GCC Code Coverage Report
Directory: ../../../ffmpeg/ Exec Total Coverage
File: src/libavcodec/movtextdec.c Lines: 129 280 46.1 %
Date: 2019-11-18 18:00:01 Branches: 47 165 28.5 %

Line Branch Exec Source
1
/*
2
 * 3GPP TS 26.245 Timed Text decoder
3
 * Copyright (c) 2012  Philip Langdale <philipl@overt.org>
4
 *
5
 * This file is part of FFmpeg.
6
 *
7
 * FFmpeg is free software; you can redistribute it and/or
8
 * modify it under the terms of the GNU Lesser General Public
9
 * License as published by the Free Software Foundation; either
10
 * version 2.1 of the License, or (at your option) any later version.
11
 *
12
 * FFmpeg is distributed in the hope that it will be useful,
13
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15
 * Lesser General Public License for more details.
16
 *
17
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with FFmpeg; if not, write to the Free Software
19
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20
 */
21
22
#include "avcodec.h"
23
#include "ass.h"
24
#include "libavutil/avstring.h"
25
#include "libavutil/common.h"
26
#include "libavutil/bprint.h"
27
#include "libavutil/intreadwrite.h"
28
#include "libavutil/mem.h"
29
30
#define STYLE_FLAG_BOLD         (1<<0)
31
#define STYLE_FLAG_ITALIC       (1<<1)
32
#define STYLE_FLAG_UNDERLINE    (1<<2)
33
34
#define BOX_SIZE_INITIAL    40
35
36
#define STYL_BOX   (1<<0)
37
#define HLIT_BOX   (1<<1)
38
#define HCLR_BOX   (1<<2)
39
#define TWRP_BOX   (1<<3)
40
41
#define BOTTOM_LEFT     1
42
#define BOTTOM_CENTER   2
43
#define BOTTOM_RIGHT    3
44
#define MIDDLE_LEFT     4
45
#define MIDDLE_CENTER   5
46
#define MIDDLE_RIGHT    6
47
#define TOP_LEFT        7
48
#define TOP_CENTER      8
49
#define TOP_RIGHT       9
50
51
typedef struct {
52
    char *font;
53
    int fontsize;
54
    int color;
55
    int back_color;
56
    int bold;
57
    int italic;
58
    int underline;
59
    int alignment;
60
} MovTextDefault;
61
62
typedef struct {
63
    uint16_t fontID;
64
    char *font;
65
} FontRecord;
66
67
typedef struct {
68
    uint16_t style_start;
69
    uint16_t style_end;
70
    uint8_t style_flag;
71
    uint8_t fontsize;
72
    uint16_t style_fontID;
73
} StyleBox;
74
75
typedef struct {
76
    uint16_t hlit_start;
77
    uint16_t hlit_end;
78
} HighlightBox;
79
80
typedef struct {
81
   uint8_t hlit_color[4];
82
} HilightcolorBox;
83
84
typedef struct {
85
    uint8_t wrap_flag;
86
} TextWrapBox;
87
88
typedef struct {
89
    StyleBox **s;
90
    StyleBox *s_temp;
91
    HighlightBox h;
92
    HilightcolorBox c;
93
    FontRecord **ftab;
94
    FontRecord *ftab_temp;
95
    TextWrapBox w;
96
    MovTextDefault d;
97
    uint8_t box_flags;
98
    uint16_t style_entries, ftab_entries;
99
    uint64_t tracksize;
100
    int size_var;
101
    int count_s, count_f;
102
    int readorder;
103
} MovTextContext;
104
105
typedef struct {
106
    uint32_t type;
107
    size_t base_size;
108
    int (*decode)(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt);
109
} Box;
110
111
15
static void mov_text_cleanup(MovTextContext *m)
112
{
113
    int i;
114
15
    if (m->box_flags & STYL_BOX) {
115
        for(i = 0; i < m->count_s; i++) {
116
            av_freep(&m->s[i]);
117
        }
118
        av_freep(&m->s);
119
        m->count_s = 0;
120
        m->style_entries = 0;
121
    }
122
15
}
123
124
6
static void mov_text_cleanup_ftab(MovTextContext *m)
125
{
126
    int i;
127
6
