| 1 | /* |
| 2 | * Copyright © 2012,2013 Google, Inc. |
| 3 | * |
| 4 | * This is part of HarfBuzz, a text shaping library. |
| 5 | * |
| 6 | * Permission is hereby granted, without written agreement and without |
| 7 | * license or royalty fees, to use, copy, modify, and distribute this |
| 8 | * software and its documentation for any purpose, provided that the |
| 9 | * above copyright notice and the following two paragraphs appear in |
| 10 | * all copies of this software. |
| 11 | * |
| 12 | * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR |
| 13 | * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES |
| 14 | * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN |
| 15 | * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH |
| 16 | * DAMAGE. |
| 17 | * |
| 18 | * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, |
| 19 | * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND |
| 20 | * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS |
| 21 | * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO |
| 22 | * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS. |
| 23 | * |
| 24 | * Google Author(s): Behdad Esfahbod |
| 25 | */ |
| 26 | |
| 27 | #include "hb.hh" |
| 28 | |
| 29 | #ifndef HB_NO_BUFFER_SERIALIZE |
| 30 | |
| 31 | #include "hb-buffer.hh" |
| 32 | |
| 33 | |
| 34 | static const char *_hb_buffer_serialize_formats[] = { |
| 35 | "text" , |
| 36 | "json" , |
| 37 | nullptr |
| 38 | }; |
| 39 | |
| 40 | /** |
| 41 | * hb_buffer_serialize_list_formats: |
| 42 | * |
| 43 | * Returns a list of supported buffer serialization formats. |
| 44 | * |
| 45 | * Return value: (transfer none): |
| 46 | * A string array of buffer serialization formats. Should not be freed. |
| 47 | * |
| 48 | * Since: 0.9.7 |
| 49 | **/ |
| 50 | const char ** |
| 51 | hb_buffer_serialize_list_formats () |
| 52 | { |
| 53 | return _hb_buffer_serialize_formats; |
| 54 | } |
| 55 | |
| 56 | /** |
| 57 | * hb_buffer_serialize_format_from_string: |
| 58 | * @str: (array length=len) (element-type uint8_t): a string to parse |
| 59 | * @len: length of @str, or -1 if string is `NULL` terminated |
| 60 | * |
| 61 | * Parses a string into an #hb_buffer_serialize_format_t. Does not check if |
| 62 | * @str is a valid buffer serialization format, use |
| 63 | * hb_buffer_serialize_list_formats() to get the list of supported formats. |
| 64 | * |
| 65 | * Return value: |
| 66 | * The parsed #hb_buffer_serialize_format_t. |
| 67 | * |
| 68 | * Since: 0.9.7 |
| 69 | **/ |
| 70 | hb_buffer_serialize_format_t |
| 71 | hb_buffer_serialize_format_from_string (const char *str, int len) |
| 72 | { |
| 73 | /* Upper-case it. */ |
| 74 | return (hb_buffer_serialize_format_t) (hb_tag_from_string (str, len) & ~0x20202020u); |
| 75 | } |
| 76 | |
| 77 | /** |
| 78 | * hb_buffer_serialize_format_to_string: |
| 79 | * @format: an #hb_buffer_serialize_format_t to convert. |
| 80 | * |
| 81 | * Converts @format to the string corresponding it, or `NULL` if it is not a valid |
| 82 | * #hb_buffer_serialize_format_t. |
| 83 | * |
| 84 | * Return value: (transfer none): |
| 85 | * A `NULL` terminated string corresponding to @format. Should not be freed. |
| 86 | * |
| 87 | * Since: 0.9.7 |
| 88 | **/ |
| 89 | const char * |
| 90 | hb_buffer_serialize_format_to_string (hb_buffer_serialize_format_t format) |
| 91 | { |
| 92 | switch ((unsigned) format) |
| 93 | { |
| 94 | case HB_BUFFER_SERIALIZE_FORMAT_TEXT: return _hb_buffer_serialize_formats[0]; |
| 95 | case HB_BUFFER_SERIALIZE_FORMAT_JSON: return _hb_buffer_serialize_formats[1]; |
| 96 | default: |
| 97 | case HB_BUFFER_SERIALIZE_FORMAT_INVALID: return nullptr; |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | static unsigned int |
| 102 | _hb_buffer_serialize_glyphs_json (hb_buffer_t *buffer, |
| 103 | unsigned int start, |
| 104 | unsigned int end, |
