Mercurial > mplayer.hg
annotate libass/ass.c @ 19637:b79f65f98553
forgotten to pass the crc32 be2me_32()
author | nicodvb |
---|---|
date | Sat, 02 Sep 2006 19:00:30 +0000 |
parents | 4a64fecfb425 |
children | a3473d990fed |
rev | line source |
---|---|
18937 | 1 #include "config.h" |
2 | |
3 #include <stdio.h> | |
4 #include <stdlib.h> | |
5 #include <string.h> | |
6 #include <assert.h> | |
7 #include <errno.h> | |
8 #include <sys/types.h> | |
9 #include <sys/stat.h> | |
10 #include <unistd.h> | |
19374 | 11 #include <inttypes.h> |
18937 | 12 |
13 #ifdef HAVE_ENCA | |
14 #include "subreader.h" // for guess_buffer_cp | |
15 #endif | |
16 | |
17 #ifdef USE_ICONV | |
18 #include <iconv.h> | |
19 extern char *sub_cp; | |
20 #endif | |
19492 | 21 extern int extract_embedded_fonts; |
19495 | 22 extern char** ass_force_style_list; |
18937 | 23 |
24 #include "mp_msg.h" | |
25 #include "ass.h" | |
26 #include "ass_utils.h" | |
27 #include "libvo/sub.h" // for utf8_get_char | |
28 | |
29 char *get_path(char *); | |
30 | |
19492 | 31 struct parser_priv_s { |
32 enum {PST_UNKNOWN = 0, PST_INFO, PST_STYLES, PST_EVENTS, PST_FONTS} state; | |
33 char* fontname; | |
34 char* fontdata; | |
35 int fontdata_size; | |
36 int fontdata_used; | |
37 }; | |
38 | |
18937 | 39 #define ASS_STYLES_ALLOC 20 |
40 #define ASS_EVENTS_ALLOC 200 | |
41 | |
42 void ass_free_track(ass_track_t* track) { | |
43 int i; | |
44 | |
19492 | 45 if (track->parser_priv) { |
46 if (track->parser_priv->fontname) | |
47 free(track->parser_priv->fontname); | |
48 if (track->parser_priv->fontdata) | |
49 free(track->parser_priv->fontdata); | |
50 free(track->parser_priv); | |
51 } | |
18937 | 52 if (track->style_format) |
53 free(track->style_format); | |
54 if (track->event_format) | |
55 free(track->event_format); | |
56 if (track->styles) { | |
19474
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
57 for (i = 0; i < track->n_styles; ++i) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
58 ass_free_style(track, i); |
18937 | 59 free(track->styles); |
60 } | |
61 if (track->events) { | |
19474
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
62 for (i = 0; i < track->n_events; ++i) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
63 ass_free_event(track, i); |
18937 | 64 free(track->events); |
65 } | |
66 } | |
67 | |
68 /// \brief Allocate a new style struct | |
69 /// \param track track | |
70 /// \return style id | |
71 int ass_alloc_style(ass_track_t* track) { | |
72 int sid; | |
73 | |
74 assert(track->n_styles <= track->max_styles); | |
75 | |
76 if (track->n_styles == track->max_styles) { | |
77 track->max_styles += ASS_STYLES_ALLOC; | |
78 track->styles = (ass_style_t*)realloc(track->styles, sizeof(ass_style_t)*track->max_styles); | |
79 } | |
80 | |
81 sid = track->n_styles++; | |
82 memset(track->styles + sid, 0, sizeof(ass_style_t)); | |
83 return sid; | |
84 } | |
85 | |
86 /// \brief Allocate a new event struct | |
87 /// \param track track | |
88 /// \return event id | |
89 int ass_alloc_event(ass_track_t* track) { | |
90 int eid; | |
91 | |
92 assert(track->n_events <= track->max_events); | |
93 | |
94 if (track->n_events == track->max_events) { | |
95 track->max_events += ASS_EVENTS_ALLOC; | |
96 track->events = (ass_event_t*)realloc(track->events, sizeof(ass_event_t)*track->max_events); | |
97 } | |
98 | |
99 eid = track->n_events++; | |
100 memset(track->events + eid, 0, sizeof(ass_event_t)); | |
101 return eid; | |
102 } | |
103 | |
19474
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
104 void ass_free_event(ass_track_t* track, int eid) { |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
105 ass_event_t* event = track->events + eid; |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
106 if (event->Name) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
107 free(event->Name); |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
108 if (event->Effect) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
109 free(event->Effect); |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
110 if (event->Text) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
111 free(event->Text); |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
112 } |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
113 |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
114 void ass_free_style(ass_track_t* track, int sid) { |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
