1 /*
2 * Copyright © 2011 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 #ifndef SHAPE_OPTIONS_HH
28 #define SHAPE_OPTIONS_HH
29
30 #include "options.hh"
31
32 struct shape_options_t
33 {
~shape_options_tshape_options_t34 ~shape_options_t ()
35 {
36 g_free (direction);
37 g_free (language);
38 g_free (script);
39 free (features);
40 g_strfreev (shapers);
41 }
42
43 void add_options (option_parser_t *parser);
44
setup_buffershape_options_t45 void setup_buffer (hb_buffer_t *buffer)
46 {
47 hb_buffer_set_direction (buffer, hb_direction_from_string (direction, -1));
48 hb_buffer_set_script (buffer, hb_script_from_string (script, -1));
49 hb_buffer_set_language (buffer, hb_language_from_string (language, -1));
50 hb_buffer_set_flags (buffer, (hb_buffer_flags_t)
51 (HB_BUFFER_FLAG_DEFAULT |
52 (bot ? HB_BUFFER_FLAG_BOT : 0) |
53 (eot ? HB_BUFFER_FLAG_EOT : 0) |
54 (preserve_default_ignorables ? HB_BUFFER_FLAG_PRESERVE_DEFAULT_IGNORABLES : 0) |
55 (remove_default_ignorables ? HB_BUFFER_FLAG_REMOVE_DEFAULT_IGNORABLES : 0) |
56 0));
57 hb_buffer_set_invisible_glyph (buffer, invisible_glyph);
58 hb_buffer_set_not_found_glyph (buffer, not_found_glyph);
59 hb_buffer_set_cluster_level (buffer, cluster_level);
60 hb_buffer_guess_segment_properties (buffer);
61 }
62
copy_buffer_propertiesshape_options_t63 static void copy_buffer_properties (hb_buffer_t *dst, hb_buffer_t *src)
64 {
65 hb_segment_properties_t props;
66 hb_buffer_get_segment_properties (src, &props);
67 hb_buffer_set_segment_properties (dst, &props);
68 hb_buffer_set_flags (dst, hb_buffer_get_flags (src));
69 hb_buffer_set_cluster_level (dst, hb_buffer_get_cluster_level (src));
70 }
71
populate_buffershape_options_t72 void populate_buffer (hb_buffer_t *buffer, const char *text, int text_len,
73 const char *text_before, const char *text_after)
74 {
75 hb_buffer_clear_contents (buffer);
76 if (text_before) {
77 unsigned int len = strlen (text_before);
78 hb_buffer_add_utf8 (buffer, text_before, len, len, 0);
79 }
80 hb_buffer_add_utf8 (buffer, text, text_len, 0, text_len);
81 if (text_after) {
82 hb_buffer_add_utf8 (buffer, text_after, -1, 0, 0);
83 }
84
85 if (!utf8_clusters) {
86 /* Reset cluster values to refer to Unicode character index
87 * instead of UTF-8 index. */
88 unsigned int num_glyphs = hb_buffer_get_length (buffer);
89 hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, nullptr);
90 for (unsigned int i = 0; i < num_glyphs; i++)
91 {
92 info->cluster = i;
93 info++;
94 }
95 }
96
97 setup_buffer (buffer);
98 }
99
shapeshape_options_t100 hb_bool_t shape (hb_font_t *font, hb_buffer_t *buffer, const char **error=nullptr)
101 {
102 hb_buffer_t *text_buffer = nullptr;
103 if (verify)
104 {
105 text_buffer = hb_buffer_create ();
106 hb_buffer_append (text_buffer, buffer, 0, -1);
107 }
108
109 if (!hb_shape_full (font, buffer, features, num_features, shapers))
110 {
111 if (error)
112 *error = "All shapers failed.";
113 goto fail;
114 }
115
116 if (normalize_glyphs)
117 hb_buffer_normalize_glyphs (buffer);
118
119 if (verify && !verify_buffer (buffer, text_buffer, font, error))
120 goto fail;
121
122 if (text_buffer)
123 hb_buffer_destroy (text_buffer);
124
125 return true;
126