| 105 | char *buf, |
| 106 | unsigned int buf_size, |
| 107 | unsigned int *buf_consumed, |
| 108 | hb_font_t *font, |
| 109 | hb_buffer_serialize_flags_t flags) |
| 110 | { |
| 111 | hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr); |
| 112 | hb_glyph_position_t *pos = (flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS) ? |
| 113 | nullptr : hb_buffer_get_glyph_positions (buffer, nullptr); |
| 114 | |
| 115 | *buf_consumed = 0; |
| 116 | hb_position_t x = 0, y = 0; |
| 117 | for (unsigned int i = start; i < end; i++) |
| 118 | { |
| 119 | char b[1024]; |
| 120 | char *p = b; |
| 121 | |
| 122 | /* In the following code, we know b is large enough that no overflow can happen. */ |
| 123 | |
| 124 | #define APPEND(s) HB_STMT_START { strcpy (p, s); p += strlen (s); } HB_STMT_END |
| 125 | |
| 126 | if (i) |
| 127 | *p++ = ','; |
| 128 | else |
| 129 | *p++ = '['; |
| 130 | |
| 131 | *p++ = '{'; |
| 132 | |
| 133 | APPEND ("\"g\":" ); |
| 134 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES)) |
| 135 | { |
| 136 | char g[128]; |
| 137 | hb_font_glyph_to_string (font, info[i].codepoint, g, sizeof (g)); |
| 138 | *p++ = '"'; |
| 139 | for (char *q = g; *q; q++) |
| 140 | { |
| 141 | if (unlikely (*q == '"' || *q == '\\')) |
| 142 | *p++ = '\\'; |
| 143 | *p++ = *q; |
| 144 | } |
| 145 | *p++ = '"'; |
| 146 | } |
| 147 | else |
| 148 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%u" , info[i].codepoint)); |
| 149 | |
| 150 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) { |
| 151 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"cl\":%u" , info[i].cluster)); |
| 152 | } |
| 153 | |
| 154 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS)) |
| 155 | { |
| 156 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"dx\":%d,\"dy\":%d" , |
| 157 | x+pos[i].x_offset, y+pos[i].y_offset)); |
| 158 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_ADVANCES)) |
| 159 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"ax\":%d,\"ay\":%d" , |
| 160 | pos[i].x_advance, pos[i].y_advance)); |
| 161 | } |
| 162 | |
| 163 | if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_FLAGS) |
| 164 | { |
| 165 | if (info[i].mask & HB_GLYPH_FLAG_DEFINED) |
| 166 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"fl\":%u" , info[i].mask & HB_GLYPH_FLAG_DEFINED)); |
| 167 | } |
| 168 | |
| 169 | if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS) |
| 170 | { |
| 171 | hb_glyph_extents_t extents; |
| 172 | hb_font_get_glyph_extents(font, info[i].codepoint, &extents); |
| 173 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"xb\":%d,\"yb\":%d" , |
| 174 | extents.x_bearing, extents.y_bearing)); |
| 175 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"w\":%d,\"h\":%d" , |
| 176 | extents.width, extents.height)); |
| 177 | } |
| 178 | |
| 179 | *p++ = '}'; |
| 180 | if (i == end-1) |
| 181 | *p++ = ']'; |
| 182 | |
| 183 | unsigned int l = p - b; |
| 184 | if (buf_size > l) |
| 185 | { |
| 186 | hb_memcpy (buf, b, l); |
| 187 | buf += l; |
| 188 | buf_size -= l; |
| 189 | *buf_consumed += l; |
| 190 | *buf = '\0'; |
| 191 | } else |
| 192 | return i - start; |
| 193 | |
| 194 | if (pos && (flags & HB_BUFFER_SERIALIZE_FLAG_NO_ADVANCES)) |
| 195 | { |
| 196 | x += pos[i].x_advance; |
| 197 | y += pos[i].y_advance; |
| 198 | } |
| 199 | } |
| 200 | |
| 201 | return end - start; |
| 202 | } |
| 203 | |
| 204 | static unsigned int |
| 205 | _hb_buffer_serialize_unicode_json (hb_buffer_t *buffer, |
| 206 | unsigned int start, |
| 207 | unsigned int end, |
| 208 | char *buf, |
| 209 | unsigned int buf_size, |
| 210 | unsigned int *buf_consumed, |
| 211 | hb_buffer_serialize_flags_t flags) |
| 212 | { |
| 213 | hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr); |
| 214 | |
| 215 | *buf_consumed = 0; |
| 216 | for (unsigned int i = start; i < end; i++) |
