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-buffer-private.hh"
28
29
30 static const char *serialize_formats[] = {
31 "text",
32 "json",
33 nullptr
34 };
35
36 /**
37 * hb_buffer_serialize_list_formats:
38 *
39 * Returns a list of supported buffer serialization formats.
40 *
41 * Return value: (transfer none):
42 * A string array of buffer serialization formats. Should not be freed.
43 *
44 * Since: 0.9.7
45 **/
46 const char **
47 hb_buffer_serialize_list_formats (void)
48 {
49 return serialize_formats;
50 }
51
52 /**
53 * hb_buffer_serialize_format_from_string:
54 * @str: (array length=len) (element-type uint8_t): a string to parse
55 * @len: length of @str, or -1 if string is %NULL terminated
56 *
57 * Parses a string into an #hb_buffer_serialize_format_t. Does not check if
58 * @str is a valid buffer serialization format, use
59 * hb_buffer_serialize_list_formats() to get the list of supported formats.
60 *
61 * Return value:
62 * The parsed #hb_buffer_serialize_format_t.
63 *
64 * Since: 0.9.7
65 **/
66 hb_buffer_serialize_format_t
67 hb_buffer_serialize_format_from_string (const char *str, int len)
68 {
69 /* Upper-case it. */
70 return (hb_buffer_serialize_format_t) (hb_tag_from_string (str, len) & ~0x20202020u);
71 }
72
73 /**
74 * hb_buffer_serialize_format_to_string:
75 * @format: an #hb_buffer_serialize_format_t to convert.
76 *
77 * Converts @format to the string corresponding it, or %NULL if it is not a valid
78 * #hb_buffer_serialize_format_t.
79 *
80 * Return value: (transfer none):
81 * A %NULL terminated string corresponding to @format. Should not be freed.
82 *
83 * Since: 0.9.7
84 **/
85 const char *
86 hb_buffer_serialize_format_to_string (hb_buffer_serialize_format_t format)
87 {
88 switch (format)
89 {
90 case HB_BUFFER_SERIALIZE_FORMAT_TEXT: return serialize_formats[0];
91 case HB_BUFFER_SERIALIZE_FORMAT_JSON: return serialize_formats[1];
92 default:
93 case HB_BUFFER_SERIALIZE_FORMAT_INVALID: return nullptr;
94 }
95 }
96
97 static unsigned int
98 _hb_buffer_serialize_glyphs_json (hb_buffer_t *buffer,
99 unsigned int start,
100 unsigned int end,
101 char *buf,
102 unsigned int buf_size,
103 unsigned int *buf_consumed,
104 hb_font_t *font,
105 hb_buffer_serialize_flags_t flags)
106 {
107 hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr);
108 hb_glyph_position_t *pos = (flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS) ?
109 nullptr : hb_buffer_get_glyph_positions (buffer, nullptr);
110
111 *buf_consumed = 0;
112 for (unsigned int i = start; i < end; i++)
113 {
114 char b[1024];
115 char *p = b;
116
117 /* In the following code, we know b is large enough that no overflow can happen. */
118
119 #define APPEND(s) HB_STMT_START { strcpy (p, s); p += strlen (s); } HB_STMT_END
120
121 if (i)
122 *p++ = ',';
123
124 *p++ = '{';
125
126 APPEND ("\"g\":");
127 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES))
128 {
129 char g[128];
130 hb_font_glyph_to_string (font, info[i].codepoint, g, sizeof (g));
131 *p++ = '"';
132 for (char *q = g; *q; q++) {
133 if (*q == '"')
134 *p++ = '\\';
135 *p++ = *q;
136 }
137 *p++ = '"';
138 }
139 else
140 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%u", info[i].codepoint));
141
142 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) {
143 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"cl\":%u", info[i].cluster));
144 }
145
146 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS))
147 {
148 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"dx\":%d,\"dy\":%d",
149 pos[i].x_offset, pos[i].y_offset));
150 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"ax\":%d,\"ay\":%d",
151 pos[i].x_advance, pos[i].y_advance));
152 }
153
154 if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_FLAGS)
155 {
156 if (info[i].mask & HB_GLYPH_FLAG_DEFINED)
157 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"fl\":%u", info[i].mask & HB_GLYPH_FLAG_DEFINED));
158 }
159
160 if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS)
161 {
162 hb_glyph_extents_t extents;
163 hb_font_get_glyph_extents(font, info[i].codepoint, &extents);
164 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"xb\":%d,\"yb\":%d",
165 extents.x_bearing, extents.y_bearing));
166 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",\"w\":%d,\"h\":%d",
167 extents.width, extents.height));
168 }
169
170 *p++ = '}';
171
172 unsigned int l = p - b;
173 if (buf_size > l)
174 {
175 memcpy (buf, b, l);
176 buf += l;
177 buf_size -= l;
178 *buf_consumed += l;
179 *buf = '\0';
180 } else
181 return i - start;
182 }
183
184 return end - start;
185 }
186
187 static unsigned int
188 _hb_buffer_serialize_glyphs_text (hb_buffer_t *buffer,
189 unsigned int start,
190 unsigned int end,
191 char *buf,
192 unsigned int buf_size,
193 unsigned int *buf_consumed,
194 hb_font_t *font,
195 hb_buffer_serialize_flags_t flags)
196 {
197 hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr);
198 hb_glyph_position_t *pos = (flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS) ?
