]> git.tuebingen.mpg.de Git - paraslash.git/blobdiff - mp4.c
mp4: Simplify parse_tag().
[paraslash.git] / mp4.c
diff --git a/mp4.c b/mp4.c
index d9f86fc7414685c5448252f8e8048c2979935d88..d8aea78379aeabc80987481229c2e537cef8a93f 100644 (file)
--- a/mp4.c
+++ b/mp4.c
 #include "mp4.h"
 
 struct mp4_track {
-       int32_t type;
+       bool is_audio;
        int32_t channelCount;
-       int32_t sampleSize;
        uint16_t sampleRate;
-       int32_t audioType;
-
-       /* stsd */
-       int32_t stsd_entry_count;
 
        /* stsz */
        int32_t stsz_sample_size;
@@ -42,18 +37,6 @@ struct mp4_track {
        int32_t stco_entry_count;
        int32_t *stco_chunk_offset;
 
-       /* ctts */
-       int32_t ctts_entry_count;
-       int32_t *ctts_sample_count;
-       int32_t *ctts_sample_offset;
-
-       /* esde */
-       uint8_t *decoderConfig;
-       int32_t decoderConfigLen;
-
-       uint32_t maxBitrate;
-       uint32_t avgBitrate;
-
        uint32_t timeScale;
        uint64_t duration;
 };
@@ -61,19 +44,16 @@ struct mp4_track {
 #define MAX_TRACKS 1024
 
 struct mp4 {
-       /* stream to read from */
-       struct mp4_callback *stream;
+       const struct mp4_callback *cb;
        int64_t current_position;
 
        uint64_t moov_offset;
        uint64_t moov_size;
        uint8_t last_atom;
        uint64_t file_size;
-       uint32_t error;
 
-       /* mvhd */
-       int32_t time_scale;
-       int32_t duration;
+       uint32_t read_error;
+       uint32_t error;
 
        /* incremental track index while reading the file */
        int32_t total_tracks;
@@ -94,10 +74,10 @@ static int32_t read_data(struct mp4 *f, void *data, uint32_t size)
 {
        int32_t result = 1;
 
-       result = f->stream->read(f->stream->user_data, data, size);
+       result = f->cb->read(f->cb->user_data, data, size);
 
        if (result < size)
-               f->stream->read_error++;
+               f->read_error++;
 
        f->current_position += size;
 
@@ -112,23 +92,11 @@ static uint64_t read_int64(struct mp4 *f)
        return read_u64_be(data);
 }
 
-/* comnapre 2 atom names, returns 1 for equal, 0 for unequal */
-static int32_t atom_compare(int8_t a1, int8_t b1, int8_t c1, int8_t d1,
+static bool atom_compare(int8_t a1, int8_t b1, int8_t c1, int8_t d1,
                int8_t a2, int8_t b2, int8_t c2, int8_t d2)
 {
-       if (a1 == a2 && b1 == b2 && c1 == c2 && d1 == d2)
-               return 1;
-       else
-               return 0;
+       return a1 == a2 && b1 == b2 && c1 == c2 && d1 == d2;
 }
-
-enum tracks {
-       TRACK_UNKNOWN = 0,
-       TRACK_AUDIO  = 1,
-       TRACK_VIDEO = 2,
-       TRACK_SYSTEM = 3
-};
-
 enum atoms {
        /* atoms with subatoms */
        ATOM_MOOV = 1,
@@ -420,7 +388,7 @@ static int need_parse_when_meta_only(uint8_t atom_type)
 
 static int32_t set_position(struct mp4 *f, int64_t position)
 {
-       f->stream->seek(f->stream->user_data, position);
+       f->cb->seek(f->cb->user_data, position);
        f->current_position = position;
 