| 217 | { |
| 218 | char b[1024]; |
| 219 | char *p = b; |
| 220 | |
| 221 | if (i) |
| 222 | *p++ = ','; |
| 223 | else |
| 224 | *p++ = '['; |
| 225 | |
| 226 | *p++ = '{'; |
| 227 | |
| 228 | APPEND ("\"u\":" ); |
| 229 | |
| 230 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%u" , info[i].codepoint)); |
| 231 | |
| 232 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) { |
| 233 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"cl\":%u" , info[i].cluster)); |
| 234 | } |
| 235 | |
| 236 | *p++ = '}'; |
| 237 | |
| 238 | if (i == end-1) |
| 239 | *p++ = ']'; |
| 240 | |
| 241 | unsigned int l = p - b; |
| 242 | if (buf_size > l) |
| 243 | { |
| 244 | hb_memcpy (buf, b, l); |
| 245 | buf += l; |
| 246 | buf_size -= l; |
| 247 | *buf_consumed += l; |
| 248 | *buf = '\0'; |
| 249 | } else |
| 250 | return i - start; |
| 251 | |
| 252 | } |
| 253 | |
| 254 | return end - start; |
| 255 | } |
| 256 | |
| 257 | static unsigned int |
| 258 | _hb_buffer_serialize_glyphs_text (hb_buffer_t *buffer, |
| 259 | unsigned int start, |
| 260 | unsigned int end, |
| 261 | char *buf, |
| 262 | unsigned int buf_size, |
| 263 | unsigned int *buf_consumed, |
| 264 | hb_font_t *font, |
| 265 | hb_buffer_serialize_flags_t flags) |
| 266 | { |
| 267 | hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr); |
| 268 | hb_glyph_position_t *pos = (flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS) ? |
| 269 | nullptr : hb_buffer_get_glyph_positions (buffer, nullptr); |
| 270 | |
| 271 | *buf_consumed = 0; |
| 272 | hb_position_t x = 0, y = 0; |
| 273 | for (unsigned int i = start; i < end; i++) |
| 274 | { |
| 275 | char b[1024]; |
| 276 | char *p = b; |
| 277 | |
| 278 | /* In the following code, we know b is large enough that no overflow can happen. */ |
| 279 | |
| 280 | if (i) |
| 281 | *p++ = '|'; |
| 282 | else |
| 283 | *p++ = '['; |
| 284 | |
| 285 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES)) |
| 286 | { |
| 287 | /* TODO Escape delimiters we use. */ |
| 288 | hb_font_glyph_to_string (font, info[i].codepoint, p, 128); |
| 289 | p += strlen (p); |
| 290 | } |
| 291 | else |
| 292 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%u" , info[i].codepoint)); |
| 293 | |
| 294 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) { |
| 295 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "=%u" , info[i].cluster)); |
| 296 | } |
| 297 | |
| 298 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS)) |
| 299 | { |
| 300 | if (x+pos[i].x_offset || y+pos[i].y_offset) |
| 301 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "@%d,%d" , x+pos[i].x_offset, y+pos[i].y_offset)); |
| 302 | |
| 303 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_ADVANCES)) |
| 304 | { |
| 305 | *p++ = '+'; |
| 306 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%d" , pos[i].x_advance)); |
| 307 | if (pos[i].y_advance) |
| 308 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",%d" , pos[i].y_advance)); |
| 309 | } |
| 310 | } |
| 311 | |
| 312 | if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_FLAGS) |
| 313 | { |
| 314 | if (info[i].mask & HB_GLYPH_FLAG_DEFINED) |
| 315 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "#%X" , info[i].mask &HB_GLYPH_FLAG_DEFINED)); |
| 316 | } |
| 317 | |
| 318 | if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS) |
| 319 | { |
| 320 | hb_glyph_extents_t extents; |
| 321 | hb_font_get_glyph_extents(font, info[i].codepoint, &extents); |
| 322 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "<%d,%d,%d,%d>" , extents.x_bearing, extents.y_bearing, extents.width, extents.height)); |
| 323 | } |
| 324 | |
| 325 | if (i == end-1) { |
| 326 | *p++ = ']'; |
| 327 | } |
| 328 | |
| 329 | unsigned int l = p - b; |
| 330 | if (buf_size > l) |
| 331 | { |
| 332 | hb_memcpy (buf, b, l); |
| 333 | buf += l; |