115 ass_style_t* style = track->styles + sid; |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
116 if (style->Name) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
117 free(style->Name); |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
118 if (style->FontName) |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
119 free(style->FontName); |
18937 | 120 } |
121 | |
122 static int events_compare_f(const void* a_, const void* b_) { | |
123 ass_event_t* a = (ass_event_t*)a_; | |
124 ass_event_t* b = (ass_event_t*)b_; | |
125 if (a->Start < b->Start) | |
126 return -1; | |
127 else if (a->Start > b->Start) | |
128 return 1; | |
129 else | |
130 return 0; | |
131 } | |
132 | |
133 /// \brief Sort events by start time | |
134 /// \param tid track id | |
135 static void sort_events(ass_track_t* track) { | |
136 qsort(track->events, track->n_events, sizeof(ass_event_t), events_compare_f); | |
137 } | |
138 | |
139 // ============================================================================================== | |
140 | |
141 static void skip_spaces(char** str) { | |
142 char* p = *str; | |
143 while ((*p==' ') || (*p=='\t')) | |
144 ++p; | |
145 *str = p; | |
146 } | |
147 | |
148 static void rskip_spaces(char** str, char* limit) { | |
149 char* p = *str; | |
150 while ((p >= limit) && ((*p==' ') || (*p=='\t'))) | |
151 --p; | |
152 *str = p; | |
153 } | |
154 | |
155 /** | |
156 * \brief find style by name | |
157 * \param track track | |
158 * \param name style name | |
159 * \return index in track->styles | |
160 * Returnes 0 if no styles found => expects at least 1 style. | |
161 * Parsing code always adds "Default" style in the end. | |
162 */ | |
163 static int lookup_style(ass_track_t* track, char* name) { | |
164 int i; | |
19567 | 165 if (*name == '*') ++name; // FIXME: what does '*' really mean ? |
18937 | 166 for (i=0; i<track->n_styles; ++i) { |
167 // FIXME: mb strcasecmp ? | |
168 if (strcmp(track->styles[i].Name, name) == 0) | |
169 return i; | |
170 } | |
171 i = track->default_style; | |
172 mp_msg(MSGT_GLOBAL, MSGL_WARN, "[%p] Warning: no style named '%s' found, using '%s'\n", track, name, track->styles[i].Name); | |
173 return i; // use the first style | |
174 } | |
175 | |
176 static uint32_t string2color(char* p) { | |
177 uint32_t tmp; | |
178 (void)strtocolor(&p, &tmp); | |
179 return tmp; | |
180 } | |
181 | |
182 static long long string2timecode(char* p) { | |
183 unsigned h, m, s, ms; | |
184 long long tm; | |
185 int res = sscanf(p, "%1d:%2d:%2d.%2d", &h, &m, &s, &ms); | |
186 if (res < 4) { | |
187 mp_msg(MSGT_GLOBAL, MSGL_WARN, "bad timestamp\n"); | |
188 return 0; | |
189 } | |
190 tm = ((h * 60 + m) * 60 + s) * 1000 + ms * 10; | |
191 return tm; | |
192 } | |
193 | |
194 /** | |
195 * \brief converts numpad-style align to align. | |
196 */ | |
197 static int numpad2align(int val) { | |
198 int res, v; | |
199 v = (val - 1) / 3; // 0, 1 or 2 for vertical alignment | |
200 if (v != 0) v = 3 - v; | |
201 res = ((val - 1) % 3) + 1; // horizontal alignment | |
202 res += v*4; | |
203 return res; | |
204 } | |
205 | |
206 #define NEXT(str,token) \ | |
207 token = next_token(&str); \ | |
208 if (!token) break; | |
209 | |
210 #define ANYVAL(name,func) \ | |
211 } else if (strcasecmp(tname, #name) == 0) { \ | |
212 target->name = func(token); \ | |
213 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "%s = %s\n", #name, token); | |
19495 | 214 |
215 #define STRVAL(name) \ | |
216 } else if (strcasecmp(tname, #name) == 0) { \ | |
217 if (target->name != NULL) free(target->name); \ | |
218 target->name = strdup(token); \ | |
219 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "%s = %s\n", #name, token); | |
220 | |
18937 | 221 #define COLORVAL(name) ANYVAL(name,string2color) |
222 #define INTVAL(name) ANYVAL(name,atoi) | |
223 #define FPVAL(name) ANYVAL(name,atof) | |
224 #define TIMEVAL(name) ANYVAL(name,string2timecode) | |
225 #define STYLEVAL(name) \ | |
226 } else if (strcasecmp(tname, #name) == 0) { \ | |
227 target->name = lookup_style(track, token); \ | |
228 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "%s = %s\n", #name, token); | |
229 | |
230 #define ALIAS(alias,name) \ | |
231 if (strcasecmp(tname, #alias) == 0) {tname = #name;} | |
232 | |
233 static char* next_token(char** str) { | |
234 char* p = *str; | |
235 char* start; | |
236 skip_spaces(&p); | |
237 if (*p == '\0') { | |
238 *str = p; | |
239 return 0; | |
240 } | |
241 start = p; // start of the token | |
242 for (; (*p != '\0') && (*p != ','); ++p) {} | |
