000; |
||
| 888 | |||
| 889 | obj = fz_dict_gets(dict, "DW2"); |
||
| 890 | if (obj) |
||
| 891 | { |
||
| 892 | dw2y = fz_to_int(fz_array_get(obj, 0)); |
||
| 893 | dw2w = fz_to_int(fz_array_get(obj, 1)); |
||
| 894 | } |
||
| 895 | |||
| 896 | pdf_set_default_vmtx(fontdesc, dw2y, dw2w); |
||
| 897 | |||
| 898 | widths = fz_dict_gets(dict, "W2"); |
||
| 899 | if (widths) |
||
| 900 | { |
||
| 901 | int c0, c1, w, x, y; |
||
| 902 | |||
| 903 | for (i = 0; i < fz_array_len(widths); ) |
||
| 904 | { |
||
| 905 | c0 = fz_to_int(fz_array_get(widths, i)); |
||
| 906 | obj = fz_array_get(widths, i + 1); |
||
| 907 | if (fz_is_array(obj)) |
||
| 908 | { |
||
| 909 | for (k = 0; k * 3 < fz_array_len(obj); k ++) |
||
| 910 | { |
||
| 911 | w = fz_to_int(fz_array_get(obj, k * 3 + 0)); |
||
| 912 | x = fz_to_int(fz_array_get(obj, k * 3 + 1)); |
||
| 913 | y = fz_to_int(fz_array_get(obj, k * 3 + 2)); |
||
| 914 | pdf_add_vmtx(fontdesc, c0 + k, c0 + k, x, y, w); |
||
| 915 | } |
||
| 916 | i += 2; |
||
| 917 | } |
||
| 918 | else |
||
| 919 | { |
||
| 920 | c1 = fz_to_int(obj); |
||
| 921 | w = fz_to_int(fz_array_get(widths, i + 2)); |
||
| 922 | x = fz_to_int(fz_array_get(widths, i + 3)); |
||
| 923 | y = fz_to_int(fz_array_get(widths, i + 4)); |
||
| 924 | pdf_add_vmtx(fontdesc, c0, c1, x, y, w); |
||
| 925 | i += 5; |
||
| 926 | } |
||
| 927 | } |
||
| 928 | } |
||
| 929 | |||
| 930 | pdf_end_vmtx(fontdesc); |
||
| 931 | } |
||
| 932 | |||
| 933 | *fontdescp = fontdesc; |
||
| 934 | return fz_okay; |
||
| 935 | |||
| 936 | cleanup: |
||
| 937 | pdf_drop_font(fontdesc); |
||
| 938 | return fz_rethrow(error, "cannot load cid font (%d %d R)", fz_to_num(dict), fz_to_gen(dict)); |
||
| 939 | } |
||
| 940 | |||
| 941 | static fz_error |
||
| 942 | pdf_load_type0_font(pdf_font_desc **fontdescp, pdf_xref *xref, fz_obj *dict) |
||
| 943 | { |
||
| 944 | fz_error error; |
||
| 945 | fz_obj *dfonts; |
||
| 946 | fz_obj *dfont; |
||
| 947 | fz_obj *subtype; |
||
| 948 | fz_obj *encoding; |
||
| 949 | fz_obj *to_unicode; |
||
| 950 | |||
| 951 | dfonts = fz_dict_gets(dict, "DescendantFonts"); |
||
| 952 | if (!dfonts) |
||
| 953 | return fz_throw("cid font is missing descendant fonts"); |
||
| 954 | |||
| 955 | dfont = fz_array_get(dfonts, 0); |
||
| 956 | |||
| 957 | subtype = fz_dict_gets(dfont, "Subtype"); |
||
| 958 | encoding = fz_dict_gets(dict, "Encoding"); |
||
| 959 | to_unicode = fz_dict_gets(dict, "ToUnicode"); |
||
| 960 | |||
| 961 | if (fz_is_name(subtype) && !strcmp(fz_to_name(subtype), "CIDFontType0")) |
||
| 962 | error = load_cid_font(fontdescp, xref, dfont, encoding, to_unicode); |
||
| 963 | else if (fz_is_name(subtype) && !strcmp(fz_to_name(subtype), "CIDFontType2")) |
||
