| 1 | /***************************************************************************/ |
| 2 | /* */ |
| 3 | /* ttcmap.c */ |
| 4 | /* */ |
| 5 | /* TrueType character mapping table (cmap) support (body). */ |
| 6 | /* */ |
| 7 | /* Copyright 1996-2000 by */ |
| 8 | /* David Turner, Robert Wilhelm, and Werner Lemberg. */ |
| 9 | /* */ |
| 10 | /* This file is part of the FreeType project, and may only be used, */ |
| 11 | /* modified, and distributed under the terms of the FreeType project */ |
| 12 | /* license, LICENSE.TXT. By continuing to use, modify, or distribute */ |
| 13 | /* this file you indicate that you have read the license and */ |
| 14 | /* understand and accept it fully. */ |
| 15 | /* */ |
| 16 | /***************************************************************************/ |
| 17 | |
| 18 | |
| 19 | #include <freetype/internal/ftdebug.h> |
| 20 | #include <freetype/internal/tterrors.h> |
| 21 | |
| 22 | |
| 23 | #ifdef FT_FLAT_COMPILE |
| 24 | |
| 25 | #include "ttload.h" |
| 26 | #include "ttcmap.h" |
| 27 | |
| 28 | #else |
| 29 | |
| 30 | #include <sfnt/ttload.h> |
| 31 | #include <sfnt/ttcmap.h> |
| 32 | |
| 33 | #endif |
| 34 | |
| 35 | |
| 36 | /*************************************************************************/ |
| 37 | /* */ |
| 38 | /* The macro FT_COMPONENT is used in trace mode. It is an implicit */ |
| 39 | /* parameter of the FT_TRACE() and FT_ERROR() macros, used to print/log */ |
| 40 | /* messages during execution. */ |
| 41 | /* */ |
| 42 | #undef FT_COMPONENT |
| 43 | #define FT_COMPONENT trace_ttcmap |
| 44 | |
| 45 | |
| 46 | static FT_UInt code_to_index0( TT_CMapTable* charmap, |
| 47 | FT_ULong char_code ); |
| 48 | static FT_UInt code_to_index2( TT_CMapTable* charmap, |
| 49 | FT_ULong char_code ); |
| 50 | static FT_UInt code_to_index4( TT_CMapTable* charmap, |
| 51 | FT_ULong char_code ); |
| 52 | static FT_UInt code_to_index6( TT_CMapTable* charmap, |
| 53 | FT_ULong char_code ); |
| 54 | |
| 55 | |
| 56 | /*************************************************************************/ |
| 57 | /* */ |
| 58 | /* <Function> */ |
| 59 | /* TT_CharMap_Load */ |
| 60 | /* */ |
| 61 | /* <Description> */ |
| 62 | /* Loads a given TrueType character map into memory. */ |
| 63 | /* */ |
| 64 | /* <Input> */ |
| 65 | /* face :: A handle to the parent face object. */ |
| 66 | /* stream :: A handle to the current stream object. */ |
| 67 | /* */ |
| 68 | /* <InOut> */ |
| 69 | /* table :: A pointer to a cmap object. */ |
| 70 | /* */ |
| 71 | /* <Return> */ |
| 72 | /* FreeType error code. 0 means success. */ |
| 73 | /* */ |
| 74 | /* <Note> */ |
| 75 | /* The function assumes that the stream is already in use (i.e., */ |
| 76 | /* opened). In case of error, all partially allocated tables are */ |