243 if (*p == '\0') { | |
244 *str = p; // eos found, str will point to '\0' at exit | |
245 } else { | |
246 *p = '\0'; | |
247 *str = p + 1; // ',' found, str will point to the next char (beginning of the next token) | |
248 } | |
249 --p; // end of current token | |
250 rskip_spaces(&p, start); | |
251 if (p < start) | |
252 p = start; // empty token | |
253 else | |
254 ++p; // the first space character, or '\0' | |
255 *p = '\0'; | |
256 return start; | |
257 } | |
258 /** | |
259 * \brief Parse the tail of Dialogue line | |
260 * \param track track | |
261 * \param event parsed data goes here | |
262 * \param str string to parse, zero-terminated | |
263 * \param n_ignored number of format options to skip at the beginning | |
264 */ | |
265 static int process_event_tail(ass_track_t* track, ass_event_t* event, char* str, int n_ignored) | |
266 { | |
267 char* token; | |
268 char* tname; | |
269 char* p = str; | |
270 int i; | |
271 ass_event_t* target = event; | |
272 | |
273 char* format = strdup(track->event_format); | |
274 char* q = format; // format scanning pointer | |
275 | |
276 for (i = 0; i < n_ignored; ++i) { | |
277 NEXT(q, tname); | |
278 } | |
279 | |
280 while (1) { | |
281 NEXT(q, tname); | |
282 if (strcasecmp(tname, "Text") == 0) { | |
283 char* last; | |
284 event->Text = strdup(p); | |
19612 | 285 if (*event->Text != 0) { |
286 last = event->Text + strlen(event->Text) - 1; | |
287 if (last >= event->Text && *last == '\r') | |
288 *last = 0; | |
289 } | |
18937 | 290 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "Text = %s\n", event->Text); |
291 event->Duration -= event->Start; | |
292 free(format); | |
293 return 0; // "Text" is always the last | |
294 } | |
295 NEXT(p, token); | |
296 | |
297 ALIAS(End,Duration) // temporarily store end timecode in event->Duration | |
298 if (0) { // cool ;) | |
19635 | 299 INTVAL(Layer) |
18937 | 300 STYLEVAL(Style) |
301 STRVAL(Name) | |
302 STRVAL(Effect) | |
303 INTVAL(MarginL) | |
304 INTVAL(MarginR) | |
305 INTVAL(MarginV) | |
306 TIMEVAL(Start) | |
307 TIMEVAL(Duration) | |
308 } | |
309 } | |
310 free(format); | |
311 return 1; | |
312 } | |
313 | |
314 /** | |
19495 | 315 * \brief Parse command line style overrides (--ass-force-style option) |
316 * \param track track to apply overrides to | |
317 * The format for overrides is [StyleName.]Field=Value | |
318 */ | |
19584
586e7d259d05
Apply -ass-force-style also to tracks generated from subdata.
eugeni
parents:
19567
diff
changeset
|
319 void process_force_style(ass_track_t* track) { |
19495 | 320 char **fs, *eq, *dt, *style, *tname, *token; |
321 ass_style_t* target; | |
322 int sid; | |
323 | |
324 if (!ass_force_style_list) return; | |
325 | |
326 for (fs = ass_force_style_list; *fs; ++fs) { | |
327 eq = strchr(*fs, '='); | |
328 if (!eq) | |
329 continue; | |
330 *eq = '\0'; | |
331 token = eq + 1; | |
332 | |
333 dt = strchr(*fs, '.'); | |
334 if (dt) { | |
335 *dt = '\0'; | |
336 style = *fs; | |
337 tname = dt + 1; | |
338 } else { | |
339 style = NULL; | |
340 tname = *fs; | |
341 } | |
342 for (sid = 0; sid < track->n_styles; ++sid) { | |
343 if (style == NULL || strcasecmp(track->styles[sid].Name, style) == 0) { | |
344 target = track->styles + sid; | |
345 if (0) { | |
346 STRVAL(FontName) | |
347 COLORVAL(PrimaryColour) | |
348 COLORVAL(SecondaryColour) | |
349 COLORVAL(OutlineColour) | |
350 COLORVAL(BackColour) | |
351 INTVAL(FontSize) | |
352 INTVAL(Bold) | |
353 INTVAL(Italic) | |
354 INTVAL(Underline) | |
355 INTVAL(StrikeOut) | |
356 INTVAL(Spacing) | |
357 INTVAL(Angle) | |
358 INTVAL(BorderStyle) | |
359 INTVAL(Alignment) | |
360 INTVAL(MarginL) | |
361 INTVAL(MarginR) | |
362 INTVAL(MarginV) | |
363 INTVAL(Encoding) | |
364 FPVAL(ScaleX) | |
365 FPVAL(ScaleY) | |
366 FPVAL(Outline) | |
367 FPVAL(Shadow) | |
368 } | |
369 } | |
370 } | |
371 *eq = '='; | |
372 if (dt) *dt = '.'; | |
373 } | |
374 } | |
375 | |
376 /** | |
18937 | 377 * \brief Parse the Style line |
378 * \param track track | |
379 * \param str string to parse, zero-terminated | |
380 * Allocates a new style struct. | |
381 */ | |
382 static int process_style(ass_track_t* track, char *str) | |
383 { | |
384 | |
385 char* token; | |
386 char* tname; | |
387 char* p = str; | |
388 char* format; | |
389 char* q; // format scanning pointer | |
390 int sid; | |
391 ass_style_t* style; | |
392 ass_style_t* target; | |
393 | |
394 if (!track->style_format) { | |
395 // no style format header | |