| 334 | buf_size -= l; |
| 335 | *buf_consumed += l; |
| 336 | *buf = '\0'; |
| 337 | } else |
| 338 | return i - start; |
| 339 | |
| 340 | if (pos && (flags & HB_BUFFER_SERIALIZE_FLAG_NO_ADVANCES)) |
| 341 | { |
| 342 | x += pos[i].x_advance; |
| 343 | y += pos[i].y_advance; |
| 344 | } |
| 345 | } |
| 346 | |
| 347 | return end - start; |
| 348 | } |
| 349 | |
| 350 | |
| 351 | static unsigned int |
| 352 | _hb_buffer_serialize_unicode_text (hb_buffer_t *buffer, |
| 353 | unsigned int start, |
| 354 | unsigned int end, |
| 355 | char *buf, |
| 356 | unsigned int buf_size, |
| 357 | unsigned int *buf_consumed, |
| 358 | hb_buffer_serialize_flags_t flags) |
| 359 | { |
| 360 | hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr); |
| 361 | *buf_consumed = 0; |
| 362 | for (unsigned int i = start; i < end; i++) |
| 363 | { |
| 364 | char b[1024]; |
| 365 | char *p = b; |
| 366 | |
| 367 | if (i) |
| 368 | *p++ = '|'; |
| 369 | else |
| 370 | *p++ = '<'; |
| 371 | |
| 372 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "U+%04X" , info[i].codepoint)); |
| 373 | |
| 374 | if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) { |
| 375 | p += hb_max (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "=%u" , info[i].cluster)); |
| 376 | } |
| 377 | |
| 378 | if (i == end-1) |
| 379 | *p++ = '>'; |
| 380 | |
| 381 | unsigned int l = p - b; |
| 382 | if (buf_size > l) |
| 383 | { |
| 384 | hb_memcpy (buf, b, l); |
| 385 | buf += l; |
| 386 | buf_size -= l; |
| 387 | *buf_consumed += l; |
| 388 | *buf = '\0'; |
| 389 | } else |
| 390 | return i - start; |
| 391 | } |
| 392 | return end - start; |
| 393 | } |
| 394 | |
| 395 | /** |
| 396 | * hb_buffer_serialize_glyphs: |
| 397 | * @buffer: an #hb_buffer_t buffer. |
| 398 | * @start: the first item in @buffer to serialize. |
| 399 | * @end: the last item in @buffer to serialize. |
| 400 | * @buf: (out) (array length=buf_size) (element-type uint8_t): output string to |
| 401 | * write serialized buffer into. |
| 402 | * @buf_size: the size of @buf. |
| 403 | * @buf_consumed: (out) (optional): if not `NULL`, will be set to the number of bytes written into @buf. |
| 404 | * @font: (nullable): the #hb_font_t used to shape this buffer, needed to |
| 405 | * read glyph names and extents. If `NULL`, an empty font will be used. |
| 406 | * @format: the #hb_buffer_serialize_format_t to use for formatting the output. |
| 407 | * @flags: the #hb_buffer_serialize_flags_t that control what glyph properties |
| 408 | * to serialize. |
| 409 | * |
| 410 | * Serializes @buffer into a textual representation of its glyph content, |
| 411 | * useful for showing the contents of the buffer, for example during debugging. |
| 412 | * There are currently two supported serialization formats: |
| 413 | * |
| 414 | * ## text |
| 415 | * A human-readable, plain text format. |
| 416 | * The serialized glyphs will look something like: |
| 417 | * |
| 418 | * ``` |
| 419 | * [uni0651=0@518,0+0|uni0628=0+1897] |
| 420 | * ``` |
| 421 | * |
| 422 | * - The serialized glyphs are delimited with `[` and `]`. |
| 423 | * - Glyphs are separated with `|` |
| 424 | * - Each glyph starts with glyph name, or glyph index if |
| 425 | * #HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES flag is set. Then, |
| 426 | * - If #HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS is not set, `=` then #hb_glyph_info_t.cluster. |
| 427 | * - If #HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS is not set, the #hb_glyph_position_t in the format: |
| 428 | * - If both #hb_glyph_position_t.x_offset and #hb_glyph_position_t.y_offset are not 0, `@x_offset,y_offset`. Then, |
| 429 | * - `+x_advance`, then `,y_advance` if #hb_glyph_position_t.y_advance is not 0. Then, |
| 430 | * - If #HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS is set, the #hb_glyph_extents_t in the format `<x_bearing,y_bearing,width,height>` |
