1 /*
2 * Copyright (c) 2012-2013 Clément Bœsch <u pkh me>
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 #include "avformat.h"
22 #include "subtitles.h"
23 #include "avio_internal.h"
24 #include "libavutil/avassert.h"
25 #include "libavutil/avstring.h"
26
ff_text_init_avio(void * s,FFTextReader * r,AVIOContext * pb)27 void ff_text_init_avio(void *s, FFTextReader *r, AVIOContext *pb)
28 {
29 int i;
30 r->pb = pb;
31 r->buf_pos = r->buf_len = 0;
32 r->type = FF_UTF_8;
33 for (i = 0; i < 2; i++)
34 r->buf[r->buf_len++] = avio_r8(r->pb);
35 if (strncmp("\xFF\xFE", r->buf, 2) == 0) {
36 r->type = FF_UTF16LE;
37 r->buf_pos += 2;
38 } else if (strncmp("\xFE\xFF", r->buf, 2) == 0) {
39 r->type = FF_UTF16BE;
40 r->buf_pos += 2;
41 } else {
42 r->buf[r->buf_len++] = avio_r8(r->pb);
43 if (strncmp("\xEF\xBB\xBF", r->buf, 3) == 0) {
44 // UTF8
45 r->buf_pos += 3;
46 }
47 }
48 if (s && (r->type == FF_UTF16LE || r->type == FF_UTF16BE))
49 av_log(s, AV_LOG_INFO,
50 "UTF16 is automatically converted to UTF8, do not specify a character encoding\n");
51 }
52
ff_text_init_buf(FFTextReader * r,void * buf,size_t size)53 void ff_text_init_buf(FFTextReader *r, void *buf, size_t size)
54 {
55 memset(&r->buf_pb, 0, sizeof(r->buf_pb));
56 ffio_init_context(&r->buf_pb, buf, size, 0, NULL, NULL, NULL, NULL);
57 ff_text_init_avio(NULL, r, &r->buf_pb);
58 }
59
ff_text_pos(FFTextReader * r)60 int64_t ff_text_pos(FFTextReader *r)
61 {
62 return avio_tell(r->pb) - r->buf_len + r->buf_pos;
63 }
64
ff_text_r8(FFTextReader * r)65 int ff_text_r8(FFTextReader *r)
66 {
67 uint32_t val;
68 uint8_t tmp;
69 if (r->buf_pos < r->buf_len)
70 return r->buf[r->buf_pos++];
71 if (r->type == FF_UTF16LE) {
72 GET_UTF16(val, avio_rl16(r->pb), return 0;)
73 } else if (r->type == FF_UTF16BE) {
74 GET_UTF16(val, avio_rb16(r->pb), return 0;)
75 } else {
76 return avio_r8(r->pb);
77 }
78 if (!val)
79 return 0;
80 r->buf_pos = 0;
81 r->buf_len = 0;
82 PUT_UTF8(val, tmp, r->buf[r->buf_len++] = tmp;)
83 return r->buf[r->buf_pos++]; // buf_len is at least 1
84 }
85
ff_text_read(FFTextReader * r,char * buf,size_t size)86 void ff_text_read(FFTextReader *r, char *buf, size_t size)
87 {
88 for ( ; size > 0; size--)
89 *buf++ = ff_text_r8(r);
90 }
91
ff_text_eof(FFTextReader * r)92 int ff_text_eof(FFTextReader *r)
93 {
94 return r->buf_pos >= r->buf_len && avio_feof(r->pb);
95 }
96
ff_text_peek_r8(FFTextReader * r)97 int ff_text_peek_r8(FFTextReader *r)
98 {
99 int c;
100 if (r->buf_pos < r->buf_len)
101 return r->buf[r->buf_pos];
102 c = ff_text_r8(r);
103 if (!avio_feof(r->pb)) {
104 r->buf_pos = 0;
105 r->buf_len = 1;
106 r->buf[0] = c;
107 }
108 return c;
109 }
110
ff_subtitles_queue_insert(FFDemuxSubtitlesQueue * q,const uint8_t * event,size_t len,int merge)111 AVPacket *ff_subtitles_queue_insert(FFDemuxSubtitlesQueue *q,
112 const uint8_t *event, size_t len, int merge)
113 {
114 AVPacket *subs, *sub;
115
116 if (merge && q->nb_subs > 0) {
117 /* merge with previous event */
118
119 int old_len;