396 // probably an ancient script version | |
397 if (track->track_type == TRACK_TYPE_SSA) | |
398 track->style_format = strdup("Name, Fontname, Fontsize, PrimaryColour, SecondaryColour," | |
399 "TertiaryColour, BackColour, Bold, Italic, BorderStyle, Outline," | |
400 "Shadow, Alignment, MarginL, MarginR, MarginV, AlphaLevel, Encoding"); | |
401 else | |
402 track->style_format = strdup("Name, Fontname, Fontsize, PrimaryColour, SecondaryColour," | |
403 "OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut," | |
404 "ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow," | |
405 "Alignment, MarginL, MarginR, MarginV, Encoding"); | |
406 } | |
407 | |
408 q = format = strdup(track->style_format); | |
409 | |
410 mp_msg(MSGT_GLOBAL, MSGL_V, "[%p] Style: %s\n", track, str); | |
411 | |
412 sid = ass_alloc_style(track); | |
413 | |
414 style = track->styles + sid; | |
415 target = style; | |
416 // fill style with some default values | |
417 style->ScaleX = 100.; | |
418 style->ScaleY = 100.; | |
419 | |
420 while (1) { | |
421 NEXT(q, tname); | |
422 NEXT(p, token); | |
423 | |
424 // ALIAS(TertiaryColour,OutlineColour) // ignore TertiaryColour; it appears only in SSA, and is overridden by BackColour | |
425 | |
426 if (0) { // cool ;) | |
427 STRVAL(Name) | |
428 if ((strcmp(target->Name, "Default")==0) || (strcmp(target->Name, "*Default")==0)) | |
429 track->default_style = sid; | |
430 STRVAL(FontName) | |
431 COLORVAL(PrimaryColour) | |
432 COLORVAL(SecondaryColour) | |
433 COLORVAL(OutlineColour) // TertiaryColor | |
434 COLORVAL(BackColour) | |
435 // SSA uses BackColour for both outline and shadow | |
436 // this will destroy SSA's TertiaryColour, but i'm not going to use it anyway | |
437 if (track->track_type == TRACK_TYPE_SSA) | |
438 target->OutlineColour = target->BackColour; | |
439 INTVAL(FontSize) | |
440 INTVAL(Bold) | |
441 INTVAL(Italic) | |
442 INTVAL(Underline) | |
443 INTVAL(StrikeOut) | |
444 INTVAL(Spacing) | |
445 INTVAL(Angle) | |
446 INTVAL(BorderStyle) | |
447 INTVAL(Alignment) | |
448 if (track->track_type == TRACK_TYPE_ASS) | |
449 target->Alignment = numpad2align(target->Alignment); | |
450 INTVAL(MarginL) | |
451 INTVAL(MarginR) | |
452 INTVAL(MarginV) | |
453 INTVAL(Encoding) | |
454 FPVAL(ScaleX) | |
455 FPVAL(ScaleY) | |
456 FPVAL(Outline) | |
457 FPVAL(Shadow) | |
458 } | |
459 } | |
460 style->ScaleX /= 100.; | |
461 style->ScaleY /= 100.; | |
462 if (!style->Name) | |
463 style->Name = strdup("Default"); | |
464 if (!style->FontName) | |
465 style->FontName = strdup("Arial"); | |
466 free(format); | |
467 return 0; | |
468 | |
469 } | |
470 | |
19492 | 471 static int process_styles_line(ass_track_t* track, char *str) |
472 { | |
473 if (!strncmp(str,"Format:", 7)) { | |
474 char* p = str + 7; | |
475 skip_spaces(&p); | |
476 track->style_format = strdup(p); | |
477 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "Style format: %s\n", track->style_format); | |
478 } else if (!strncmp(str,"Style:", 6)) { | |
479 char* p = str + 6; | |
480 skip_spaces(&p); | |
481 process_style(track, p); | |
482 } | |
483 return 0; | |
484 } | |
485 | |
486 static int process_info_line(ass_track_t* track, char *str) | |
487 { | |
488 if (!strncmp(str, "PlayResX:", 9)) { | |
489 track->PlayResX = atoi(str + 9); | |
490 } else if (!strncmp(str,"PlayResY:", 9)) { | |
491 track->PlayResY = atoi(str + 9); | |
492 } else if (!strncmp(str,"Timer:", 6)) { | |
493 track->Timer = atof(str + 6); | |
494 } else if (!strncmp(str,"WrapStyle:", 10)) { | |
495 track->WrapStyle = atoi(str + 10); | |
496 } | |
497 return 0; | |
498 } | |
499 | |
500 static int process_events_line(ass_track_t* track, char *str) | |
501 { | |
502 if (!strncmp(str, "Format:", 7)) { | |
503 char* p = str + 7; | |
504 skip_spaces(&p); | |
505 track->event_format = strdup(p); | |
506 mp_msg(MSGT_GLOBAL, MSGL_DBG2, "Event format: %s\n", track->event_format); | |
507 } else if (!strncmp(str, "Dialogue:", 9)) { | |
508 // This should never be reached for embedded subtitles. | |
509 // They have slightly different format and are parsed in ass_process_chunk, | |
510 // called directly from demuxer | |
511 int eid; | |
512 ass_event_t* event; | |
513 | |
514 str += 9; | |
515 skip_spaces(&str); | |
516 | |
517 eid = ass_alloc_event(track); | |
518 event = track->events + eid; | |
519 | |
520 process_event_tail(track, event, str, 0); | |
521 } else { | |
522 mp_msg(MSGT_GLOBAL, MSGL_V, "Not understood: %s \n", str); | |