127 fail:
128 if (text_buffer)
129 hb_buffer_destroy (text_buffer);
130
131 return false;
132 }
133
verify_buffershape_options_t134 bool verify_buffer (hb_buffer_t *buffer,
135 hb_buffer_t *text_buffer,
136 hb_font_t *font,
137 const char **error=nullptr)
138 {
139 if (!verify_buffer_monotone (buffer, error))
140 return false;
141 if (!verify_buffer_safe_to_break (buffer, text_buffer, font, error))
142 return false;
143 return true;
144 }
145
verify_buffer_monotoneshape_options_t146 bool verify_buffer_monotone (hb_buffer_t *buffer, const char **error=nullptr)
147 {
148 /* Check that clusters are monotone. */
149 if (cluster_level == HB_BUFFER_CLUSTER_LEVEL_MONOTONE_GRAPHEMES ||
150 cluster_level == HB_BUFFER_CLUSTER_LEVEL_MONOTONE_CHARACTERS)
151 {
152 bool is_forward = HB_DIRECTION_IS_FORWARD (hb_buffer_get_direction (buffer));
153
154 unsigned int num_glyphs;
155 hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, &num_glyphs);
156
157 for (unsigned int i = 1; i < num_glyphs; i++)
158 if (info[i-1].cluster != info[i].cluster &&
159 (info[i-1].cluster < info[i].cluster) != is_forward)
160 {
161 if (error)
162 *error = "clusters are not monotone.";
163 return false;
164 }
165 }
166
167 return true;
168 }
169
verify_buffer_safe_to_breakshape_options_t170 bool verify_buffer_safe_to_break (hb_buffer_t *buffer,
171 hb_buffer_t *text_buffer,
172 hb_font_t *font,
173 const char **error=nullptr)
174 {
175 if (cluster_level != HB_BUFFER_CLUSTER_LEVEL_MONOTONE_GRAPHEMES &&
176 cluster_level != HB_BUFFER_CLUSTER_LEVEL_MONOTONE_CHARACTERS)
177 {
178 /* Cannot perform this check without monotone clusters.
179 * Then again, unsafe-to-break flag is much harder to use without
180 * monotone clusters. */
181 return true;
182 }
183
184 /* Check that breaking up shaping at safe-to-break is indeed safe. */
185
186 hb_buffer_t *fragment = hb_buffer_create ();
187 hb_buffer_t *reconstruction = hb_buffer_create ();
188 copy_buffer_properties (reconstruction, buffer);
189
190 unsigned int num_glyphs;
191 hb_glyph_info_t *info = hb_buffer_get_glyph_infos (buffer, &num_glyphs);
192
193 unsigned int num_chars;
194 hb_glyph_info_t *text = hb_buffer_get_glyph_infos (text_buffer, &num_chars);
195
196 /* Chop text and shape fragments. */
197 bool forward = HB_DIRECTION_IS_FORWARD (hb_buffer_get_direction (buffer));
198 unsigned int start = 0;
199 unsigned int text_start = forward ? 0 : num_chars;
200 unsigned int text_end = text_start;
201 for (unsigned int end = 1; end < num_glyphs + 1; end++)
202 {
203 if (end < num_glyphs &&
204 (info[end].cluster == info[end-1].cluster ||
205 info[end-(forward?0:1)].mask & HB_GLYPH_FLAG_UNSAFE_TO_BREAK))
206 continue;
207
208 /* Shape segment corresponding to glyphs start..end. */
209 if (end == num_glyphs)
210 {
211 if (forward)
212 text_end = num_chars;
213 else
214 text_start = 0;
215 }
216 else
217 {
218 if (forward)
219 {
220 unsigned int cluster = info[end].cluster;
221 while (text_end < num_chars && text[text_end].cluster < cluster)
222 text_end++;
223 }
224 else
225 {
226 unsigned int cluster = info[end - 1].cluster;
227 while (text_start && text[text_start - 1].cluster >= cluster)
228 text_start--;
229 }
230 }
231 assert (text_start < text_end);