| 77 | /* released. */ |
| 78 | /* */ |
| 79 | LOCAL_FUNC |
| 80 | FT_Error TT_CharMap_Load( TT_Face face, |
| 81 | TT_CMapTable* cmap, |
| 82 | FT_Stream stream ) |
| 83 | { |
| 84 | FT_Error error; |
| 85 | FT_Memory memory; |
| 86 | FT_UShort num_SH, num_Seg, i; |
| 87 | |
| 88 | FT_UShort u, l; |
| 89 | |
| 90 | TT_CMap0* cmap0; |
| 91 | TT_CMap2* cmap2; |
| 92 | TT_CMap4* cmap4; |
| 93 | TT_CMap6* cmap6; |
| 94 | |
| 95 | TT_CMap2SubHeader* cmap2sub; |
| 96 | TT_CMap4Segment* segments; |
| 97 | |
| 98 | |
| 99 | if ( cmap->loaded ) |
| 100 | return TT_Err_Ok; |
| 101 | |
| 102 | memory = stream->memory; |
| 103 | |
| 104 | if ( FILE_Seek( cmap->offset ) ) |
| 105 | return error; |
| 106 | |
| 107 | switch ( cmap->format ) |
| 108 | { |
| 109 | case 0: |
| 110 | cmap0 = &cmap->c.cmap0; |
| 111 | |
| 112 | if ( ALLOC( cmap0->glyphIdArray, 256L ) || |
| 113 | FILE_Read( cmap0->glyphIdArray, 256L ) ) |
| 114 | goto Fail; |
| 115 | |
| 116 | cmap->get_index = code_to_index0; |
| 117 | break; |
| 118 | |
| 119 | case 2: |
| 120 | num_SH = 0; |
| 121 | cmap2 = &cmap->c.cmap2; |
| 122 | |
| 123 | /* allocate subheader keys */ |
| 124 | |
| 125 | if ( ALLOC_ARRAY( cmap2->subHeaderKeys, 256, FT_UShort ) || |
| 126 | ACCESS_Frame( 512L ) ) |
| 127 | goto Fail; |
| 128 | |
| 129 | for ( i = 0; i < 256; i++ ) |
| 130 | { |
| 131 | u = GET_UShort() / 8; |
| 132 | cmap2->subHeaderKeys[i] = u; |
| 133 | |
| 134 | if ( num_SH < u ) |
| 135 | num_SH = u; |
| 136 | } |
| 137 | |
| 138 | FORGET_Frame(); |
| 139 | |
| 140 | /* load subheaders */ |
| 141 | |
| 142 | cmap2->numGlyphId = l = |
| 143 | ( ( cmap->length - 2L * ( 256 + 3 ) - num_SH * 8L ) & 0xFFFF ) / 2; |
| 144 | |
| 145 | if ( ALLOC_ARRAY( cmap2->subHeaders, |
| 146 | num_SH + 1, |
| 147 | TT_CMap2SubHeader ) || |
| 148 | ACCESS_Frame( ( num_SH + 1 ) * 8L ) ) |
| 149 | goto Fail; |
| 150 | |
| 151 | cmap2sub = cmap2->subHeaders; |
| 152 | |
| 153 | for ( i = 0; i <= num_SH; i++ ) |
| 154 | { |
| 155 | cmap2sub->firstCode = GET_UShort(); |
| 156 | cmap2sub->entryCount = GET_UShort(); |
| 157 | cmap2sub->idDelta = GET_Short(); |
| 158 | /* we apply the location offset immediately */ |
| 159 | cmap2sub->idRangeOffset = GET_UShort() - ( num_SH - i ) * 8 - 2; |
| 160 | |
| 161 | cmap2sub++; |
| 162 | } |
| 163 | |
| 164 | FORGET_Frame(); |
| 165 | |
| 166 | /* load glyph IDs */ |
| 167 | |
| 168 | if ( ALLOC_ARRAY( cmap2->glyphIdArray, l, FT_UShort ) || |
| 169 | ACCESS_Frame( l * 2L ) ) |
| 170 | goto Fail; |
| 171 | |
| 172 | for ( i = 0; i < l; i++ ) |
| 173 | cmap2->glyphIdArray[i] = GET_UShort(); |
| 174 | |
| 175 | FORGET_Frame(); |
| 176 | |
| 177 | cmap->get_index = code_to_index2; |
| 178 | break; |
| 179 | |
| 180 | case 4: |
| 181 | cmap4 = &cmap->c.cmap4; |