        return 0;
@@ -476,7 +444,7 @@ static int32_t read_stsz(struct mp4 *f)
        if (t->stsz_sample_size != 0)
                return 0;
        t->stsz_table = para_malloc(t->stsz_sample_count * sizeof(int32_t));
-       for (i = 0; i < t->stsz_sample_count && !f->stream->read_error; i++)
+       for (i = 0; i < t->stsz_sample_count && !f->read_error; i++)
                t->stsz_table[i] = read_int32(f);
        return 0;
 }
@@ -501,41 +469,13 @@ static int32_t read_stts(struct mp4 *f)
        t->stts_sample_delta = para_malloc(t->stts_entry_count
                * sizeof (int32_t));
        /* CVE-2017-9254 */
-       for (i = 0; i < t->stts_entry_count && !f->stream->read_error; i++) {
+       for (i = 0; i < t->stts_entry_count && !f->read_error; i++) {
                t->stts_sample_count[i] = read_int32(f);
                t->stts_sample_delta[i] = read_int32(f);
        }
        return 1;
 }
 
-static int32_t read_ctts(struct mp4 *f)
-{
-       int32_t i;
-       struct mp4_track *t;
-
-       if (f->total_tracks == 0)
-               return f->error++;
-       t = f->track[f->total_tracks - 1];
-       if (t->ctts_entry_count)
-               return 0;
-
-       read_char(f);   /* version */
-       read_int24(f);  /* flags */
-       t->ctts_entry_count = read_int32(f);
-
-       t->ctts_sample_count = para_malloc(t->ctts_entry_count
-               * sizeof (int32_t));
-       t->ctts_sample_offset = para_malloc(t->ctts_entry_count
-               * sizeof (int32_t));
-
-       /* CVE-2017-9257 */
-       for (i = 0; i < t->ctts_entry_count && !f->stream->read_error; i++) {
-               t->ctts_sample_count[i] = read_int32(f);
-               t->ctts_sample_offset[i] = read_int32(f);
-       }
-       return 1;
-}
-
 static int32_t read_stsc(struct mp4 *f)
 {
        int32_t i;
@@ -555,7 +495,7 @@ static int32_t read_stsc(struct mp4 *f)
                sizeof (int32_t));
 
        /* CVE-2017-9255 */
-       for (i = 0; i < t->stsc_entry_count && !f->stream->read_error; i++) {
+       for (i = 0; i < t->stsc_entry_count && !f->read_error; i++) {
                t->stsc_first_chunk[i] = read_int32(f);
                t->stsc_samples_per_chunk[i] = read_int32(f);
                t->stsc_sample_desc_index[i] = read_int32(f);
@@ -578,7 +518,7 @@ static int32_t read_stco(struct mp4 *f)
        t->stco_chunk_offset = para_malloc(t->stco_entry_count
                * sizeof(int32_t));
        /* CVE-2017-9256 */
-       for (i = 0; i < t->stco_entry_count && !f->stream->read_error; i++)
+       for (i = 0; i < t->stco_entry_count && !f->read_error; i++)
                t->stco_chunk_offset[i] = read_int32(f);
        return 0;
 }
@@ -591,74 +531,6 @@ static uint16_t read_int16(struct mp4 *f)
        return read_u16_be(data);
 }
 