199 nullptr : hb_buffer_get_glyph_positions (buffer, nullptr);
200
201 *buf_consumed = 0;
202 for (unsigned int i = start; i < end; i++)
203 {
204 char b[1024];
205 char *p = b;
206
207 /* In the following code, we know b is large enough that no overflow can happen. */
208
209 if (i)
210 *p++ = '|';
211
212 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES))
213 {
214 hb_font_glyph_to_string (font, info[i].codepoint, p, 128);
215 p += strlen (p);
216 }
217 else
218 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%u", info[i].codepoint));
219
220 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS)) {
221 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "=%u", info[i].cluster));
222 }
223
224 if (!(flags & HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS))
225 {
226 if (pos[i].x_offset || pos[i].y_offset)
227 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "@%d,%d", pos[i].x_offset, pos[i].y_offset));
228
229 *p++ = '+';
230 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "%d", pos[i].x_advance));
231 if (pos[i].y_advance)
232 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), ",%d", pos[i].y_advance));
233 }
234
235 if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_FLAGS)
236 {
237 if (info[i].mask &HB_GLYPH_FLAG_DEFINED)
238 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "#%X", info[i].mask &HB_GLYPH_FLAG_DEFINED));
239 }
240
241 if (flags & HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS)
242 {
243 hb_glyph_extents_t extents;
244 hb_font_get_glyph_extents(font, info[i].codepoint, &extents);
245 p += MAX (0, snprintf (p, ARRAY_LENGTH (b) - (p - b), "<%d,%d,%d,%d>", extents.x_bearing, extents.y_bearing, extents.width, extents.height));
246 }
247
248 unsigned int l = p - b;
249 if (buf_size > l)
250 {
251 memcpy (buf, b, l);
252 buf += l;
253 buf_size -= l;
254 *buf_consumed += l;
255 *buf = '\0';
256 } else
257 return i - start;
258 }
259
260 return end - start;
261 }
262
263 /**
264 * hb_buffer_serialize_glyphs:
265 * @buffer: an #hb_buffer_t buffer.
266 * @start: the first item in @buffer to serialize.
267 * @end: the last item in @buffer to serialize.
268 * @buf: (out) (array length=buf_size) (element-type uint8_t): output string to
269 * write serialized buffer into.
270 * @buf_size: the size of @buf.
271 * @buf_consumed: (out) (allow-none): if not %NULL, will be set to the number of byes written into @buf.
272 * @font: (allow-none): the #hb_font_t used to shape this buffer, needed to
273 * read glyph names and extents. If %NULL, and empty font will be used.
274 * @format: the #hb_buffer_serialize_format_t to use for formatting the output.
275 * @flags: the #hb_buffer_serialize_flags_t that control what glyph properties
276 * to serialize.
277 *
278 * Serializes @buffer into a textual representation of its glyph content,
279 * useful for showing the contents of the buffer, for example during debugging.
280 * There are currently two supported serialization formats:
281 *
282 * ## text
283 * A human-readable, plain text format.
284 * The serialized glyphs will look something like:
285 *
286 * ```
287 * [uni0651=0@518,0+0|uni0628=0+1897]
288 * ```
289 * - The serialized glyphs are delimited with `[` and `]`.
290 * - Glyphs are separated with `|`
291 * - Each glyph starts with glyph name, or glyph index if
292 * #HB_BUFFER_SERIALIZE_FLAG_NO_GLYPH_NAMES flag is set. Then,
293 * - If #HB_BUFFER_SERIALIZE_FLAG_NO_CLUSTERS is not set, `=` then #hb_glyph_info_t.cluster.
294 * - If #HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS is not set, the #hb_glyph_position_t in the format:
295 * - If both #hb_glyph_position_t.x_offset and #hb_glyph_position_t.y_offset are not 0, `@x_offset,y_offset`. Then,
296 * - `+x_advance`, then `,y_advance` if #hb_glyph_position_t.y_advance is not 0. Then,
297 * - If #HB_BUFFER_SERIALIZE_FLAG_GLYPH_EXTENTS is set, the
298 * #hb_glyph_extents_t in the format
299 * `<x_bearing,y_bearing,width,height>`
300 *
301 * ## json
302 * TODO.
303 *
304 * Return value:
305 * The number of serialized items.