232
233 if (0)
234 printf("start %d end %d text start %d end %d\n", start, end, text_start, text_end);
235
236 hb_buffer_clear_contents (fragment);
237 copy_buffer_properties (fragment, buffer);
238
239 hb_buffer_flags_t flags = hb_buffer_get_flags (fragment);
240 if (0 < text_start)
241 flags = (hb_buffer_flags_t) (flags & ~HB_BUFFER_FLAG_BOT);
242 if (text_end < num_chars)
243 flags = (hb_buffer_flags_t) (flags & ~HB_BUFFER_FLAG_EOT);
244 hb_buffer_set_flags (fragment, flags);
245
246 hb_buffer_append (fragment, text_buffer, text_start, text_end);
247 if (!hb_shape_full (font, fragment, features, num_features, shapers))
248 {
249 if (error)
250 *error = "All shapers failed while shaping fragment.";
251 hb_buffer_destroy (reconstruction);
252 hb_buffer_destroy (fragment);
253 return false;
254 }
255 hb_buffer_append (reconstruction, fragment, 0, -1);
256
257 start = end;
258 if (forward)
259 text_start = text_end;
260 else
261 text_end = text_start;
262 }
263
264 bool ret = true;
265 hb_buffer_diff_flags_t diff = hb_buffer_diff (reconstruction, buffer, (hb_codepoint_t) -1, 0);
266 if (diff)
267 {
268 if (error)
269 *error = "Safe-to-break test failed.";
270 ret = false;
271
272 /* Return the reconstructed result instead so it can be inspected. */
273 hb_buffer_set_length (buffer, 0);
274 hb_buffer_append (buffer, reconstruction, 0, -1);
275 }
276
277 hb_buffer_destroy (reconstruction);
278 hb_buffer_destroy (fragment);
279
280 return ret;
281 }
282
shape_closureshape_options_t283 void shape_closure (const char *text, int text_len,
284 hb_font_t *font, hb_buffer_t *buffer,
285 hb_set_t *glyphs)
286 {
287 hb_buffer_reset (buffer);
288 hb_buffer_add_utf8 (buffer, text, text_len, 0, text_len);
289 setup_buffer (buffer);
290 hb_ot_shape_glyphs_closure (font, buffer, features, num_features, glyphs);
291 }
292
293 /* Buffer properties */
294 char *direction = nullptr;
295 char *language = nullptr;
296 char *script = nullptr;
297
298 /* Buffer flags */
299 hb_bool_t bot = false;
300 hb_bool_t eot = false;
301 hb_bool_t preserve_default_ignorables = false;
302 hb_bool_t remove_default_ignorables = false;
303
304 hb_feature_t *features = nullptr;
305 unsigned int num_features = 0;
306 char **shapers = nullptr;
307 hb_bool_t utf8_clusters = false;
308 hb_codepoint_t invisible_glyph = 0;
309 hb_codepoint_t not_found_glyph = 0;
310 hb_buffer_cluster_level_t cluster_level = HB_BUFFER_CLUSTER_LEVEL_DEFAULT;
311 hb_bool_t normalize_glyphs = false;
312 hb_bool_t verify = false;
313 unsigned int num_iterations = 1;
314 };
315
316
317 static gboolean
parse_shapers(const char * name G_GNUC_UNUSED,const char * arg,gpointer data,GError ** error)318 parse_shapers (const char *name G_GNUC_UNUSED,
319 const char *arg,
320 gpointer data,
321 GError **error)
322 {
323 shape_options_t *shape_opts = (shape_options_t *) data;
324 char **shapers = g_strsplit (arg, ",", 0);
325
326 for (char **shaper = shapers; *shaper; shaper++)
327 {
328 bool found = false;
329 for (const char **hb_shaper = hb_shape_list_shapers (); *hb_shaper; hb_shaper++) {
330 if (strcmp (*shaper, *hb_shaper) == 0)
331 {
332 found = true;
333 break;
334 }
335 }
336 if (!found)
337 {