-static uint32_t read_mp4_descr_length(struct mp4 *f)
-{
-       uint8_t b;
-       uint8_t numBytes = 0;
-       uint32_t length = 0;
-
-       do {
-               b = read_char(f);
-               numBytes++;
-               length = (length << 7) | (b & 0x7F);
-       } while ((b & 0x80) && numBytes < 4);
-
-       return length;
-}
-static int32_t read_esds(struct mp4 *f)
-{
-       uint8_t tag;
-       uint32_t temp;
-       struct mp4_track *t;
-
-       if (f->total_tracks == 0)
-               return f->error++;
-       t = f->track[f->total_tracks - 1];
-       read_char(f);   /* version */
-       read_int24(f);  /* flags */
-       /* get and verify ES_DescrTag */
-       tag = read_char(f);
-       if (tag == 0x03) {
-               /* read length */
-               if (read_mp4_descr_length(f) < 5 + 15) {
-                       return 1;
-               }
-               /* skip 3 bytes */
-               read_int24(f);
-       } else {
-               /* skip 2 bytes */
-               read_int16(f);
-       }
-
-       /* get and verify DecoderConfigDescrTab */
-       if (read_char(f) != 0x04) {
-               return 1;
-       }
-
-       /* read length */
-       temp = read_mp4_descr_length(f);
-       if (temp < 13)
-               return 1;
-
-       t->audioType = read_char(f);
-       read_int32(f);  //0x15000414 ????
-       t->maxBitrate = read_int32(f);
-       t->avgBitrate = read_int32(f);
-
-       /* get and verify DecSpecificInfoTag */
-       if (read_char(f) != 0x05) {
-               return 1;
-       }
-
-       /* read length */
-       t->decoderConfigLen = read_mp4_descr_length(f);
-       free(t->decoderConfig);
-       t->decoderConfig = para_malloc(t->decoderConfigLen);
-       read_data(f, t->decoderConfig, t->decoderConfigLen);
-       /* will skip the remainder of the atom */
-       return 0;
-}
-
 static int32_t read_mp4a(struct mp4 *f)
 {
        int32_t i;
@@ -679,7 +551,7 @@ static int32_t read_mp4a(struct mp4 *f)
        read_int32(f);  /* reserved */
 
        t->channelCount = read_int16(f);
-       t->sampleSize = read_int16(f);
+       read_int16(f);
 
        read_int16(f);
        read_int16(f);
@@ -689,14 +561,12 @@ static int32_t read_mp4a(struct mp4 *f)
        read_int16(f);
 
        atom_read_header(f, &atom_type, &header_size);
-       if (atom_type == ATOM_ESDS)
-               read_esds(f);
        return 0;
 }
 
 static int32_t read_stsd(struct mp4 *f)
 {
-       int32_t i;
+       int32_t i, entry_count;
        uint8_t header_size = 0;
        struct mp4_track *t;
 
@@ -708,63 +578,27 @@ static int32_t read_stsd(struct mp4 *f)
        read_char(f);   /* version */
        read_int24(f);  /* flags */
 
-       t->stsd_entry_count = read_int32(f);
+       entry_count = read_int32(f);
 
        /* CVE-2017-9253 */
-       for (i = 0; i < t->stsd_entry_count && !f->stream->read_error; i++) {
+       for (i = 0; i < entry_count && !f->read_error; i++) {
                uint64_t skip = get_position(f);
                uint64_t size;
                uint8_t atom_type = 0;
                size = atom_read_header(f, &atom_type, &header_size);
                skip += size;
-
-               if (atom_type == ATOM_MP4A) {
-                       t->type = TRACK_AUDIO;
+               t->is_audio = atom_type == ATOM_MP4A;
+               if (t->is_audio)
                        read_mp4a(f);
-               } else if (atom_type == ATOM_MP4V) {
-                       t->type = TRACK_VIDEO;
-               } else if (atom_type == ATOM_MP4S) {
-                       t->type = TRACK_SYSTEM;
-               } else {
-                       t->type = TRACK_UNKNOWN;
-               }
                set_position(f, skip);
        }
 
        return 0;
 }
 
-static int32_t read_mvhd(struct mp4 *f)
-{
-       int32_t i;
-
-       read_char(f);   /* version */
-       read_int24(f);  /* flags */
-       read_int32(f); /* creation_time */
-       read_int32(f); /* modification_time */
-       f->time_scale = read_int32(f);
-       f->duration = read_int32(f);
-       read_int32(f); /* preferred_rate */
-       read_int16(f); /* preferred_volume */
-       for (i = 0; i < 10; i++)
-               read_char(f); /* reserved */
-       for (i = 0; i < 9; i++)
-               read_int32(f); /* matrix */
-       read_int32(f); /* preview_time */
-       read_int32(f); /* preview_duration */
-       read_int32(f); /* poster_time */
-       read_int32(f); /* selection_time */
-       read_int32(f); /* selection_duration */
-       read_int32(f); /* current_time */
-       read_int32(f); /* next_track_id */
-       return 0;
-}
-
 static int32_t tag_add_field(struct mp4_metadata *tags, const char *item,
                const char *value, int32_t len)
 {
-       if (!item || (item && !*item) || !value)
-               return 0;
        tags->tags = para_realloc(tags->tags,
                (tags->count + 1) * sizeof(struct mp4_tag));
        tags->tags[tags->count].item = para_strdup(item);
@@ -811,15 +645,6 @@ static const char *ID3v1GenreList[] = {
        "Anime", "JPop", "SynthPop",
 };
 