523 } | |
524 return 0; | |
525 } | |
526 | |
527 // Copied from mkvtoolnix | |
528 static unsigned char* decode_chars(unsigned char c1, unsigned char c2, | |
529 unsigned char c3, unsigned char c4, unsigned char* dst, int cnt) | |
530 { | |
531 uint32_t value; | |
532 unsigned char bytes[3]; | |
533 int i; | |
534 | |
535 value = ((c1 - 33) << 18) + ((c2 - 33) << 12) + ((c3 - 33) << 6) + (c4 - 33); | |
536 bytes[2] = value & 0xff; | |
537 bytes[1] = (value & 0xff00) >> 8; | |
538 bytes[0] = (value & 0xff0000) >> 16; | |
539 | |
540 for (i = 0; i < cnt; ++i) | |
541 *dst++ = bytes[i]; | |
542 return dst; | |
543 } | |
544 | |
545 static int decode_font(ass_track_t* track) | |
546 { | |
547 unsigned char* p; | |
548 unsigned char* q; | |
549 int i; | |
550 int size; // original size | |
551 int dsize; // decoded size | |
552 unsigned char* buf = 0; | |
553 | |
554 mp_msg(MSGT_GLOBAL, MSGL_V, "font: %d bytes encoded data \n", track->parser_priv->fontdata_used); | |
555 size = track->parser_priv->fontdata_used; | |
556 if (size % 4 == 1) { | |
557 mp_msg(MSGT_GLOBAL, MSGL_ERR, "bad encoded data size\n"); | |
558 goto error_decode_font; | |
559 } | |
560 buf = malloc(size / 4 * 3 + 2); | |
561 q = buf; | |
562 for (i = 0, p = (unsigned char*)track->parser_priv->fontdata; i < size / 4; i++, p+=4) { | |
563 q = decode_chars(p[0], p[1], p[2], p[3], q, 3); | |
564 } | |
565 if (size % 4 == 2) { | |
566 q = decode_chars(p[0], p[1], 0, 0, q, 1); | |
567 } else if (size % 4 == 3) { | |
568 q = decode_chars(p[0], p[1], p[2], 0, q, 2); | |
569 } | |
570 dsize = q - buf; | |
571 assert(dsize <= size / 4 * 3 + 2); | |
572 | |
573 if (extract_embedded_fonts) | |
574 ass_process_font(track->parser_priv->fontname, (char*)buf, dsize); | |
575 | |
576 error_decode_font: | |
577 if (buf) free(buf); | |
578 free(track->parser_priv->fontname); | |
579 free(track->parser_priv->fontdata); | |
580 track->parser_priv->fontname = 0; | |
581 track->parser_priv->fontdata = 0; | |
582 track->parser_priv->fontdata_size = 0; | |
583 track->parser_priv->fontdata_used = 0; | |
584 return 0; | |
585 } | |
586 | |
587 static char* validate_fname(char* name); | |
588 | |
589 static int process_fonts_line(ass_track_t* track, char *str) | |
590 { | |
591 int len; | |
592 | |
593 if (!strncmp(str, "fontname:", 9)) { | |
594 char* p = str + 9; | |
595 skip_spaces(&p); | |
596 if (track->parser_priv->fontname) { | |
597 decode_font(track); | |
598 } | |
599 track->parser_priv->fontname = validate_fname(p); | |
600 mp_msg(MSGT_GLOBAL, MSGL_V, "fontname: %s\n", track->parser_priv->fontname); | |
601 return 0; | |
602 } | |
603 | |
604 if (!track->parser_priv->fontname) { | |
605 mp_msg(MSGT_GLOBAL, MSGL_V, "Not understood: %s \n", str); | |
606 return 0; | |
607 } | |
608 | |
609 len = strlen(str); | |
610 if (len > 80) { | |
611 mp_msg(MSGT_GLOBAL, MSGL_WARN, "Font line too long: %d, %s\n", len, str); | |
612 return 0; | |
613 } | |
614 if (track->parser_priv->fontdata_used + len > track->parser_priv->fontdata_size) { | |
615 track->parser_priv->fontdata_size += 100 * 1024; | |
616 track->parser_priv->fontdata = realloc(track->parser_priv->fontdata, track->parser_priv->fontdata_size); | |
617 } | |
618 memcpy(track->parser_priv->fontdata + track->parser_priv->fontdata_used, str, len); | |
619 track->parser_priv->fontdata_used += len; | |
620 | |
621 return 0; | |
622 } | |
623 | |
18937 | 624 /** |
625 * \brief Parse a header line | |
626 * \param track track | |
627 * \param str string to parse, zero-terminated | |
628 */ | |
19492 | 629 static int process_line(ass_track_t* track, char *str) |
18937 | 630 { |
19492 | 631 if (strstr(str, "[Script Info]")) { // FIXME: strstr to skip possible BOM at the beginning of the script |
632 track->parser_priv->state = PST_INFO; | |
633 } else if (!strncmp(str, "[V4 Styles]", 11)) { | |
634 track->parser_priv->state = PST_STYLES; | |
635 track->track_type = TRACK_TYPE_SSA; | |
636 } else if (!strncmp(str, "[V4+ Styles]", 12)) { | |
637 track->parser_priv->state = PST_STYLES; | |
638 track->track_type = TRACK_TYPE_ASS; | |
639 } else if (!strncmp(str, "[Events]", 8)) { | |
640 track->parser_priv->state = PST_EVENTS; | |
641 } else if (!strncmp(str, "[Fonts]", 7)) { | |
642 track->parser_priv->state = PST_FONTS; | |
643 } else { | |
644 switch (track->parser_priv->state) { | |
645 case PST_INFO: | |
646 process_info_line(track, str); | |
647 break; | |
648 case PST_STYLES: | |
649 process_styles_line(track, str); | |