| 182 | |
| 183 | /* load header */ |
| 184 | |
| 185 | if ( ACCESS_Frame( 8L ) ) |
| 186 | goto Fail; |
| 187 | |
| 188 | cmap4->segCountX2 = GET_UShort(); |
| 189 | cmap4->searchRange = GET_UShort(); |
| 190 | cmap4->entrySelector = GET_UShort(); |
| 191 | cmap4->rangeShift = GET_UShort(); |
| 192 | |
| 193 | num_Seg = cmap4->segCountX2 / 2; |
| 194 | |
| 195 | FORGET_Frame(); |
| 196 | |
| 197 | /* load segments */ |
| 198 | |
| 199 | if ( ALLOC_ARRAY( cmap4->segments, |
| 200 | num_Seg, |
| 201 | TT_CMap4Segment ) || |
| 202 | ACCESS_Frame( ( num_Seg * 4 + 1 ) * 2L ) ) |
| 203 | goto Fail; |
| 204 | |
| 205 | segments = cmap4->segments; |
| 206 | |
| 207 | for ( i = 0; i < num_Seg; i++ ) |
| 208 | segments[i].endCount = GET_UShort(); |
| 209 | |
| 210 | (void)GET_UShort(); |
| 211 | |
| 212 | for ( i = 0; i < num_Seg; i++ ) |
| 213 | segments[i].startCount = GET_UShort(); |
| 214 | |
| 215 | for ( i = 0; i < num_Seg; i++ ) |
| 216 | segments[i].idDelta = GET_Short(); |
| 217 | |
| 218 | for ( i = 0; i < num_Seg; i++ ) |
| 219 | segments[i].idRangeOffset = GET_UShort(); |
| 220 | |
| 221 | FORGET_Frame(); |
| 222 | |
| 223 | cmap4->numGlyphId = l = |
| 224 | ( ( cmap->length - ( 16L + 8L * num_Seg ) ) & 0xFFFF ) / 2; |
| 225 | |
| 226 | /* load IDs */ |
| 227 | |
| 228 | if ( ALLOC_ARRAY( cmap4->glyphIdArray, l, FT_UShort ) || |
| 229 | ACCESS_Frame( l * 2L ) ) |
| 230 | goto Fail; |
| 231 | |
| 232 | for ( i = 0; i < l; i++ ) |
| 233 | cmap4->glyphIdArray[i] = GET_UShort(); |
| 234 | |
| 235 | FORGET_Frame(); |
| 236 | |
| 237 | cmap->get_index = code_to_index4; |
| 238 | |
| 239 | cmap4->last_segment = cmap4->segments; |
| 240 | break; |
| 241 | |
| 242 | case 6: |
| 243 | cmap6 = &cmap->c.cmap6; |
| 244 | |
| 245 | if ( ACCESS_Frame( 4L ) ) |
| 246 | goto Fail; |
| 247 | |
| 248 | cmap6->firstCode = GET_UShort(); |
| 249 | cmap6->entryCount = GET_UShort(); |
| 250 | |
| 251 | FORGET_Frame(); |
| 252 | |
| 253 | l = cmap6->entryCount; |
| 254 | |
| 255 | if ( ALLOC_ARRAY( cmap6->glyphIdArray, |
| 256 | cmap6->entryCount, |
| 257 | FT_Short ) || |
| 258 | ACCESS_Frame( l * 2L ) ) |
| 259 | goto Fail; |
| 260 | |
| 261 | for ( i = 0; i < l; i++ ) |
| 262 | cmap6->glyphIdArray[i] = GET_UShort(); |
| 263 | |
| 264 | FORGET_Frame(); |
| 265 | cmap->get_index = code_to_index6; |
| 266 | break; |
| 267 | |
| 268 | default: /* corrupt character mapping table */ |
| 269 | return TT_Err_Invalid_CharMap_Format; |
| 270 | |
| 271 | } |
| 272 | |
| 273 | return TT_Err_Ok; |
| 274 | |
| 275 | Fail: |
| 276 | TT_CharMap_Free( face, cmap ); |
| 277 | return error; |
| 278 | } |
| 279 | |
| 280 | |
| 281 | /*************************************************************************/ |
| 282 | /* */ |
| 283 | /* <Function> */ |