120 sub = &q->subs[q->nb_subs - 1];
121 old_len = sub->size;
122 if (av_grow_packet(sub, len) < 0)
123 return NULL;
124 memcpy(sub->data + old_len, event, len);
125 } else {
126 /* new event */
127
128 if (q->nb_subs >= INT_MAX/sizeof(*q->subs) - 1)
129 return NULL;
130 subs = av_fast_realloc(q->subs, &q->allocated_size,
131 (q->nb_subs + 1) * sizeof(*q->subs));
132 if (!subs)
133 return NULL;
134 q->subs = subs;
135 sub = &subs[q->nb_subs];
136 if (av_new_packet(sub, len) < 0)
137 return NULL;
138 q->nb_subs++;
139 sub->flags |= AV_PKT_FLAG_KEY;
140 sub->pts = sub->dts = 0;
141 memcpy(sub->data, event, len);
142 }
143 return sub;
144 }
145
cmp_pkt_sub_ts_pos(const void * a,const void * b)146 static int cmp_pkt_sub_ts_pos(const void *a, const void *b)
147 {
148 const AVPacket *s1 = a;
149 const AVPacket *s2 = b;
150 if (s1->pts == s2->pts)
151 return FFDIFFSIGN(s1->pos, s2->pos);
152 return FFDIFFSIGN(s1->pts , s2->pts);
153 }
154
cmp_pkt_sub_pos_ts(const void * a,const void * b)155 static int cmp_pkt_sub_pos_ts(const void *a, const void *b)
156 {
157 const AVPacket *s1 = a;
158 const AVPacket *s2 = b;
159 if (s1->pos == s2->pos) {
160 if (s1->pts == s2->pts)
161 return 0;
162 return s1->pts > s2->pts ? 1 : -1;
163 }
164 return s1->pos > s2->pos ? 1 : -1;
165 }
166
drop_dups(void * log_ctx,FFDemuxSubtitlesQueue * q)167 static void drop_dups(void *log_ctx, FFDemuxSubtitlesQueue *q)
168 {
169 int i, drop = 0;
170
171 for (i = 1; i < q->nb_subs; i++) {
172 const int last_id = i - 1 - drop;
173 const AVPacket *last = &q->subs[last_id];
174
175 if (q->subs[i].pts == last->pts &&
176 q->subs[i].duration == last->duration &&
177 q->subs[i].stream_index == last->stream_index &&
178 !strcmp(q->subs[i].data, last->data)) {
179
180 av_packet_unref(&q->subs[i]);
181 drop++;
182 } else if (drop) {
183 q->subs[last_id + 1] = q->subs[i];
184 memset(&q->subs[i], 0, sizeof(q->subs[i])); // for safety
185 }
186 }
187
188 if (drop) {
189 q->nb_subs -= drop;
190 av_log(log_ctx, AV_LOG_WARNING, "Dropping %d duplicated subtitle events\n", drop);
191 }
192 }
193
ff_subtitles_queue_finalize(void * log_ctx,FFDemuxSubtitlesQueue * q)194 void ff_subtitles_queue_finalize(void *log_ctx, FFDemuxSubtitlesQueue *q)
195 {
196 int i;
197
198 if (!q->nb_subs)
199 return;
200
201 qsort(q->subs, q->nb_subs, sizeof(*q->subs),
202 q->sort == SUB_SORT_TS_POS ? cmp_pkt_sub_ts_pos
203 : cmp_pkt_sub_pos_ts);
204 for (i = 0; i < q->nb_subs; i++)
205 if (q->subs[i].duration < 0 && i < q->nb_subs - 1)
206 q->subs[i].duration = q->subs[i + 1].pts - q->subs[i].pts;
207
208 if (!q->keep_duplicates)
209 drop_dups(log_ctx, q);
210 }
211
ff_subtitles_queue_read_packet(FFDemuxSubtitlesQueue * q,AVPacket * pkt)212 int ff_subtitles_queue_read_packet(FFDemuxSubtitlesQueue *q, AVPacket *pkt)
213 {
214 AVPacket *sub = q->subs + q->current_sub_idx;
215 int ret;
216
217 if (q->current_sub_idx == q->nb_subs)
218 return AVERROR_EOF;
219 if ((ret = av_packet_ref(pkt, sub)) < 0) {
220 return ret;