| 964 | error = load_cid_font(fontdescp, xref, dfont, encoding, to_unicode); |
||
| 965 | else |
||
| 966 | error = fz_throw("syntaxerror: unknown cid font type"); |
||
| 967 | if (error) |
||
| 968 | return fz_rethrow(error, "cannot load descendant font (%d %d R)", fz_to_num(dfont), fz_to_gen(dfont)); |
||
| 969 | |||
| 970 | return fz_okay; |
||
| 971 | } |
||
| 972 | |||
| 973 | /* |
||
| 974 | * FontDescriptor |
||
| 975 | */ |
||
| 976 | |||
| 977 | static fz_error |
||
| 978 | pdf_load_font_descriptor(pdf_font_desc *fontdesc, pdf_xref *xref, fz_obj *dict, char *collection, char *basefont) |
||
| 979 | { |
||
| 980 | fz_error error; |
||
| 981 | fz_obj *obj1, *obj2, *obj3, *obj; |
||
| 982 | char *fontname; |
||
| 983 | char *origname; |
||
| 984 | FT_Face face; |
||
| 985 | |||
| 986 | if (!strchr(basefont, ',') || strchr(basefont, '+')) |
||
| 987 | origname = fz_to_name(fz_dict_gets(dict, "FontName")); |
||
| 988 | else |
||
| 989 | origname = basefont; |
||
| 990 | fontname = clean_font_name(origname); |
||
| 991 | |||
| 992 | fontdesc->flags = fz_to_int(fz_dict_gets(dict, "Flags")); |
||
| 993 | fontdesc->italic_angle = fz_to_real(fz_dict_gets(dict, "ItalicAngle")); |
||
| 994 | fontdesc->ascent = fz_to_real(fz_dict_gets(dict, "Ascent")); |
||
| 995 | fontdesc->descent = fz_to_real(fz_dict_gets(dict, "Descent")); |
||
| 996 | fontdesc->cap_height = fz_to_real(fz_dict_gets(dict, "CapHeight")); |
||
| 997 | fontdesc->x_height = fz_to_real(fz_dict_gets(dict, "XHeight")); |
||
| 998 | fontdesc->missing_width = fz_to_real(fz_dict_gets(dict, "MissingWidth")); |
||
| 999 | |||
| 1000 | obj1 = fz_dict_gets(dict, "FontFile"); |
||
| 1001 | obj2 = fz_dict_gets(dict, "FontFile2"); |
||
| 1002 | obj3 = fz_dict_gets(dict, "FontFile3"); |
||
| 1003 | obj = obj1 ? obj1 : obj2 ? obj2 : obj3; |
||
| 1004 | |||
| 1005 | if (fz_is_indirect(obj)) |
||
| 1006 | { |
||
| 1007 | error = pdf_load_embedded_font(fontdesc, xref, obj); |
||
| 1008 | if (error) |
||
| 1009 | { |
||
| 1010 | fz_catch(error, "ignored error when loading embedded font, attempting to load system font"); |
||
| 1011 | if (origname != fontname) |
||
| 1012 | error = pdf_load_builtin_font(fontdesc, fontname); |
||
| 1013 | else |
||
| 1014 | error = pdf_load_system_font(fontdesc, fontname, collection); |
||
| 1015 | if (error) |
||
| 1016 | return fz_rethrow(error, "cannot load font descriptor (%d %d R)", fz_to_num(dict), fz_to_gen(dict)); |
||
| 1017 | } |
||
| 1018 | } |
||
| 1019 | else |
||
| 1020 | { |
||
| 1021 | if (origname != fontname) |
||
| 1022 | error = pdf_load_builtin_font(fontdesc, fontname); |
||
| 1023 | else |
||
| 1024 | error = pdf_load_system_font(fontdesc, fontname, collection); |
||
| 1025 | if (error) |
||