| 431 | * |
| 432 | * ## json |
| 433 | * A machine-readable, structured format. |
| 434 | * The serialized glyphs will look something like: |
| 435 | * |
| 436 | * ``` |
| 437 | * [{"g":"uni0651","cl":0,"dx":518,"dy":0,"ax":0,"ay":0}, |
| 438 | * {"g":"uni0628","cl":0,"dx":0,"dy":0,"ax":1897,"ay":0}] |
| 439 | * ``` |
| 440 | * |
| 441 | * Each glyph is a JSON object, with the following properties: |
| 442 | * - `g`: the glyph name or glyph index if |
| 443 | * #HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES flag is set. |
| 444 | * - `cl`: #hb_glyph_info_t.cluster if |
| 445 | * #HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS is not set. |
| 446 | * - `dx`,`dy`,`ax`,`ay`: #hb_glyph_position_t.x_offset, #hb_glyph_position_t.y_offset, |
| 447 | * #hb_glyph_position_t.x_advance and #hb_glyph_position_t.y_advance |
| 448 | * respectively, if #HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS is not set. |
| 449 | * - `xb`,`yb`,`w`,`h`: #hb_glyph_extents_t.x_bearing, #hb_glyph_extents_t.y_bearing, |
| 450 | * #hb_glyph_extents_t.width and #hb_glyph_extents_t.height respectively if |
| 451 | * #HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS is set. |
| 452 | * |
| 453 | * Return value: |
| 454 | * The number of serialized items. |
| 455 | * |
| 456 | * Since: 0.9.7 |
| 457 | **/ |
| 458 | unsigned int |
| 459 | hb_buffer_serialize_glyphs (hb_buffer_t *buffer, |
| 460 | unsigned int start, |
| 461 | unsigned int end, |
| 462 | char *buf, |
| 463 | unsigned int buf_size, |
| 464 | unsigned int *buf_consumed, |
| 465 | hb_font_t *font, |
| 466 | hb_buffer_serialize_format_t format, |
| 467 | hb_buffer_serialize_flags_t flags) |
| 468 | { |
| 469 | end = hb_clamp (end, start, buffer->len); |
| 470 | start = hb_min (start, end); |
| 471 | |
| 472 | unsigned int sconsumed; |
| 473 | if (!buf_consumed) |
| 474 | buf_consumed = &sconsumed; |
| 475 | *buf_consumed = 0; |
| 476 | if (buf_size) |
| 477 | *buf = '\0'; |
| 478 | |
| 479 | buffer->assert_glyphs (); |
| 480 | |
| 481 | if (!buffer->have_positions) |
| 482 | flags |= HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS; |
| 483 | |
| 484 | if (unlikely (start == end)) |
| 485 | return 0; |
| 486 | |
| 487 | if (!font) |
| 488 | font = hb_font_get_empty (); |
| 489 | |
| 490 | switch (format) |
| 491 | { |
| 492 | case HB_BUFFER_SERIALIZE_FORMAT_TEXT: |
| 493 | return _hb_buffer_serialize_glyphs_text (buffer, start, end, |
| 494 | buf, buf_size, buf_consumed, |
| 495 | font, flags); |
| 496 | |
| 497 | case HB_BUFFER_SERIALIZE_FORMAT_JSON: |
| 498 | return _hb_buffer_serialize_glyphs_json (buffer, start, end, |
| 499 | buf, buf_size, buf_consumed, |
| 500 | font, flags); |
| 501 | |
| 502 | default: |
| 503 | case HB_BUFFER_SERIALIZE_FORMAT_INVALID: |
| 504 | return 0; |
| 505 | |
| 506 | } |
| 507 | } |
| 508 | |
| 509 | /** |
| 510 | * hb_buffer_serialize_unicode: |
| 511 | * @buffer: an #hb_buffer_t buffer. |
| 512 | * @start: the first item in @buffer to serialize. |
| 513 | * @end: the last item in @buffer to serialize. |
| 514 | * @buf: (out) (array length=buf_size) (element-type uint8_t): output string to |
| 515 | * write serialized buffer into. |
| 516 | * @buf_size: the size of @buf. |
| 517 | * @buf_consumed: (out) (optional): if not `NULL`, will be set to the number of bytes written into @buf. |
| 518 | * @format: the #hb_buffer_serialize_format_t to use for formatting the output. |
| 519 | * @flags: the #hb_buffer_serialize_flags_t that control what glyph properties |
| 520 | * to serialize. |
| 521 | * |
| 522 | * Serializes @buffer into a textual representation of its content, |
| 523 | * when the buffer contains Unicode codepoints (i.e., before shaping). This is |
| 524 | * useful for showing the contents of the buffer, for example during debugging. |