338 g_set_error (error, G_OPTION_ERROR, G_OPTION_ERROR_BAD_VALUE,
339 "Unknown or unsupported shaper: %s", *shaper);
340 g_strfreev (shapers);
341 return false;
342 }
343 }
344
345 g_strfreev (shape_opts->shapers);
346 shape_opts->shapers = shapers;
347 return true;
348 }
349
350 static G_GNUC_NORETURN gboolean
list_shapers(const char * name G_GNUC_UNUSED,const char * arg G_GNUC_UNUSED,gpointer data G_GNUC_UNUSED,GError ** error G_GNUC_UNUSED)351 list_shapers (const char *name G_GNUC_UNUSED,
352 const char *arg G_GNUC_UNUSED,
353 gpointer data G_GNUC_UNUSED,
354 GError **error G_GNUC_UNUSED)
355 {
356 for (const char **shaper = hb_shape_list_shapers (); *shaper; shaper++)
357 g_printf ("%s\n", *shaper);
358
359 exit(0);
360 }
361
362
363 static gboolean
parse_features(const char * name G_GNUC_UNUSED,const char * arg,gpointer data,GError ** error G_GNUC_UNUSED)364 parse_features (const char *name G_GNUC_UNUSED,
365 const char *arg,
366 gpointer data,
367 GError **error G_GNUC_UNUSED)
368 {
369 shape_options_t *shape_opts = (shape_options_t *) data;
370 char *s = (char *) arg;
371 size_t l = strlen (s);
372 char *p;
373
374 shape_opts->num_features = 0;
375 g_free (shape_opts->features);
376 shape_opts->features = nullptr;
377
378 /* if the string is quoted, strip the quotes */
379 if (s[0] == s[l - 1] && (s[0] == '\"' || s[0] == '\''))
380 {
381 s[l - 1] = '\0';
382 s++;
383 }
384
385 if (!*s)
386 return true;
387
388 /* count the features first, so we can allocate memory */
389 p = s;
390 do {
391 shape_opts->num_features++;
392 p = strchr (p, ',');
393 if (p)
394 p++;
395 } while (p);
396
397 shape_opts->features = (hb_feature_t *) calloc (shape_opts->num_features, sizeof (*shape_opts->features));
398 if (!shape_opts->features)
399 return false;
400
401 /* now do the actual parsing */
402 p = s;
403 shape_opts->num_features = 0;
404 while (p && *p) {
405 char *end = strchr (p, ',');
406 if (hb_feature_from_string (p, end ? end - p : -1, &shape_opts->features[shape_opts->num_features]))
407 shape_opts->num_features++;
408 p = end ? end + 1 : nullptr;
409 }
410
411 return true;
412 }
413
414 void
add_options(option_parser_t * parser)415 shape_options_t::add_options (option_parser_t *parser)
416 {
417 GOptionEntry entries[] =
418 {
419 {"list-shapers", 0, G_OPTION_FLAG_NO_ARG,
420 G_OPTION_ARG_CALLBACK, (gpointer) &list_shapers, "List available shapers and quit", nullptr},
421 {"shaper", 0, G_OPTION_FLAG_HIDDEN,
422 G_OPTION_ARG_CALLBACK, (gpointer) &parse_shapers, "Hidden duplicate of --shapers", nullptr},
423 {"shapers", 0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_shapers, "Set comma-separated list of shapers to try","list"},
424 {"direction", 0, 0, G_OPTION_ARG_STRING, &this->direction, "Set text direction (default: auto)", "ltr/rtl/ttb/btt"},
425 {"language", 0, 0, G_OPTION_ARG_STRING, &this->language, "Set text language (default: $LANG)", "BCP 47 tag"},
426 {"script", 0, 0, G_OPTION_ARG_STRING, &this->script, "Set text script (default: auto)", "ISO-15924 tag"},
427 {"bot", 0, 0, G_OPTION_ARG_NONE, &this->bot, "Treat text as beginning-of-paragraph", nullptr},
428 {"eot", 0, 0, G_OPTION_ARG_NONE, &this->eot, "Treat text as end-of-paragraph", nullptr},