| 284 | /* TT_CharMap_Free */ |
| 285 | /* */ |
| 286 | /* <Description> */ |
| 287 | /* Destroys a character mapping table. */ |
| 288 | /* */ |
| 289 | /* <Input> */ |
| 290 | /* face :: A handle to the parent face object. */ |
| 291 | /* cmap :: A handle to a cmap object. */ |
| 292 | /* */ |
| 293 | /* <Return> */ |
| 294 | /* FreeType error code. 0 means success. */ |
| 295 | /* */ |
| 296 | LOCAL_FUNC |
| 297 | FT_Error TT_CharMap_Free( TT_Face face, |
| 298 | TT_CMapTable* cmap ) |
| 299 | { |
| 300 | FT_Memory memory; |
| 301 | |
| 302 | |
| 303 | if ( !cmap ) |
| 304 | return TT_Err_Ok; |
| 305 | |
| 306 | memory = face->root.driver->root.memory; |
| 307 | |
| 308 | switch ( cmap->format ) |
| 309 | { |
| 310 | case 0: |
| 311 | FREE( cmap->c.cmap0.glyphIdArray ); |
| 312 | break; |
| 313 | |
| 314 | case 2: |
| 315 | FREE( cmap->c.cmap2.subHeaderKeys ); |
| 316 | FREE( cmap->c.cmap2.subHeaders ); |
| 317 | FREE( cmap->c.cmap2.glyphIdArray ); |
| 318 | break; |
| 319 | |
| 320 | case 4: |
| 321 | FREE( cmap->c.cmap4.segments ); |
| 322 | FREE( cmap->c.cmap4.glyphIdArray ); |
| 323 | cmap->c.cmap4.segCountX2 = 0; |
| 324 | break; |
| 325 | |
| 326 | case 6: |
| 327 | FREE( cmap->c.cmap6.glyphIdArray ); |
| 328 | cmap->c.cmap6.entryCount = 0; |
| 329 | break; |
| 330 | |
| 331 | default: |
| 332 | /* invalid table format, do nothing */ |
| 333 | ; |
| 334 | } |
| 335 | |
| 336 | cmap->loaded = FALSE; |
| 337 | return TT_Err_Ok; |
| 338 | } |
| 339 | |
| 340 | |
| 341 | /*************************************************************************/ |
| 342 | /* */ |
| 343 | /* <Function> */ |
| 344 | /* code_to_index0 */ |
| 345 | /* */ |
| 346 | /* <Description> */ |
| 347 | /* Converts the character code into a glyph index. Uses format 0. */ |
| 348 | /* `charCode' must be in the range 0x00-0xFF (otherwise 0 is */ |
| 349 | /* returned). */ |
| 350 | /* */ |
| 351 | /* <Input> */ |
| 352 | /* charCode :: The wanted character code. */ |
| 353 | /* cmap0 :: A pointer to a cmap table in format 0. */ |
| 354 | /* */ |
| 355 | /* <Return> */ |
| 356 | /* Glyph index into the glyphs array. 0 if the glyph does not exist. */ |
| 357 | /* */ |
| 358 | static |
| 359 | FT_UInt code_to_index0( TT_CMapTable* cmap, |
| 360 | FT_ULong charCode ) |
| 361 | { |
| 362 | TT_CMap0* cmap0 = &cmap->c.cmap0; |
| 363 | |
| 364 | |
| 365 | return ( charCode <= 0xFF ? cmap0->glyphIdArray[charCode] : 0 ); |
| 366 | } |
| 367 | |
| 368 | |
| 369 | /*************************************************************************/ |
| 370 | /* */ |
| 371 | /* <Function> */ |
| 372 | /* code_to_index2 */ |
| 373 | /* */ |
| 374 | /* <Description> */ |
| 375 | /* Converts the character code into a glyph index. Uses format 2. */ |
| 376 | /* */ |
| 377 | /* <Input> */ |