| 525 | * There are currently two supported serialization formats: |
| 526 | * |
| 527 | * ## text |
| 528 | * A human-readable, plain text format. |
| 529 | * The serialized codepoints will look something like: |
| 530 | * |
| 531 | * ``` |
| 532 | * <U+0651=0|U+0628=1> |
| 533 | * ``` |
| 534 | * |
| 535 | * - Glyphs are separated with `|` |
| 536 | * - Unicode codepoints are expressed as zero-padded four (or more) |
| 537 | * digit hexadecimal numbers preceded by `U+` |
| 538 | * - If #HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS is not set, the cluster |
| 539 | * will be indicated with a `=` then #hb_glyph_info_t.cluster. |
| 540 | * |
| 541 | * ## json |
| 542 | * A machine-readable, structured format. |
| 543 | * The serialized codepoints will be a list of objects with the following |
| 544 | * properties: |
| 545 | * - `u`: the Unicode codepoint as a decimal integer |
| 546 | * - `cl`: #hb_glyph_info_t.cluster if |
| 547 | * #HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS is not set. |
| 548 | * |
| 549 | * For example: |
| 550 | * |
| 551 | * ``` |
| 552 | * [{u:1617,cl:0},{u:1576,cl:1}] |
| 553 | * ``` |
| 554 | * |
| 555 | * Return value: |
| 556 | * The number of serialized items. |
| 557 | * |
| 558 | * Since: 2.7.3 |
| 559 | **/ |
| 560 | unsigned int |
| 561 | hb_buffer_serialize_unicode (hb_buffer_t *buffer, |
| 562 | unsigned int start, |
| 563 | unsigned int end, |
| 564 | char *buf, |
| 565 | unsigned int buf_size, |
| 566 | unsigned int *buf_consumed, |
| 567 | hb_buffer_serialize_format_t format, |
| 568 | hb_buffer_serialize_flags_t flags) |
| 569 | { |
| 570 | end = hb_clamp (end, start, buffer->len); |
| 571 | start = hb_min (start, end); |
| 572 | |
| 573 | unsigned int sconsumed; |
| 574 | if (!buf_consumed) |
| 575 | buf_consumed = &sconsumed; |
| 576 | *buf_consumed = 0; |
| 577 | if (buf_size) |
| 578 | *buf = '\0'; |
| 579 | |
| 580 | buffer->assert_unicode (); |
| 581 | |
| 582 | if (unlikely (start == end)) |
| 583 | return 0; |
| 584 | |
| 585 | switch (format) |
| 586 | { |
| 587 | case HB_BUFFER_SERIALIZE_FORMAT_TEXT: |
| 588 | return _hb_buffer_serialize_unicode_text (buffer, start, end, |
| 589 | buf, buf_size, buf_consumed, flags); |
| 590 | |
| 591 | case HB_BUFFER_SERIALIZE_FORMAT_JSON: |
| 592 | return _hb_buffer_serialize_unicode_json (buffer, start, end, |
| 593 | buf, buf_size, buf_consumed, flags); |
| 594 | |
| 595 | default: |
| 596 | case HB_BUFFER_SERIALIZE_FORMAT_INVALID: |
| 597 | return 0; |
| 598 | |
| 599 | } |
| 600 | } |
| 601 | |
| 602 | static unsigned int |
| 603 | _hb_buffer_serialize_invalid (hb_buffer_t *buffer, |
| 604 | unsigned int start, |
| 605 | unsigned int end, |
| 606 | char *buf, |
| 607 | unsigned int buf_size, |
| 608 | unsigned int *buf_consumed, |
| 609 | hb_buffer_serialize_format_t format, |
| 610 | hb_buffer_serialize_flags_t flags) |
| 611 | { |
| 612 | assert (!buffer->len); |
| 613 | |
| 614 | unsigned int sconsumed; |
| 615 | if (!buf_consumed) |
| 616 | buf_consumed = &sconsumed; |
| 617 | if (buf_size < 3) |
| 618 | return 0; |
| 619 | if (format == HB_BUFFER_SERIALIZE_FORMAT_JSON) { |
| 620 | *buf++ = '['; |
| 621 | *buf++ = ']'; |
| 622 | *buf = '\0'; |
| 623 | } else if (format == HB_BUFFER_SERIALIZE_FORMAT_TEXT) { |
| 624 | *buf++ = '!'; |
| 625 | *buf++ = '!'; |
| 626 | *buf = '\0'; |
| 627 | } |
| 628 | *buf_consumed = 2; |
| 629 | return 0; |
| 630 | } |
| 631 | |
| 632 | /** |
| 633 | * hb_buffer_serialize: |
| 634 | * @buffer: an #hb_buffer_t buffer. |
| 635 | * @start: the first item in @buffer to serialize. |
| 636 | * @end: the last item in @buffer to serialize. |
| 637 | * @buf: (out) (array length=buf_size) (element-type uint8_t): output string to |
| 638 | * write serialized buffer into. |
| 639 | * @buf_size: the size of @buf. |