-static const char *meta_index_to_genre(uint32_t idx)
-{
-       if (idx > 0 && idx <= sizeof (ID3v1GenreList) / sizeof (ID3v1GenreList[0])) {
-               return ID3v1GenreList[idx - 1];
-       } else {
-               return 0;
-       }
-}
-
 static char *read_string(struct mp4 *f, uint32_t length)
 {
        char *str = para_malloc(length + 1);
@@ -831,232 +656,47 @@ static char *read_string(struct mp4 *f, uint32_t length)
        return str;
 }
 
-static int32_t set_metadata_name(uint8_t atom_type, char **name)
+static const char *get_metadata_name(uint8_t atom_type)
 {
-       static char *tag_names[] = {
-               "unknown", "title", "artist", "writer", "album",
-               "date", "tool", "comment", "genre", "track",
-               "disc", "compilation", "genre", "tempo", "cover",
-               "album_artist", "contentgroup", "lyrics", "description",
-               "network", "show", "episodename",
-               "sorttitle", "sortalbum", "sortartist", "sortalbumartist",
-               "sortwriter", "sortshow",
-               "season", "episode", "podcast"
-       };
-       uint8_t tag_idx = 0;
-
        switch (atom_type) {
-       case ATOM_TITLE:
-               tag_idx = 1;
-               break;
-       case ATOM_ARTIST:
-               tag_idx = 2;
-               break;
-       case ATOM_WRITER:
-               tag_idx = 3;
-               break;
-       case ATOM_ALBUM:
-               tag_idx = 4;
-               break;
-       case ATOM_DATE:
-               tag_idx = 5;
-               break;
-       case ATOM_TOOL:
-               tag_idx = 6;
-               break;
-       case ATOM_COMMENT:
-               tag_idx = 7;
-               break;
-       case ATOM_GENRE1:
-               tag_idx = 8;
-               break;
-       case ATOM_TRACK:
-               tag_idx = 9;
-               break;
-       case ATOM_DISC:
-               tag_idx = 10;
-               break;
-       case ATOM_COMPILATION:
-               tag_idx = 11;
-               break;
-       case ATOM_GENRE2:
-               tag_idx = 12;
-               break;
-       case ATOM_TEMPO:
-               tag_idx = 13;
-               break;
-       case ATOM_COVER:
-               tag_idx = 14;
-               break;
-       case ATOM_ALBUM_ARTIST:
-               tag_idx = 15;
-               break;
-       case ATOM_CONTENTGROUP:
-               tag_idx = 16;
-               break;
-       case ATOM_LYRICS:
-               tag_idx = 17;
-               break;
-       case ATOM_DESCRIPTION:
-               tag_idx = 18;
-               break;
-       case ATOM_NETWORK:
-               tag_idx = 19;
-               break;
-       case ATOM_SHOW:
-               tag_idx = 20;
-               break;
-       case ATOM_EPISODENAME:
-               tag_idx = 21;
-               break;
-       case ATOM_SORTTITLE:
-               tag_idx = 22;
-               break;
-       case ATOM_SORTALBUM:
-               tag_idx = 23;
-               break;
-       case ATOM_SORTARTIST:
-               tag_idx = 24;
-               break;
-       case ATOM_SORTALBUMARTIST:
-               tag_idx = 25;
-               break;
-       case ATOM_SORTWRITER:
-               tag_idx = 26;
-               break;
-       case ATOM_SORTSHOW:
-               tag_idx = 27;
-               break;
-       case ATOM_SEASON:
-               tag_idx = 28;
-               break;
-       case ATOM_EPISODE:
-               tag_idx = 29;
-               break;
-       case ATOM_PODCAST:
-               tag_idx = 30;
-               break;
-       default:
-               tag_idx = 0;
-               break;
-       }
-
-       *name = para_strdup(tag_names[tag_idx]);
-       return 0;
-}
-
-static uint32_t min_body_size(uint8_t atom_type)
-{
-       switch(atom_type) {
-       case ATOM_GENRE2:
-       case ATOM_TEMPO:
-               return 10;
-       case ATOM_TRACK:
-               return sizeof (char) /* version */
-                       + sizeof(uint8_t) * 3 /* flags */
-                       + sizeof(uint32_t) /* reserved */
-                       + sizeof(uint16_t) /* leading uint16_t */
-                       + sizeof(uint16_t) /* track */
-                       + sizeof(uint16_t); /* totaltracks */
-       case ATOM_DISC:
-               return sizeof (char) /* version */
-                       + sizeof(uint8_t) * 3 /* flags */
-                       + sizeof(uint32_t) /* reserved */
-                       + sizeof(uint16_t) /* disc */
-                       + sizeof(uint16_t); /* totaldiscs */
-       default: assert(false);
+       case ATOM_TITLE: return "title";
+       case ATOM_ARTIST: return "artist";
+       case ATOM_ALBUM: return "album";
+       case ATOM_DATE: return "date";
+       case ATOM_COMMENT: return "comment";
+       default: return "unknown";
        }
 }
 