| 378 | /* charCode :: The wanted character code. */ |
| 379 | /* cmap2 :: A pointer to a cmap table in format 2. */ |
| 380 | /* */ |
| 381 | /* <Return> */ |
| 382 | /* Glyph index into the glyphs array. 0 if the glyph does not exist. */ |
| 383 | /* */ |
| 384 | static |
| 385 | FT_UInt code_to_index2( TT_CMapTable* cmap, |
| 386 | FT_ULong charCode ) |
| 387 | { |
| 388 | FT_UInt result, index1, offset; |
| 389 | FT_UInt char_lo; |
| 390 | FT_ULong char_hi; |
| 391 | TT_CMap2SubHeader* sh2; |
| 392 | TT_CMap2* cmap2; |
| 393 | |
| 394 | |
| 395 | cmap2 = &cmap->c.cmap2; |
| 396 | result = 0; |
| 397 | char_lo = (FT_UInt)( charCode & 0xFF ); |
| 398 | char_hi = charCode >> 8; |
| 399 | |
| 400 | if ( char_hi == 0 ) |
| 401 | { |
| 402 | /* an 8-bit character code -- we use the subHeader 0 in this case */ |
| 403 | /* to test whether the character code is in the charmap */ |
| 404 | if ( cmap2->subHeaderKeys[char_lo] == 0 ) |
| 405 | result = cmap2->glyphIdArray[char_lo]; |
| 406 | } |
| 407 | else |
| 408 | { |
| 409 | /* a 16-bit character code */ |
| 410 | index1 = cmap2->subHeaderKeys[char_hi & 0xFF]; |
| 411 | if ( index1 ) |
| 412 | { |
| 413 | sh2 = cmap2->subHeaders + index1; |
| 414 | char_lo -= sh2->firstCode; |
| 415 | |
| 416 | if ( char_lo < sh2->entryCount ) |
| 417 | { |
| 418 | offset = sh2->idRangeOffset / 2 + char_lo; |
| 419 | if ( offset < cmap2->numGlyphId ) |
| 420 | { |
| 421 | result = cmap2->glyphIdArray[offset]; |
| 422 | if ( result ) |
| 423 | result = ( result + sh2->idDelta ) & 0xFFFF; |
| 424 | } |
| 425 | } |
| 426 | } |
| 427 | } |
| 428 | |
| 429 | return result; |
| 430 | } |
| 431 | |
| 432 | |
| 433 | /*************************************************************************/ |
| 434 | /* */ |
| 435 | /* <Function> */ |
| 436 | /* code_to_index4 */ |
| 437 | /* */ |
| 438 | /* <Description> */ |
| 439 | /* Converts the character code into a glyph index. Uses format 4. */ |
| 440 | /* */ |
| 441 | /* <Input> */ |
| 442 | /* charCode :: The wanted character code. */ |
| 443 | /* cmap4 :: A pointer to a cmap table in format 4. */ |
| 444 | /* */ |
| 445 | /* <Return> */ |
| 446 | /* Glyph index into the glyphs array. 0 if the glyph does not exist. */ |
| 447 | /* */ |
| 448 | static |
| 449 | FT_UInt code_to_index4( TT_CMapTable* cmap, |
| 450 | FT_ULong charCode ) |
| 451 | { |
| 452 | FT_UInt result, index1, segCount; |
| 453 | TT_CMap4* cmap4; |
| 454 | TT_CMap4Segment *seg4, *limit; |
| 455 | |
| 456 | |
| 457 | cmap4 = &cmap->c.cmap4; |
| 458 | result = 0; |
| 459 | segCount = cmap4->segCountX2 / 2; |
| 460 | seg4 = cmap4->segments; |
| 461 | limit = seg4 + segCount; |
| 462 | |
| 463 | /* check against the last segment */ |
| 464 | seg4 = cmap4->last_segment; |
| 465 | |