| 640 | * @buf_consumed: (out) (optional): if not `NULL`, will be set to the number of bytes written into @buf. |
| 641 | * @font: (nullable): the #hb_font_t used to shape this buffer, needed to |
| 642 | * read glyph names and extents. If `NULL`, an empty font will be used. |
| 643 | * @format: the #hb_buffer_serialize_format_t to use for formatting the output. |
| 644 | * @flags: the #hb_buffer_serialize_flags_t that control what glyph properties |
| 645 | * to serialize. |
| 646 | * |
| 647 | * Serializes @buffer into a textual representation of its content, whether |
| 648 | * Unicode codepoints or glyph identifiers and positioning information. This is |
| 649 | * useful for showing the contents of the buffer, for example during debugging. |
| 650 | * See the documentation of hb_buffer_serialize_unicode() and |
| 651 | * hb_buffer_serialize_glyphs() for a description of the output format. |
| 652 | * |
| 653 | * Return value: |
| 654 | * The number of serialized items. |
| 655 | * |
| 656 | * Since: 2.7.3 |
| 657 | **/ |
| 658 | unsigned int |
| 659 | hb_buffer_serialize (hb_buffer_t *buffer, |
| 660 | unsigned int start, |
| 661 | unsigned int end, |
| 662 | char *buf, |
| 663 | unsigned int buf_size, |
| 664 | unsigned int *buf_consumed, |
| 665 | hb_font_t *font, |
| 666 | hb_buffer_serialize_format_t format, |
| 667 | hb_buffer_serialize_flags_t flags) |
| 668 | { |
| 669 | switch (buffer->content_type) |
| 670 | { |
| 671 | |
| 672 | case HB_BUFFER_CONTENT_TYPE_GLYPHS: |
| 673 | return hb_buffer_serialize_glyphs (buffer, start, end, buf, buf_size, |
| 674 | buf_consumed, font, format, flags); |
| 675 | |
| 676 | case HB_BUFFER_CONTENT_TYPE_UNICODE: |
| 677 | return hb_buffer_serialize_unicode (buffer, start, end, buf, buf_size, |
| 678 | buf_consumed, format, flags); |
| 679 | |
| 680 | case HB_BUFFER_CONTENT_TYPE_INVALID: |
| 681 | default: |
| 682 | return _hb_buffer_serialize_invalid (buffer, start, end, buf, buf_size, |
| 683 | buf_consumed, format, flags); |
| 684 | } |
| 685 | } |
| 686 | |
| 687 | static bool |
| 688 | parse_int (const char *pp, const char *end, int32_t *pv) |
| 689 | { |
| 690 | int v; |
| 691 | const char *p = pp; |
| 692 | if (unlikely (!hb_parse_int (&p, end, &v, true/* whole buffer */))) |
| 693 | return false; |
| 694 | |
| 695 | *pv = v; |
| 696 | return true; |
| 697 | } |
| 698 | |
| 699 | static bool |
| 700 | parse_uint (const char *pp, const char *end, uint32_t *pv) |
| 701 | { |
| 702 | unsigned int v; |
| 703 | const char *p = pp; |
| 704 | if (unlikely (!hb_parse_uint (&p, end, &v, true/* whole buffer */))) |
| 705 | return false; |
| 706 | |
| 707 | *pv = v; |
| 708 | return true; |
| 709 | } |
| 710 | |
| 711 | static bool |
| 712 | parse_hex (const char *pp, const char *end, uint32_t *pv) |
| 713 | { |
| 714 | unsigned int v; |
| 715 | const char *p = pp; |
| 716 | if (unlikely (!hb_parse_uint (&p, end, &v, true/* whole buffer */, 16))) |
| 717 | return false; |
| 718 | |
| 719 | *pv = v; |
| 720 | return true; |
| 721 | } |
| 722 | |
| 723 | #include "hb-buffer-deserialize-json.hh" |
| 724 | #include "hb-buffer-deserialize-text-glyphs.hh" |
| 725 | #include "hb-buffer-deserialize-text-unicode.hh" |
| 726 | |
| 727 | /** |
| 728 | * hb_buffer_deserialize_glyphs: |
| 729 | * @buffer: an #hb_buffer_t buffer. |
| 730 | * @buf: (array length=buf_len): string to deserialize |
| 731 | * @buf_len: the size of @buf, or -1 if it is `NULL`-terminated |
| 732 | * @end_ptr: (out) (optional): output pointer to the character after last |
| 733 | * consumed one. |
| 734 | * @font: (nullable): font for getting glyph IDs |
| 735 | * @format: the #hb_buffer_serialize_format_t of the input @buf |
| 736 | * |
| 737 | * Deserializes glyphs @buffer from textual representation in the format |
| 738 | * produced by hb_buffer_serialize_glyphs(). |
| 739 | * |