221 }
222
223 pkt->dts = pkt->pts;
224 q->current_sub_idx++;
225 return 0;
226 }
227
search_sub_ts(const FFDemuxSubtitlesQueue * q,int64_t ts)228 static int search_sub_ts(const FFDemuxSubtitlesQueue *q, int64_t ts)
229 {
230 int s1 = 0, s2 = q->nb_subs - 1;
231
232 if (s2 < s1)
233 return AVERROR(ERANGE);
234
235 for (;;) {
236 int mid;
237
238 if (s1 == s2)
239 return s1;
240 if (s1 == s2 - 1)
241 return q->subs[s1].pts <= q->subs[s2].pts ? s1 : s2;
242 mid = (s1 + s2) / 2;
243 if (q->subs[mid].pts <= ts)
244 s1 = mid;
245 else
246 s2 = mid;
247 }
248 }
249
ff_subtitles_queue_seek(FFDemuxSubtitlesQueue * q,AVFormatContext * s,int stream_index,int64_t min_ts,int64_t ts,int64_t max_ts,int flags)250 int ff_subtitles_queue_seek(FFDemuxSubtitlesQueue *q, AVFormatContext *s, int stream_index,
251 int64_t min_ts, int64_t ts, int64_t max_ts, int flags)
252 {
253 if (flags & AVSEEK_FLAG_BYTE) {
254 return AVERROR(ENOSYS);
255 } else if (flags & AVSEEK_FLAG_FRAME) {
256 if (ts < 0 || ts >= q->nb_subs)
257 return AVERROR(ERANGE);
258 q->current_sub_idx = ts;
259 } else {
260 int i, idx = search_sub_ts(q, ts);
261 int64_t ts_selected;
262
263 if (idx < 0)
264 return idx;
265 for (i = idx; i < q->nb_subs && q->subs[i].pts < min_ts; i++)
266 if (stream_index == -1 || q->subs[i].stream_index == stream_index)
267 idx = i;
268 for (i = idx; i > 0 && q->subs[i].pts > max_ts; i--)
269 if (stream_index == -1 || q->subs[i].stream_index == stream_index)
270 idx = i;
271
272 ts_selected = q->subs[idx].pts;
273 if (ts_selected < min_ts || ts_selected > max_ts)
274 return AVERROR(ERANGE);
275
276 /* look back in the latest subtitles for overlapping subtitles */
277 for (i = idx - 1; i >= 0; i--) {
278 int64_t pts = q->subs[i].pts;
279 if (q->subs[i].duration <= 0 ||
280 (stream_index != -1 && q->subs[i].stream_index != stream_index))
281 continue;
282 if (pts >= min_ts && pts > ts_selected - q->subs[i].duration)
283 idx = i;
284 else
285 break;
286 }
287
288 /* If the queue is used to store multiple subtitles streams (like with
289 * VobSub) and the stream index is not specified, we need to make sure
290 * to focus on the smallest file position offset for a same timestamp;
291 * queue is ordered by pts and then filepos, so we can take the first
292 * entry for a given timestamp. */
293 if (stream_index == -1)
294 while (idx > 0 && q->subs[idx - 1].pts == q->subs[idx].pts)
295 idx--;
296
297 q->current_sub_idx = idx;
298 }
299 return 0;
300 }
301
ff_subtitles_queue_clean(FFDemuxSubtitlesQueue * q)302 void ff_subtitles_queue_clean(FFDemuxSubtitlesQueue *q)
303 {
304 int i;
305
306 for (i = 0; i < q->nb_subs; i++)
307 av_packet_unref(&q->subs[i]);
308 av_freep(&q->subs);
309 q->nb_subs = q->allocated_size = q->current_sub_idx = 0;
310 }
311
ff_smil_extract_next_text_chunk(FFTextReader * tr,AVBPrint * buf,char * c)312 int ff_smil_extract_next_text_chunk(FFTextReader *tr, AVBPrint *buf, char *c)
313 {
314 int i = 0;
315 char end_chr;
316
317 if (!*c) // cached char?