306 *
307 * Since: 0.9.7
308 **/
309 unsigned int
310 hb_buffer_serialize_glyphs (hb_buffer_t *buffer,
311 unsigned int start,
312 unsigned int end,
313 char *buf,
314 unsigned int buf_size,
315 unsigned int *buf_consumed,
316 hb_font_t *font,
317 hb_buffer_serialize_format_t format,
318 hb_buffer_serialize_flags_t flags)
319 {
320 assert (start <= end && end <= buffer->len);
321
322 unsigned int sconsumed;
323 if (!buf_consumed)
324 buf_consumed = &sconsumed;
325 *buf_consumed = 0;
326 if (buf_size)
327 *buf = '\0';
328
329 assert ((!buffer->len && buffer->content_type == HB_BUFFER_CONTENT_TYPE_INVALID) ||
330 buffer->content_type == HB_BUFFER_CONTENT_TYPE_GLYPHS);
331
332 if (!buffer->have_positions)
333 flags |= HB_BUFFER_SERIALIZE_FLAG_NO_POSITIONS;
334
335 if (unlikely (start == end))
336 return 0;
337
338 if (!font)
339 font = hb_font_get_empty ();
340
341 switch (format)
342 {
343 case HB_BUFFER_SERIALIZE_FORMAT_TEXT:
344 return _hb_buffer_serialize_glyphs_text (buffer, start, end,
345 buf, buf_size, buf_consumed,
346 font, flags);
347
348 case HB_BUFFER_SERIALIZE_FORMAT_JSON:
349 return _hb_buffer_serialize_glyphs_json (buffer, start, end,
350 buf, buf_size, buf_consumed,
351 font, flags);
352
353 default:
354 case HB_BUFFER_SERIALIZE_FORMAT_INVALID:
355 return 0;
356
357 }
358 }
359
360
361 static hb_bool_t
362 parse_uint (const char *pp, const char *end, uint32_t *pv)
363 {
364 char buf[32];
365 unsigned int len = MIN (ARRAY_LENGTH (buf) - 1, (unsigned int) (end - pp));
366 strncpy (buf, pp, len);
367 buf[len] = '\0';
368
369 char *p = buf;
370 char *pend = p;
371 uint32_t v;
372
373 errno = 0;
374 v = strtol (p, &pend, 10);
375 if (errno || p == pend || pend - p != end - pp)
376 return false;
377
378 *pv = v;
379 return true;
380 }
381
382 static hb_bool_t
383 parse_int (const char *pp, const char *end, int32_t *pv)
384 {
385 char buf[32];
386 unsigned int len = MIN (ARRAY_LENGTH (buf) - 1, (unsigned int) (end - pp));
387 strncpy (buf, pp, len);
388 buf[len] = '\0';
389
390 char *p = buf;
391 char *pend = p;
392 int32_t v;
393
394 errno = 0;
395 v = strtol (p, &pend, 10);
396 if (errno || p == pend || pend - p != end - pp)
397 return false;
398
399 *pv = v;
400 return true;
401 }
402
403 #include "hb-buffer-deserialize-json.hh"
404 #include "hb-buffer-deserialize-text.hh"
405
406 /**
407 * hb_buffer_deserialize_glyphs:
408 * @buffer: an #hb_buffer_t buffer.
409 * @buf: (array length=buf_len):
410 * @buf_len:
411 * @end_ptr: (out):
412 * @font:
413 * @format:
414 *
415 *
416 *
417 * Return value:
418 *
419 * Since: 0.9.7
420 **/
421 hb_bool_t
422 hb_buffer_deserialize_glyphs (hb_buffer_t *buffer,
423 const char *buf,
424 int buf_len, /* -1 means nul-terminated */
425 const char **end_ptr, /* May be nullptr */
426 hb_font_t *font, /* May be nullptr */
427 hb_buffer_serialize_format_t format)
428 {
429 const char *end;
430 if (!end_ptr)
431 end_ptr = &end;
432 *end_ptr = buf;
433
434 assert ((!buffer->len && buffer->content_type == HB_BUFFER_CONTENT_TYPE_INVALID) ||
435 buffer->content_type == HB_BUFFER_CONTENT_TYPE_GLYPHS);
436
437 if (buf_len == -1)
438 buf_len = strlen (buf);
439
440 if (!buf_len)
441 {
442 *end_ptr = buf;
443 return false;
444 }
445
446 hb_buffer_set_content_type (buffer, HB_BUFFER_CONTENT_TYPE_GLYPHS);
447
448 if (!font)
449 font = hb_font_get_empty ();
450
451 switch (format)
452 {
453 case HB_BUFFER_SERIALIZE_FORMAT_TEXT:
454 return _hb_buffer_deserialize_glyphs_text (buffer,
455 buf, buf_len, end_ptr,
456 font);
457
458 case HB_BUFFER_SERIALIZE_FORMAT_JSON:
459 return _hb_buffer_deserialize_glyphs_json (buffer,
460 buf, buf_len, end_ptr,
461 font);
462
463 default:
464 case HB_BUFFER_SERIALIZE_FORMAT_INVALID:
465 return false;
466
467 }
468 }