| 740 | * Return value: `true` if parse was successful, `false` if an error |
| 741 | * occurred. |
| 742 | * |
| 743 | * Since: 0.9.7 |
| 744 | **/ |
| 745 | hb_bool_t |
| 746 | hb_buffer_deserialize_glyphs (hb_buffer_t *buffer, |
| 747 | const char *buf, |
| 748 | int buf_len, /* -1 means nul-terminated */ |
| 749 | const char **end_ptr, /* May be NULL */ |
| 750 | hb_font_t *font, /* May be NULL */ |
| 751 | hb_buffer_serialize_format_t format) |
| 752 | { |
| 753 | const char *end; |
| 754 | if (!end_ptr) |
| 755 | end_ptr = &end; |
| 756 | *end_ptr = buf; |
| 757 | |
| 758 | buffer->assert_glyphs (); |
| 759 | |
| 760 | if (unlikely (hb_object_is_immutable (buffer))) |
| 761 | { |
| 762 | if (end_ptr) |
| 763 | *end_ptr = buf; |
| 764 | return false; |
| 765 | } |
| 766 | |
| 767 | if (buf_len == -1) |
| 768 | buf_len = strlen (buf); |
| 769 | |
| 770 | if (!buf_len) |
| 771 | { |
| 772 | *end_ptr = buf; |
| 773 | return false; |
| 774 | } |
| 775 | |
| 776 | hb_buffer_set_content_type (buffer, HB_BUFFER_CONTENT_TYPE_GLYPHS); |
| 777 | |
| 778 | if (!font) |
| 779 | font = hb_font_get_empty (); |
| 780 | |
| 781 | switch (format) |
| 782 | { |
| 783 | case HB_BUFFER_SERIALIZE_FORMAT_TEXT: |
| 784 | return _hb_buffer_deserialize_text_glyphs (buffer, |
| 785 | buf, buf_len, end_ptr, |
| 786 | font); |
| 787 | |
| 788 | case HB_BUFFER_SERIALIZE_FORMAT_JSON: |
| 789 | return _hb_buffer_deserialize_json (buffer, |
| 790 | buf, buf_len, end_ptr, |
| 791 | font); |
| 792 | |
| 793 | default: |
| 794 | case HB_BUFFER_SERIALIZE_FORMAT_INVALID: |
| 795 | return false; |
| 796 | |
| 797 | } |
| 798 | } |
| 799 | |
| 800 | |
| 801 | /** |
| 802 | * hb_buffer_deserialize_unicode: |
| 803 | * @buffer: an #hb_buffer_t buffer. |
| 804 | * @buf: (array length=buf_len): string to deserialize |
| 805 | * @buf_len: the size of @buf, or -1 if it is `NULL`-terminated |
| 806 | * @end_ptr: (out) (optional): output pointer to the character after last |
| 807 | * consumed one. |
| 808 | * @format: the #hb_buffer_serialize_format_t of the input @buf |
| 809 | * |
| 810 | * Deserializes Unicode @buffer from textual representation in the format |
| 811 | * produced by hb_buffer_serialize_unicode(). |
| 812 | * |
| 813 | * Return value: `true` if parse was successful, `false` if an error |
| 814 | * occurred. |
| 815 | * |
| 816 | * Since: 2.7.3 |
| 817 | **/ |
| 818 | hb_bool_t |
| 819 | hb_buffer_deserialize_unicode (hb_buffer_t *buffer, |
| 820 | const char *buf, |
| 821 | int buf_len, /* -1 means nul-terminated */ |
| 822 | const char **end_ptr, /* May be NULL */ |
| 823 | hb_buffer_serialize_format_t format) |
| 824 | { |
| 825 | const char *end; |
| 826 | if (!end_ptr) |
| 827 | end_ptr = &end; |
| 828 | *end_ptr = buf; |
| 829 | |
| 830 | buffer->assert_unicode (); |
| 831 | |
| 832 | if (unlikely (hb_object_is_immutable (buffer))) |
| 833 | { |
| 834 | if (end_ptr) |
| 835 | *end_ptr = buf; |
| 836 | return false; |
| 837 | } |
| 838 | |
| 839 | if (buf_len == -1) |
| 840 | buf_len = strlen (buf); |
| 841 | |
| 842 | if (!buf_len) |
| 843 | { |
| 844 | *end_ptr = buf; |
| 845 | return false; |
| 846 | } |
| 847 | |
| 848 | hb_buffer_set_content_type (buffer, HB_BUFFER_CONTENT_TYPE_UNICODE); |
| 849 | |
| 850 | hb_font_t* font = hb_font_get_empty (); |
| 851 | |
| 852 | switch (format) |
| 853 | { |
| 854 | case HB_BUFFER_SERIALIZE_FORMAT_TEXT: |
| 855 | return _hb_buffer_deserialize_text_unicode (buffer, |
| 856 | buf, buf_len, end_ptr, |
| 857 | font); |
| 858 | |
| 859 | case HB_BUFFER_SERIALIZE_FORMAT_JSON: |
| 860 | return _hb_buffer_deserialize_json (buffer, |
| 861 | buf, buf_len, end_ptr, |
| 862 | font); |
| 863 | |
| 864 | default: |
| 865 | case HB_BUFFER_SERIALIZE_FORMAT_INVALID: |
| 866 | return false; |
| 867 | |
| 868 | } |
| 869 | } |
| 870 | |
| 871 | |
| 872 | #endif |
| 873 | |