    if (m->ftab_temp)
128
        av_freep(&m->ftab_temp->font);
129
6
    av_freep(&m->ftab_temp);
130
6
    if (m->ftab) {
131
12
        for(i = 0; i < m->count_f; i++) {
132
6
            av_freep(&m->ftab[i]->font);
133
6
            av_freep(&m->ftab[i]);
134
        }
135
    }
136
6
    av_freep(&m->ftab);
137
6
}
138
139
6
static int mov_text_tx3g(AVCodecContext *avctx, MovTextContext *m)
140
{
141
6
    uint8_t *tx3g_ptr = avctx->extradata;
142
    int i, box_size, font_length;
143
    int8_t v_align, h_align;
144
    int style_fontID;
145
    StyleBox s_default;
146
147
6
    m->count_f = 0;
148
6
    m->ftab_entries = 0;
149
6
    box_size = BOX_SIZE_INITIAL; /* Size till ftab_entries */
150
6
    if (avctx->extradata_size < box_size)
151
        return -1;
152
153
    // Display Flags
154
6
    tx3g_ptr += 4;
155
    // Alignment
156
6
    h_align = *tx3g_ptr++;
157
6
    v_align = *tx3g_ptr++;
158
6
    if (h_align == 0) {
159
        if (v_align == 0)
160
            m->d.alignment = TOP_LEFT;
161
        if (v_align == 1)
162
            m->d.alignment = MIDDLE_LEFT;
163
        if (v_align == -1)
164
            m->d.alignment = BOTTOM_LEFT;
165
    }
166
6
    if (h_align == 1) {
167
6
        if (v_align == 0)
168
            m->d.alignment = TOP_CENTER;
169
6
        if (v_align == 1)
170
            m->d.alignment = MIDDLE_CENTER;
171
6
        if (v_align == -1)
172
6
            m->d.alignment = BOTTOM_CENTER;
173
    }
174
6
    if (h_align == -1) {
175
        if (v_align == 0)
176
            m->d.alignment = TOP_RIGHT;
177
        if (v_align == 1)
178
            m->d.alignment = MIDDLE_RIGHT;
179
        if (v_align == -1)
180
            m->d.alignment = BOTTOM_RIGHT;
181
    }
182
    // Background Color
183
6
    m->d.back_color = AV_RB24(tx3g_ptr);
184
6
    tx3g_ptr += 4;
185
    // BoxRecord
186
6
    tx3g_ptr += 8;
187
    // StyleRecord
188
6
    tx3g_ptr += 4;
189
    // fontID
190
6
    style_fontID = AV_RB16(tx3g_ptr);
191
6
    tx3g_ptr += 2;
192
    // face-style-flags
193
6
    s_default.style_flag = *tx3g_ptr++;
194
6
    m->d.bold = s_default.style_flag & STYLE_FLAG_BOLD;
195
6
    m->d.italic = s_default.style_flag & STYLE_FLAG_ITALIC;
196
6
    m->d.underline = s_default.style_flag & STYLE_FLAG_UNDERLINE;
197
    // fontsize
198
6
    m->d.fontsize = *tx3g_ptr++;
199
    // Primary color
200
6
    m->d.color = AV_RB24(tx3g_ptr);
201
6
    tx3g_ptr += 4;
202
    // FontRecord
203
    // FontRecord Size
204
6
    tx3g_ptr += 4;
205
    // ftab
206
6
    tx3g_ptr += 4;
207
208
6
    m->ftab_entries = AV_RB16(tx3g_ptr);
209
6
    tx3g_ptr += 2;
210
211
12
    for (i = 0; i < m->ftab_entries; i++) {
212
213
6
        box_size += 3;
214
6
        if (avctx->extradata_size < box_size) {
215
            mov_text_cleanup_ftab(m);
216
            m->ftab_entries = 0;
217
            return -1;
218
        }
219
6
        m->ftab_temp = av_mallocz(sizeof(*m->ftab_temp));
220
6
        if (!m->ftab_temp) {
221
            mov_text_cleanup_ftab(m);
222
            return AVERROR(ENOMEM);
223
        }
224
6
        m->ftab_temp->fontID = AV_RB16(tx3g_ptr);
225
6
        tx3g_ptr += 2;
226
6
        font_length = *tx3g_ptr++;
227
228
6
        box_size = box_size + font_length;
229
6
        if (avctx->extradata_size < box_size) {
230
            mov_text_cleanup_ftab(m);
231
            m->ftab_entries = 0;
232
            return -1;
233