650 break; | |
651 case PST_EVENTS: | |
652 process_events_line(track, str); | |
653 break; | |
654 case PST_FONTS: | |
655 process_fonts_line(track, str); | |
656 break; | |
657 default: | |
658 break; | |
18937 | 659 } |
19492 | 660 } |
661 | |
662 // there is no explicit end-of-font marker in ssa/ass | |
663 if ((track->parser_priv->state != PST_FONTS) && (track->parser_priv->fontname)) | |
664 decode_font(track); | |
665 | |
666 return 0; | |
667 } | |
668 | |
669 static int process_text(ass_track_t* track, char* str) | |
670 { | |
671 char* p = str; | |
672 while(1) { | |
673 char* q; | |
674 for (;((*p=='\r')||(*p=='\n'));++p) {} | |
675 for (q=p; ((*q!='\0')&&(*q!='\r')&&(*q!='\n')); ++q) {}; | |
676 if (q==p) | |
677 break; | |
678 if (*q != '\0') | |
679 *(q++) = '\0'; | |
680 process_line(track, p); | |
681 if (*q == '\0') | |
682 break; | |
683 p = q; | |
18937 | 684 } |
685 return 0; | |
686 } | |
687 | |
688 /** | |
689 * \brief Process CodecPrivate section of subtitle stream | |
690 * \param track track | |
691 * \param data string to parse | |
692 * \param size length of data | |
19492 | 693 CodecPrivate section contains [Stream Info] and [V4+ Styles] ([V4 Styles] for SSA) sections |
18937 | 694 */ |
19492 | 695 void ass_process_codec_private(ass_track_t* track, char *data, int size) |
18937 | 696 { |
697 char* str = malloc(size + 1); | |
698 int sid; | |
699 | |
700 memcpy(str, data, size); | |
701 str[size] = '\0'; | |
702 | |
19492 | 703 process_text(track, str); |
18937 | 704 free(str); |
705 | |
706 // add "Default" style to the end | |
707 // will be used if track does not contain a default style (or even does not contain styles at all) | |
708 sid = ass_alloc_style(track); | |
709 track->styles[sid].Name = strdup("Default"); | |
710 track->styles[sid].FontName = strdup("Arial"); | |
711 | |
712 if (!track->event_format) { | |
713 // probably an mkv produced by ancient mkvtoolnix | |
714 // such files don't have [Events] and Format: headers | |
19492 | 715 track->parser_priv->state = PST_EVENTS; |
18937 | 716 if (track->track_type == TRACK_TYPE_SSA) |
717 track->event_format = strdup("Format: Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text"); | |
718 else | |
719 track->event_format = strdup("Format: Layer, Start, End, Style, Actor, MarginL, MarginR, MarginV, Effect, Text"); | |
720 } | |
19495 | 721 |
722 process_force_style(track); | |
18937 | 723 } |
724 | |
725 static int check_duplicate_event(ass_track_t* track, int ReadOrder) | |
726 { | |
727 int i; | |
728 for (i = 0; i<track->n_events - 1; ++i) // ignoring last event, it is the one we are comparing with | |
729 if (track->events[i].ReadOrder == ReadOrder) | |
730 return 1; | |
731 return 0; | |
732 } | |
733 | |
734 /** | |
735 * \brief Process a chunk of subtitle stream data. In matroska, this containes exactly 1 event (or a commentary) | |
736 * \param track track | |
737 * \param data string to parse | |
738 * \param size length of data | |
739 * \param timecode starting time of the event (milliseconds) | |
740 * \param duration duration of the event (milliseconds) | |
741 */ | |
19492 | 742 void ass_process_chunk(ass_track_t* track, char *data, int size, long long timecode, long long duration) |
18937 | 743 { |
744 char* str; | |
745 int eid; | |
746 char* p; | |
747 char* token; | |
748 ass_event_t* event; | |
749 | |
750 if (!track->event_format) { | |
751 mp_msg(MSGT_GLOBAL, MSGL_WARN, "Event format header missing\n"); | |
752 return; | |
753 } | |
754 | |
755 str = malloc(size + 1); | |
756 memcpy(str, data, size); | |
757 str[size] = '\0'; | |
19378 | 758 mp_msg(MSGT_GLOBAL, MSGL_V, "event at %" PRId64 ", +%" PRId64 ": %s \n", (int64_t)timecode, (int64_t)duration, str); |
18937 | 759 |
760 eid = ass_alloc_event(track); | |
761 event = track->events + eid; | |
762 | |
763 p = str; | |
764 | |
765 do { | |
766 NEXT(p, token); | |
767 event->ReadOrder = atoi(token); | |
768 if (check_duplicate_event(track, event->ReadOrder)) | |
769 break; | |
770 | |
771 NEXT(p, token); | |
772 event->Layer = atoi(token); | |
773 | |
774 process_event_tail(track, event, p, 3); | |
775 | |
776 event->Start = timecode; | |
777 event->Duration = duration; | |
778 | |
779 free(str); | |
780 return; | |
781 // dump_events(tid); | |
782 } while (0); | |
783 // some error | |
19474
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
784 ass_free_event(track, eid); |
07209f48e527
Add public functions for removal of styles and events.