429 {"preserve-default-ignorables",0, 0, G_OPTION_ARG_NONE, &this->preserve_default_ignorables, "Preserve Default-Ignorable characters", nullptr},
430 {"remove-default-ignorables",0, 0, G_OPTION_ARG_NONE, &this->remove_default_ignorables, "Remove Default-Ignorable characters", nullptr},
431 {"invisible-glyph", 0, 0, G_OPTION_ARG_INT, &this->invisible_glyph, "Glyph value to replace Default-Ignorables with", nullptr},
432 {"not-found-glyph", 0, 0, G_OPTION_ARG_INT, &this->not_found_glyph, "Glyph value to replace not-found characters with", nullptr},
433 {"utf8-clusters", 0, 0, G_OPTION_ARG_NONE, &this->utf8_clusters, "Use UTF8 byte indices, not char indices", nullptr},
434 {"cluster-level", 0, 0, G_OPTION_ARG_INT, &this->cluster_level, "Cluster merging level (default: 0)", "0/1/2"},
435 {"normalize-glyphs",0, 0, G_OPTION_ARG_NONE, &this->normalize_glyphs, "Rearrange glyph clusters in nominal order", nullptr},
436 {"verify", 0, 0, G_OPTION_ARG_NONE, &this->verify, "Perform sanity checks on shaping results", nullptr},
437 {"num-iterations", 'n', G_OPTION_FLAG_IN_MAIN,
438 G_OPTION_ARG_INT, &this->num_iterations, "Run shaper N times (default: 1)", "N"},
439 {nullptr}
440 };
441 parser->add_group (entries,
442 "shape",
443 "Shape options:",
444 "Options for the shaping process",
445 this);
446
447 const gchar *features_help = "Comma-separated list of font features\n"
448 "\n"
449 " Features can be enabled or disabled, either globally or limited to\n"
450 " specific character ranges. The format for specifying feature settings\n"
451 " follows. All valid CSS font-feature-settings values other than 'normal'\n"
452 " and the global values are also accepted, though not documented below.\n"
453 " CSS string escapes are not supported."
454 "\n"
455 " The range indices refer to the positions between Unicode characters,\n"
456 " unless the --utf8-clusters is provided, in which case range indices\n"
457 " refer to UTF-8 byte indices. The position before the first character\n"
458 " is always 0.\n"
459 "\n"
460 " The format is Python-esque. Here is how it all works:\n"
461 "\n"
462 " Syntax: Value: Start: End:\n"
463 "\n"
464 " Setting value:\n"
465 " \"kern\" 1 0 ∞ # Turn feature on\n"
466 " \"+kern\" 1 0 ∞ # Turn feature on\n"
467 " \"-kern\" 0 0 ∞ # Turn feature off\n"
468 " \"kern=0\" 0 0 ∞ # Turn feature off\n"
469 " \"kern=1\" 1 0 ∞ # Turn feature on\n"
470 " \"aalt=2\" 2 0 ∞ # Choose 2nd alternate\n"
471 "\n"
472 " Setting index:\n"
473 " \"kern[]\" 1 0 ∞ # Turn feature on\n"
474 " \"kern[:]\" 1 0 ∞ # Turn feature on\n"
475 " \"kern[5:]\" 1 5 ∞ # Turn feature on, partial\n"
476 " \"kern[:5]\" 1 0 5 # Turn feature on, partial\n"
477 " \"kern[3:5]\" 1 3 5 # Turn feature on, range\n"
478 " \"kern[3]\" 1 3 3+1 # Turn feature on, single char\n"
479 "\n"
480 " Mixing it all:\n"
481 "\n"
482 " \"aalt[3:5]=2\" 2 3 5 # Turn 2nd alternate on for range";
483
484 GOptionEntry entries2[] =
485 {
486 {"features", 0, 0, G_OPTION_ARG_CALLBACK, (gpointer) &parse_features, features_help, "list"},
487 {nullptr}
488 };
489 parser->add_group (entries2,
490 "features",
491 "Features options:",
492 "Options for font features used",
493 this);
494 }
495
496 #endif
497