        }
234
6
        m->ftab_temp->font = av_malloc(font_length + 1);
235
6
        if (!m->ftab_temp->font) {
236
            mov_text_cleanup_ftab(m);
237
            return AVERROR(ENOMEM);
238
        }
239
6
        memcpy(m->ftab_temp->font, tx3g_ptr, font_length);
240
6
        m->ftab_temp->font[font_length] = '\0';
241
6
        av_dynarray_add(&m->ftab, &m->count_f, m->ftab_temp);
242
6
        if (!m->ftab) {
243
            mov_text_cleanup_ftab(m);
244
            return AVERROR(ENOMEM);
245
        }
246
6
        m->ftab_temp = NULL;
247
6
        tx3g_ptr = tx3g_ptr + font_length;
248
    }
249
12
    for (i = 0; i < m->ftab_entries; i++) {
250
6
        if (style_fontID == m->ftab[i]->fontID)
251
6
            m->d.font = m->ftab[i]->font;
252
    }
253
6
    return 0;
254
}
255
256
static int decode_twrp(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
257
{
258
    m->box_flags |= TWRP_BOX;
259
    m->w.wrap_flag = *tsmb++;
260
    return 0;
261
}
262
263
static int decode_hlit(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
264
{
265
    m->box_flags |= HLIT_BOX;
266
    m->h.hlit_start = AV_RB16(tsmb);
267
    tsmb += 2;
268
    m->h.hlit_end = AV_RB16(tsmb);
269
    tsmb += 2;
270
    return 0;
271
}
272
273
static int decode_hclr(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
274
{
275
    m->box_flags |= HCLR_BOX;
276
    memcpy(m->c.hlit_color, tsmb, 4);
277
    tsmb += 4;
278
    return 0;
279
}
280
281
static int decode_styl(const uint8_t *tsmb, MovTextContext *m, AVPacket *avpkt)
282
{
283
    int i;
284
    int style_entries = AV_RB16(tsmb);
285
    tsmb += 2;
286
    // A single style record is of length 12 bytes.
287
    if (m->tracksize + m->size_var + 2 + style_entries * 12 > avpkt->size)
288
        return -1;
289
290
    m->style_entries = style_entries;
291
292
    m->box_flags |= STYL_BOX;
293
    for(i = 0; i < m->style_entries; i++) {
294
        m->s_temp = av_malloc(sizeof(*m->s_temp));
295
        if (!m->s_temp) {
296
            mov_text_cleanup(m);
297
            return AVERROR(ENOMEM);
298
        }
299
        m->s_temp->style_start = AV_RB16(tsmb);
300
        tsmb += 2;
301
        m->s_temp->style_end = AV_RB16(tsmb);
302
303
        if (   m->s_temp->style_end < m->s_temp->style_start
304
            || (m->count_s && m->s_temp->style_start < m->s[m->count_s - 1]->style_end)) {
305
            av_freep(&m->s_temp);
306
            mov_text_cleanup(m);
307
            return AVERROR(ENOMEM);
308
        }
309
310
        tsmb += 2;
311
        m->s_temp->style_fontID = AV_RB16(tsmb);
312
        tsmb += 2;
313
        m->s_temp->style_flag = AV_RB8(tsmb);
314
        tsmb++;
315
        m->s_temp->fontsize = AV_RB8(tsmb);
316
        av_dynarray_add(&m->s, &m->count_s, m->s_temp);
317
        if(!m->s) {
318
            mov_text_cleanup(m);
319
            return AVERROR(ENOMEM);
320
        }
321
        tsmb++;
322
        // text-color-rgba
323
        tsmb += 4;
324
    }
325
    return 0;
326
}
327
328
static const Box box_types[] = {
329
    { MKBETAG('s','t','y','l'), 2, decode_styl },
330
    { MKBETAG('h','l','i','t'), 4, decode_hlit },
331
    { MKBETAG('h','c','l','r'), 4, decode_hclr },
332
    { MKBETAG('t','w','r','p'), 1, decode_twrp }
333
};
334
335
const static size_t box_count = FF_ARRAY_ELEMS(box_types);
336
337
// Return byte length of the UTF-8 sequence starting at text[0]. 0 on error.