| 1026 | return fz_rethrow(error, "cannot load font descriptor (%d %d R)", fz_to_num(dict), fz_to_gen(dict)); |
||
| 1027 | } |
||
| 1028 | |||
| 1029 | fz_strlcpy(fontdesc->font->name, fontname, sizeof fontdesc->font->name); |
||
| 1030 | |||
| 1031 | /* Check for DynaLab fonts that must use hinting */ |
||
| 1032 | face = fontdesc->font->ft_face; |
||
| 1033 | if (ft_kind(face) == TRUETYPE) |
||
| 1034 | { |
||
| 1035 | if (FT_IS_TRICKY(face) || is_dynalab(fontdesc->font->name)) |
||
| 1036 | fontdesc->font->ft_hint = 1; |
||
| 1037 | } |
||
| 1038 | |||
| 1039 | return fz_okay; |
||
| 1040 | |||
| 1041 | } |
||
| 1042 | |||
| 1043 | static void |
||
| 1044 | pdf_make_width_table(pdf_font_desc *fontdesc) |
||
| 1045 | { |
||
| 1046 | fz_font *font = fontdesc->font; |
||
| 1047 | int i, k, cid, gid; |
||
| 1048 | |||
| 1049 | font->width_count = 0; |
||
| 1050 | for (i = 0; i < fontdesc->hmtx_len; i++) |
||
| 1051 | { |
||
| 1052 | for (k = fontdesc->hmtx[i].lo; k <= fontdesc->hmtx[i].hi; k++) |
||
| 1053 | { |
||
| 1054 | cid = pdf_lookup_cmap(fontdesc->encoding, k); |
||
| 1055 | gid = pdf_font_cid_to_gid(fontdesc, cid); |
||
| 1056 | if (gid > font->width_count) |
||
| 1057 | font->width_count = gid; |
||
| 1058 | } |
||
| 1059 | } |
||
| 1060 | font->width_count ++; |
||
| 1061 | |||
| 1062 | font->width_table = fz_calloc(font->width_count, sizeof(int)); |
||
| 1063 | memset(font->width_table, 0, sizeof(int) * font->width_count); |
||
| 1064 | |||
| 1065 | for (i = 0; i < fontdesc->hmtx_len; i++) |
||
| 1066 | { |
||
| 1067 | for (k = fontdesc->hmtx[i].lo; k <= fontdesc->hmtx[i].hi; k++) |
||
| 1068 | { |
||
| 1069 | cid = pdf_lookup_cmap(fontdesc->encoding, k); |
||
| 1070 | gid = pdf_font_cid_to_gid(fontdesc, cid); |
||
| 1071 | if (gid >= 0 && gid < font->width_count) |
||
| 1072 | font->width_table[gid] = fontdesc->hmtx[i].w; |
||
| 1073 | } |
||
| 1074 | } |
||
| 1075 | } |
||
| 1076 | |||
| 1077 | fz_error |
||
| 1078 | pdf_load_font(pdf_font_desc **fontdescp, pdf_xref *xref, fz_obj *rdb, fz_obj *dict) |
||
| 1079 | { |
||
| 1080 | fz_error error; |
||
| 1081 | char *subtype; |
||
| 1082 | fz_obj *dfonts; |
||
| 1083 | fz_obj *charprocs; |
||
| 1084 | |||
| 1085 | if ((*fontdescp = pdf_find_item(xref->store, pdf_drop_font, dict))) |
||
| 1086 | { |
||
| 1087 | pdf_keep_font(*fontdescp); |
||
| 1088 | return fz_okay; |
||
| 1089 | } |
||
| 1090 | |||
| 1091 | subtype = fz_to_name(fz_dict_gets(dict, "Subtype")); |
||
| 1092 | dfonts = fz_dict_gets(dict, "DescendantFonts"); |
||
| 1093 | charprocs = fz_dict_gets(dict, "CharProcs"); |
||
| 1094 | |||
| 1095 | if (subtype && !strcmp(subtype, "Type0")) |
||
| 1096 | error = pdf_load_type0_font(fontdescp, xref, dict); |
||
| 1097 | else if (subtype && !strcmp(subtype, "Type1")) |
||