eugeni
parents:
19405
diff
changeset
|
785 track->n_events--; |
18937 | 786 free(str); |
787 } | |
788 | |
789 #ifdef USE_ICONV | |
790 /** \brief recode buffer to utf-8 | |
791 * constraint: sub_cp != 0 | |
792 * \param data pointer to text buffer | |
793 * \param size buffer size | |
794 * \return a pointer to recoded buffer, caller is responsible for freeing it | |
795 **/ | |
796 static char* sub_recode(char* data, size_t size) | |
797 { | |
798 static iconv_t icdsc = (iconv_t)(-1); | |
799 char* tocp = "UTF-8"; | |
800 char* outbuf; | |
801 assert(sub_cp); | |
802 | |
803 { | |
804 char* cp_tmp = sub_cp; | |
805 #ifdef HAVE_ENCA | |
806 char enca_lang[3], enca_fallback[100]; | |
807 if (sscanf(sub_cp, "enca:%2s:%99s", enca_lang, enca_fallback) == 2 | |
808 || sscanf(sub_cp, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) { | |
809 cp_tmp = guess_buffer_cp((unsigned char*)data, size, enca_lang, enca_fallback); | |
810 } | |
811 #endif | |
812 if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){ | |
813 mp_msg(MSGT_SUBREADER,MSGL_V,"LIBSUB: opened iconv descriptor.\n"); | |
814 } else | |
815 mp_msg(MSGT_SUBREADER,MSGL_ERR,"LIBSUB: error opening iconv descriptor.\n"); | |
816 #ifdef HAVE_ENCA | |
817 if (cp_tmp) free(cp_tmp); | |
818 #endif | |
819 } | |
820 | |
821 { | |
822 size_t osize = size; | |
823 size_t ileft = size; | |
824 size_t oleft = size - 1; | |
825 char* ip; | |
826 char* op; | |
827 size_t rc; | |
828 | |
829 outbuf = malloc(size); | |
830 ip = data; | |
831 op = outbuf; | |
832 | |
833 while (ileft) { | |
834 rc = iconv(icdsc, &ip, &ileft, &op, &oleft); | |
835 if (rc == (size_t)(-1)) { | |
836 if (errno == E2BIG) { | |
837 int offset = op - outbuf; | |
838 outbuf = (char*)realloc(outbuf, osize + size); | |
839 op = outbuf + offset; | |
840 osize += size; | |
841 oleft += size; | |
842 } else { | |
843 mp_msg(MSGT_SUBREADER, MSGL_WARN, "LIBSUB: error recoding file.\n"); | |
844 return NULL; | |
845 } | |
846 } | |
847 } | |
848 outbuf[osize - oleft - 1] = 0; | |
849 } | |
850 | |
851 if (icdsc != (iconv_t)(-1)) { | |
852 (void)iconv_close(icdsc); | |
853 icdsc = (iconv_t)(-1); | |
854 mp_msg(MSGT_SUBREADER,MSGL_V,"LIBSUB: closed iconv descriptor.\n"); | |
855 } | |
856 | |
857 return outbuf; | |
858 } | |
859 #endif // ICONV | |
860 | |
861 /** | |
862 * \brief Read subtitles from file. | |
863 * \param fname file name | |
864 * \return newly allocated track | |
865 */ | |
866 ass_track_t* ass_read_file(char* fname) | |
867 { | |
868 int res; | |
869 long sz; | |
870 long bytes_read; | |
871 char* buf; | |
872 ass_track_t* track; | |
873 | |
874 FILE* fp = fopen(fname, "rb"); | |
875 if (!fp) { | |
876 mp_msg(MSGT_GLOBAL, MSGL_WARN, "ass_read_file(%s): fopen failed\n", fname); | |
877 return 0; | |
878 } | |
879 res = fseek(fp, 0, SEEK_END); | |
880 if (res == -1) { | |
881 mp_msg(MSGT_GLOBAL, MSGL_WARN, "ass_read_file(%s): fseek failed\n", fname); | |
882 fclose(fp); | |
883 return 0; | |
884 } | |
885 | |
886 sz = ftell(fp); | |
887 rewind(fp); | |
888 | |
889 if (sz > 10*1024*1024) { | |
890 mp_msg(MSGT_GLOBAL, MSGL_INFO, "ass_read_file(%s): Refusing to load subtitles larger than 10M\n", fname); | |
891 fclose(fp); | |
892 return 0; | |
893 } | |
894 | |
895 mp_msg(MSGT_GLOBAL, MSGL_V, "file size: %ld\n", sz); | |
896 | |
897 buf = malloc(sz + 1); | |
898 assert(buf); | |
899 bytes_read = 0; | |
900 do { | |
901 res = fread(buf + bytes_read, 1, sz - bytes_read, fp); | |
902 if (res <= 0) { | |
903 mp_msg(MSGT_GLOBAL, MSGL_INFO, "Read failed, %d: %s\n", errno, strerror(errno)); | |
904 fclose(fp); | |
905 free(buf); | |
906 return 0; | |
907 } | |
908 bytes_read += res; | |
909 } while (sz - bytes_read > 0); | |
910 buf[sz] = '\0'; | |
911 fclose(fp); | |
912 | |
913 #ifdef USE_ICONV | |
914 if (sub_cp) { | |
915 char* tmpbuf = sub_recode(buf, sz); | |