338
135
static int get_utf8_length_at(const char *text, const char *text_end)
339
{
340
135
    const char *start = text;
341
135
    int err = 0;
342
    uint32_t c;
343



135
    GET_UTF8(c, text < text_end ? (uint8_t)*text++ : (err = 1, 0), goto error;);
344
135
    if (err)
345
        goto error;
346
135
    return text - start;
347
error:
348
    return 0;
349
}
350
351
9
static int text_to_ass(AVBPrint *buf, const char *text, const char *text_end,
352
                       AVCodecContext *avctx)
353
{
354
9
    MovTextContext *m = avctx->priv_data;
355
9
    int i = 0;
356
9
    int j = 0;
357
9
    int text_pos = 0;
358
359

9
    if (text < text_end && m->box_flags & TWRP_BOX) {
360
        if (m->w.wrap_flag == 1) {
361
            av_bprintf(buf, "{\\q1}"); /* End of line wrap */
362
        } else {
363
            av_bprintf(buf, "{\\q2}"); /* No wrap */
364
        }
365
    }
366
367
144
    while (text < text_end) {
368
        int len;
369
370
135
        if (m->box_flags & STYL_BOX) {
371
            for (i = 0; i < m->style_entries; i++) {
372
                if (m->s[i]->style_flag && text_pos == m->s[i]->style_end) {
373
                    av_bprintf(buf, "{\\r}");
374
                }
375
            }
376
            for (i = 0; i < m->style_entries; i++) {
377
                if (m->s[i]->style_flag && text_pos == m->s[i]->style_start) {
378
                    if (m->s[i]->style_flag & STYLE_FLAG_BOLD)
379
                        av_bprintf(buf, "{\\b1}");
380
                    if (m->s[i]->style_flag & STYLE_FLAG_ITALIC)
381
                        av_bprintf(buf, "{\\i1}");
382
                    if (m->s[i]->style_flag & STYLE_FLAG_UNDERLINE)
383
                        av_bprintf(buf, "{\\u1}");
384
                    av_bprintf(buf, "{\\fs%d}", m->s[i]->fontsize);
385
                    for (j = 0; j < m->ftab_entries; j++) {
386
                        if (m->s[i]->style_fontID == m->ftab[j]->fontID)
387
                            av_bprintf(buf, "{\\fn%s}", m->ftab[j]->font);
388
                    }
389
                }
390
            }
391
        }
392
135
        if (m->box_flags & HLIT_BOX) {
393
            if (text_pos == m->h.hlit_start) {
394
                /* If hclr box is present, set the secondary color to the color
395
                 * specified. Otherwise, set primary color to white and secondary
396
                 * color to black. These colors will come from TextSampleModifier
397
                 * boxes in future and inverse video technique for highlight will
398
                 * be implemented.
399
                 */
400
                if (m->box_flags & HCLR_BOX) {
401
                    av_bprintf(buf, "{\\2c&H%02x%02x%02x&}", m->c.hlit_color[2],
402
                                m->c.hlit_color[1], m->c.hlit_color[0]);
403
                } else {
404
                    av_bprintf(buf, "{\\1c&H000000&}{\\2c&HFFFFFF&}");
405
                }
406
            }
407
            if (text_pos == m->h.hlit_end) {
408
                if (m->box_flags & HCLR_BOX) {
409
                    av_bprintf(buf, "{\\2c&H000000&}");
410
                } else {
411
                    av_bprintf(buf, "{\\1c&HFFFFFF&}{\\2c&H000000&}");
412
                }
413
            }
414
        }
415
416
135
        len = get_utf8_length_at(text, text_end);
417
135
        if (len < 1) {
418
            av_log(avctx, AV_LOG_ERROR, "invalid UTF-8 byte in subtitle\n");
419
            len = 1;
420
        }
421
270
        for (i = 0; i < len; i++) {
422
135
            switch (*text) {
423
            case '\r':
424
                break;
425
6
            case '\n':
426
6
                av_bprintf(buf, "\\N");
427
6
                break;
428
129
            default:
429
129
                av_bprint_chars(buf, *text, 1);
430
129
                break;
431
            }
432
135
            text++;
433
        }
434
135
        text_pos++;
435
    }
436
437
9
    return 0;
438
}
439
440
6
static int mov_text_init(AVCodecContext *avctx) {
441
    /*
442
     * TODO: Handle the default text style.
443
     * NB: Most players ignore styles completely, with the result that
444
     * it's very common to find files where the default style is broken
445
     * and respecting it results in a worse experience than ignoring it.
446
     */
447
    int ret;
448
6
    MovTextContext *m = avctx->priv_data;
449
6
    ret = mov_text_tx3g(avctx, m);
450
6
    if (ret == 0) {
451
6
        return ff_ass_subtitle_header(avctx, m->d.font, m->d.fontsize, m->d.color,
452
                                m->d.back_color, m->d.bold, m->d.italic,
453
                                m->d.underline, ASS_DEFAULT_BORDERSTYLE,
454
                                m->d.alignment);
455
    } else
456
        return ff_ass_subtitle_header_default(avctx);
457
}
458
459
18
static int mov_text_decode_frame(AVCodecContext *avctx,
460
                            void *data, int *got_sub_ptr, AVPacket *avpkt)
461
{
462
18
    AVSubtitle *sub = data;
463
18
    MovTextContext *m = avctx->priv_data;
464
    int ret;
465
    AVBPrint buf;
466
18
    char *ptr = avpkt->data;
467
    char *end;
468
    int text_length, tsmb_type, ret_tsmb;
469
    uint64_t tsmb_size;
470
    const uint8_t *tsmb;
471
    size_t i;
472
473

18
    if (!ptr || avpkt->size < 2)
474
        return AVERROR_INVALIDDATA;
475
476
    /*
477
     * A packet of size two with value zero is an empty subtitle
478
     * used to mark the end of the previous non-empty subtitle.