| 1098 | error = pdf_load_simple_font(fontdescp, xref, dict); |
||
| 1099 | else if (subtype && !strcmp(subtype, "MMType1")) |
||
| 1100 | error = pdf_load_simple_font(fontdescp, xref, dict); |
||
| 1101 | else if (subtype && !strcmp(subtype, "TrueType")) |
||
| 1102 | error = pdf_load_simple_font(fontdescp, xref, dict); |
||
| 1103 | else if (subtype && !strcmp(subtype, "Type3")) |
||
| 1104 | error = pdf_load_type3_font(fontdescp, xref, rdb, dict); |
||
| 1105 | else if (charprocs) |
||
| 1106 | { |
||
| 1107 | fz_warn("unknown font format, guessing type3."); |
||
| 1108 | error = pdf_load_type3_font(fontdescp, xref, rdb, dict); |
||
| 1109 | } |
||
| 1110 | else if (dfonts) |
||
| 1111 | { |
||
| 1112 | fz_warn("unknown font format, guessing type0."); |
||
| 1113 | error = pdf_load_type0_font(fontdescp, xref, dict); |
||
| 1114 | } |
||
| 1115 | else |
||
| 1116 | { |
||
| 1117 | fz_warn("unknown font format, guessing type1 or truetype."); |
||
| 1118 | error = pdf_load_simple_font(fontdescp, xref, dict); |
||
| 1119 | } |
||
| 1120 | if (error) |
||
| 1121 | return fz_rethrow(error, "cannot load font (%d %d R)", fz_to_num(dict), fz_to_gen(dict)); |
||
| 1122 | |||
| 1123 | /* Save the widths to stretch non-CJK substitute fonts */ |
||
| 1124 | if ((*fontdescp)->font->ft_substitute && !(*fontdescp)->to_ttf_cmap) |
||
| 1125 | pdf_make_width_table(*fontdescp); |
||
| 1126 | |||
| 1127 | pdf_store_item(xref->store, pdf_keep_font, pdf_drop_font, dict, *fontdescp); |
||
| 1128 | |||
| 1129 | return fz_okay; |
||
| 1130 | } |
||
| 1131 | |||
| 1132 | void |
||
| 1133 | pdf_debug_font(pdf_font_desc *fontdesc) |
||
| 1134 | { |
||
| 1135 | int i; |
||
| 1136 | |||
| 1137 | printf("fontdesc {\n"); |
||
| 1138 | |||
| 1139 | if (fontdesc->font->ft_face) |
||
| 1140 | printf("\tfreetype font\n"); |
||
| 1141 | if (fontdesc->font->t3procs) |
||
| 1142 | printf("\ttype3 font\n"); |
||
| 1143 | |||
| 1144 | printf("\twmode %d\n", fontdesc->wmode); |
||
| 1145 | printf("\tDW %d\n", fontdesc->dhmtx.w); |
||
| 1146 | |||
| 1147 | printf("\tW {\n"); |
||
| 1148 | for (i = 0; i < fontdesc->hmtx_len; i++) |
||
| 1149 | printf("\t\t<%04x> <%04x> %d\n", |
||
| 1150 | fontdesc->hmtx[i].lo, fontdesc->hmtx[i].hi, fontdesc->hmtx[i].w); |
||
| 1151 | printf("\t}\n"); |
||
| 1152 | |||
| 1153 | if (fontdesc->wmode) |
||
| 1154 | { |
||
| 1155 | printf("\tDW2 [%d %d]\n", fontdesc->dvmtx.y, fontdesc->dvmtx.w); |
||
| 1156 | printf("\tW2 {\n"); |
||
| 1157 | for (i = 0; i < fontdesc->vmtx_len; i++) |
||
| 1158 | printf("\t\t<%04x> <%04x> %d %d %d\n", fontdesc->vmtx[i].lo, fontdesc->vmtx[i].hi, |
||
| 1159 | fontdesc->vmtx[i].x, fontdesc->vmtx[i].y, fontdesc->vmtx[i].w); |
||
| 1160 | printf("\t}\n"); |
||
| 1161 | } |
||
| 1162 | } |