318 *c = ff_text_r8(tr);
319 if (!*c)
320 return 0;
321
322 end_chr = *c == '<' ? '>' : '<';
323 do {
324 av_bprint_chars(buf, *c, 1);
325 *c = ff_text_r8(tr);
326 i++;
327 } while (*c != end_chr && *c);
328 if (end_chr == '>') {
329 av_bprint_chars(buf, '>', 1);
330 *c = 0;
331 }
332 return i;
333 }
334
ff_smil_get_attr_ptr(const char * s,const char * attr)335 const char *ff_smil_get_attr_ptr(const char *s, const char *attr)
336 {
337 int in_quotes = 0;
338 const size_t len = strlen(attr);
339
340 while (*s) {
341 while (*s) {
342 if (!in_quotes && av_isspace(*s))
343 break;
344 in_quotes ^= *s == '"'; // XXX: support escaping?
345 s++;
346 }
347 while (av_isspace(*s))
348 s++;
349 if (!av_strncasecmp(s, attr, len) && s[len] == '=')
350 return s + len + 1 + (s[len + 1] == '"');
351 }
352 return NULL;
353 }
354
is_eol(char c)355 static inline int is_eol(char c)
356 {
357 return c == '\r' || c == '\n';
358 }
359
ff_subtitles_read_text_chunk(FFTextReader * tr,AVBPrint * buf)360 void ff_subtitles_read_text_chunk(FFTextReader *tr, AVBPrint *buf)
361 {
362 char eol_buf[5], last_was_cr = 0;
363 int n = 0, i = 0, nb_eol = 0;
364
365 av_bprint_clear(buf);
366
367 for (;;) {
368 char c = ff_text_r8(tr);
369
370 if (!c)
371 break;
372
373 /* ignore all initial line breaks */
374 if (n == 0 && is_eol(c))
375 continue;
376
377 /* line break buffering: we don't want to add the trailing \r\n */
378 if (is_eol(c)) {
379 nb_eol += c == '\n' || last_was_cr;
380 if (nb_eol == 2)
381 break;
382 eol_buf[i++] = c;
383 if (i == sizeof(eol_buf) - 1)
384 break;
385 last_was_cr = c == '\r';
386 continue;
387 }
388
389 /* only one line break followed by data: we flush the line breaks
390 * buffer */
391 if (i) {
392 eol_buf[i] = 0;
393 av_bprintf(buf, "%s", eol_buf);
394 i = nb_eol = 0;
395 }
396
397 av_bprint_chars(buf, c, 1);
398 n++;
399 }
400 }
401
ff_subtitles_read_chunk(AVIOContext * pb,AVBPrint * buf)402 void ff_subtitles_read_chunk(AVIOContext *pb, AVBPrint *buf)
403 {
404 FFTextReader tr;
405 tr.buf_pos = tr.buf_len = 0;
406 tr.type = 0;
407 tr.pb = pb;
408 ff_subtitles_read_text_chunk(&tr, buf);
409 }
410
ff_subtitles_read_line(FFTextReader * tr,char * buf,size_t size)411 ptrdiff_t ff_subtitles_read_line(FFTextReader *tr, char *buf, size_t size)
412 {
413 size_t cur = 0;
414 if (!size)
415 return 0;
416 while (cur + 1 < size) {
417 unsigned char c = ff_text_r8(tr);
418 if (!c)
419 return ff_text_eof(tr) ? cur : AVERROR_INVALIDDATA;
420 if (c == '\r' || c == '\n')
421 break;
422 buf[cur++] = c;
423 buf[cur] = '\0';
424 }
425 while (ff_text_peek_r8(tr) == '\r')
426 ff_text_r8(tr);
427 if (ff_text_peek_r8(tr) == '\n')
428 ff_text_r8(tr);
429 return cur;
430 }
431