479
     * We can just drop them here as we have duration information
480
     * already. If the value is non-zero, then it's technically a
481
     * bad packet.
482
     */
483
18
    if (avpkt->size == 2)
484
9
        return AV_RB16(ptr) == 0 ? 0 : AVERROR_INVALIDDATA;
485
486
    /*
487
     * The first two bytes of the packet are the length of the text string
488
     * In complex cases, there are style descriptors appended to the string
489
     * so we can't just assume the packet size is the string size.
490
     */
491
9
    text_length = AV_RB16(ptr);
492
9
    end = ptr + FFMIN(2 + text_length, avpkt->size);
493
9
    ptr += 2;
494
495
9
    mov_text_cleanup(m);
496
497
9
    tsmb_size = 0;
498
9
    m->tracksize = 2 + text_length;
499
9
    m->style_entries = 0;
500
9
    m->box_flags = 0;
501
9
    m->count_s = 0;
502
    // Note that the spec recommends lines be no longer than 2048 characters.
503
9
    av_bprint_init(&buf, 0, AV_BPRINT_SIZE_UNLIMITED);
504
9
    if (text_length + 2 != avpkt->size) {
505
        while (m->tracksize + 8 <= avpkt->size) {
506
            // A box is a minimum of 8 bytes.
507
            tsmb = ptr + m->tracksize - 2;
508
            tsmb_size = AV_RB32(tsmb);
509
            tsmb += 4;
510
            tsmb_type = AV_RB32(tsmb);
511
            tsmb += 4;
512
513
            if (tsmb_size == 1) {
514
                if (m->tracksize + 16 > avpkt->size)
515
                    break;
516
                tsmb_size = AV_RB64(tsmb);
517
                tsmb += 8;
518
                m->size_var = 16;
519
            } else
520
                m->size_var = 8;
521
            //size_var is equal to 8 or 16 depending on the size of box
522
523
            if (tsmb_size == 0) {
524
                av_log(avctx, AV_LOG_ERROR, "tsmb_size is 0\n");
525
                return AVERROR_INVALIDDATA;
526
            }
527
528
            if (tsmb_size > avpkt->size - m->tracksize)
529
                break;
530
531
            for (i = 0; i < box_count; i++) {
532
                if (tsmb_type == box_types[i].type) {
533
                    if (m->tracksize + m->size_var + box_types[i].base_size > avpkt->size)
534
                        break;
535
                    ret_tsmb = box_types[i].decode(tsmb, m, avpkt);
536
                    if (ret_tsmb == -1)
537
                        break;
538
                }
539
            }
540
            m->tracksize = m->tracksize + tsmb_size;
541
        }
542
        text_to_ass(&buf, ptr, end, avctx);
543
        mov_text_cleanup(m);
544
    } else
545
9
        text_to_ass(&buf, ptr, end, avctx);
546
547
9
    ret = ff_ass_add_rect(sub, buf.str, m->readorder++, 0, NULL, NULL);
548
9
    av_bprint_finalize(&buf, NULL);
549
9
    if (ret < 0)
550
        return ret;
551
9
    *got_sub_ptr = sub->num_rects > 0;
552
9
    return avpkt->size;
553
}
554
555
6
static int mov_text_decode_close(AVCodecContext *avctx)
556
{
557
6
    MovTextContext *m = avctx->priv_data;
558
6
    mov_text_cleanup_ftab(m);
559
6
    mov_text_cleanup(m);
560
6
    return 0;
561
}
562
563
static void mov_text_flush(AVCodecContext *avctx)
564
{
565
    MovTextContext *m = avctx->priv_data;
566
    if (!(avctx->flags2 & AV_CODEC_FLAG2_RO_FLUSH_NOOP))
567
        m->readorder = 0;
568
}
569
570
AVCodec ff_movtext_decoder = {
571
    .name         = "mov_text",
572
    .long_name    = NULL_IF_CONFIG_SMALL("3GPP Timed Text subtitle"),
573
    .type         = AVMEDIA_TYPE_SUBTITLE,
574
    .id           = AV_CODEC_ID_MOV_TEXT,
575
    .priv_data_size = sizeof(MovTextContext),
576
    .init         = mov_text_init,
577
    .decode       = mov_text_decode_frame,
578
    .close        = mov_text_decode_close,
579
    .flush        = mov_text_flush,
580
};