| 466 | /* the following is equivalent to performing two tests, as in */ |
| 467 | /* */ |
| 468 | /* if ( charCode >= seg4->startCount && charCode <= seg4->endCount ) */ |
| 469 | /* */ |
| 470 | /* Yes, that's a bit strange, but it's faster, and the idea behind */ |
| 471 | /* the cache is to significantly speed up charcode to glyph index */ |
| 472 | /* conversion. */ |
| 473 | |
| 474 | if ( (FT_ULong)(charCode - seg4->startCount) < |
| 475 | (FT_ULong)(seg4->endCount - seg4->startCount) ) |
| 476 | goto Found; |
| 477 | |
| 478 | for ( seg4 = cmap4->segments; seg4 < limit; seg4++ ) |
| 479 | { |
| 480 | /* the ranges are sorted in increasing order. If we are out of */ |
| 481 | /* the range here, the char code isn't in the charmap, so exit. */ |
| 482 | |
| 483 | if ( charCode > seg4->endCount ) |
| 484 | continue; |
| 485 | |
| 486 | if ( charCode >= seg4->startCount ) |
| 487 | goto Found; |
| 488 | } |
| 489 | return 0; |
| 490 | |
| 491 | Found: |
| 492 | cmap4->last_segment = seg4; |
| 493 | |
| 494 | /* if the idRangeOffset is 0, we can compute the glyph index */ |
| 495 | /* directly */ |
| 496 | |
| 497 | if ( seg4->idRangeOffset == 0 ) |
| 498 | result = ( charCode + seg4->idDelta ) & 0xFFFF; |
| 499 | else |
| 500 | { |
| 501 | /* otherwise, we must use the glyphIdArray to do it */ |
| 502 | index1 = seg4->idRangeOffset / 2 |
| 503 | + ( charCode - seg4->startCount ) |
| 504 | + ( seg4 - cmap4->segments ) |
| 505 | - segCount; |
| 506 | |
| 507 | if ( index1 < cmap4->numGlyphId && |
| 508 | cmap4->glyphIdArray[index1] != 0 ) |
| 509 | result = ( cmap4->glyphIdArray[index1] + seg4->idDelta ) & 0xFFFF; |
| 510 | } |
| 511 | |
| 512 | return result; |
| 513 | } |
| 514 | |
| 515 | |
| 516 | /*************************************************************************/ |
| 517 | /* */ |
| 518 | /* <Function> */ |
| 519 | /* code_to_index6 */ |
| 520 | /* */ |
| 521 | /* <Description> */ |
| 522 | /* Converts the character code into a glyph index. Uses format 6. */ |
| 523 | /* */ |
| 524 | /* <Input> */ |
| 525 | /* charCode :: The wanted character code. */ |
| 526 | /* cmap6 :: A pointer to a cmap table in format 6. */ |
| 527 | /* */ |
| 528 | /* <Return> */ |
| 529 | /* Glyph index into the glyphs array. 0 if the glyph does not exist. */ |
| 530 | /* */ |
| 531 | static |
| 532 | FT_UInt code_to_index6( TT_CMapTable* cmap, |
| 533 | FT_ULong charCode ) |
| 534 | { |
| 535 | TT_CMap6* cmap6; |
| 536 | FT_UInt result = 0; |
| 537 | |
| 538 | |
| 539 | cmap6 = &cmap->c.cmap6; |
| 540 | result = 0; |
| 541 | charCode -= cmap6->firstCode; |
| 542 | |
| 543 | if ( charCode < cmap6->entryCount ) |
| 544 | result = cmap6->glyphIdArray[charCode]; |
| 545 | |
| 546 | return result; |
| 547 | } |
| 548 | |
| 549 | |
| 550 | /* END */ |