916 free(buf); | |
917 if (!tmpbuf) | |
918 return 0; | |
919 buf = tmpbuf; | |
920 } | |
921 #endif | |
922 | |
923 track = ass_new_track(); | |
924 track->name = strdup(fname); | |
925 | |
926 // process header | |
19492 | 927 process_text(track, buf); |
928 | |
929 // there is no explicit end-of-font marker in ssa/ass | |
930 if (track->parser_priv->fontname) | |
931 decode_font(track); | |
932 | |
18937 | 933 free(buf); |
934 | |
19492 | 935 if (track->track_type == TRACK_TYPE_UNKNOWN) { |
18937 | 936 ass_free_track(track); |
937 return 0; | |
938 } | |
939 | |
19495 | 940 process_force_style(track); |
941 | |
18937 | 942 mp_msg(MSGT_GLOBAL, MSGL_INFO, "LIBASS: added subtitle file: %s (%d styles, %d events)\n", fname, track->n_styles, track->n_events); |
943 | |
944 sort_events(track); | |
945 | |
946 // dump_events(forced_tid); | |
947 return track; | |
948 } | |
949 | |
950 static char* validate_fname(char* name) | |
951 { | |
952 char* fname; | |
953 char* p; | |
954 char* q; | |
955 unsigned code; | |
956 int sz = strlen(name); | |
957 | |
958 q = fname = malloc(sz + 1); | |
959 p = name; | |
960 while (*p) { | |
961 code = utf8_get_char(&p); | |
962 if (code == 0) | |
963 break; | |
964 if ( (code > 0x7F) || | |
965 (code == '\\') || | |
966 (code == '/') || | |
967 (code == ':') || | |
968 (code == '*') || | |
969 (code == '?') || | |
970 (code == '<') || | |
971 (code == '>') || | |
972 (code == '|') || | |
973 (code == 0)) | |
974 { | |
975 *q++ = '_'; | |
976 } else { | |
977 *q++ = code; | |
978 } | |
979 if (p - name > sz) | |
980 break; | |
981 } | |
982 *q = 0; | |
983 return fname; | |
984 } | |
985 | |
986 /** | |
987 * \brief Process embedded matroska font. Saves it to ~/.mplayer/fonts. | |
988 * \param name attachment name | |
989 * \param data binary font data | |
990 * \param data_size data size | |
991 */ | |
992 void ass_process_font(const char* name, char* data, int data_size) | |
993 { | |
994 char buf[1000]; | |
995 FILE* fp = 0; | |
996 int rc; | |
997 struct stat st; | |
998 char* fname; | |
999 | |
1000 char* fonts_dir = get_path("fonts"); | |
1001 rc = stat(fonts_dir, &st); | |
1002 if (rc) { | |
1003 int res; | |
1004 #ifndef __MINGW32__ | |
1005 res = mkdir(fonts_dir, 0700); | |
1006 #else | |
1007 res = mkdir(fonts_dir); | |
1008 #endif | |
1009 if (res) { | |
1010 mp_msg(MSGT_GLOBAL, MSGL_WARN, "Failed to create: %s\n", fonts_dir); | |
1011 } | |
1012 } else if (!S_ISDIR(st.st_mode)) { | |
1013 mp_msg(MSGT_GLOBAL, MSGL_WARN, "Not a directory: %s\n", fonts_dir); | |
1014 } | |
1015 | |
1016 fname = validate_fname((char*)name); | |
1017 | |
1018 snprintf(buf, 1000, "%s/%s", fonts_dir, fname); | |
1019 free(fname); | |
1020 free(fonts_dir); | |
1021 | |
1022 fp = fopen(buf, "wb"); | |
1023 if (!fp) return; | |
1024 | |
1025 fwrite(data, data_size, 1, fp); | |
1026 fclose(fp); | |
1027 } | |
1028 | |
1029 long long ass_step_sub(ass_track_t* track, long long now, int movement) { | |
1030 int i; | |
1031 | |
1032 if (movement == 0) return 0; | |
1033 if (track->n_events == 0) return 0; | |
1034 | |
1035 if (movement < 0) | |
1036 for (i = 0; (i < track->n_events) && ((long long)(track->events[i].Start + track->events[i].Duration) <= now); ++i) {} | |
1037 else | |
1038 for (i = track->n_events - 1; (i >= 0) && ((long long)(track->events[i].Start) > now); --i) {} | |
1039 | |
1040 // -1 and n_events are ok | |
1041 assert(i >= -1); assert(i <= track->n_events); | |
1042 i += movement; | |
1043 if (i < 0) i = 0; | |
1044 if (i >= track->n_events) i = track->n_events - 1; | |
1045 return ((long long)track->events[i].Start) - now; | |
1046 } | |
1047 | |
1048 ass_track_t* ass_new_track(void) { | |
1049 ass_track_t* track = calloc(1, sizeof(ass_track_t)); | |
19492 | 1050 track->parser_priv = calloc(1, sizeof(parser_priv_t)); |
18937 | 1051 return track; |
1052 } | |
1053 |