-static int32_t parse_tag(struct mp4 *f, uint8_t parent, int32_t size)
+static void parse_tag(struct mp4 *f, uint8_t parent, int32_t size)
 {
-       uint8_t atom_type;
-       uint8_t header_size = 0;
        uint64_t subsize, sumsize;
-       char *name = NULL;
        char *data = NULL;
-       uint32_t done = 0;
        uint32_t len = 0;
        uint64_t destpos;
 
        for (
                sumsize = 0;
-               sumsize < size && !f->stream->read_error; /* CVE-2017-9222 */
+               sumsize < size && !f->read_error; /* CVE-2017-9222 */
                set_position(f, destpos), sumsize += subsize
        ) {
+               uint8_t atom_type;
+               uint8_t header_size = 0;
                subsize = atom_read_header(f, &atom_type, &header_size);
                destpos = get_position(f) + subsize - header_size;
-               if (done)
-                       continue;
-               if (atom_type == ATOM_NAME) {
-                       read_char(f);   /* version */
-                       read_int24(f);  /* flags */
-                       free(name);
-                       name = read_string(f, subsize - (header_size + 4));
-                       continue;
-               }
                if (atom_type != ATOM_DATA)
                        continue;
                read_char(f);   /* version */
                read_int24(f);  /* flags */
                read_int32(f);  /* reserved */
-
-               /* some need special attention */
-               if (parent == ATOM_GENRE2 || parent == ATOM_TEMPO) {
-                       uint16_t val;
-                       if (subsize - header_size < min_body_size(parent))
-                               continue;
-                       val = read_int16(f);
-                       if (parent == ATOM_TEMPO) {
-                               char temp[16];
-                               sprintf(temp, "%.5u BPM", val);
-                               tag_add_field(&(f-> tags), "tempo",
-                                       temp, -1);
-                       } else {
-                               const char *tmp = meta_index_to_genre(val);
-                               if (tmp)
-                                       tag_add_field (&(f->tags),
-                                               "genre", tmp, -1);
-                       }
-                       done = 1;
-               } else if (parent == ATOM_TRACK || parent == ATOM_DISC) {
-                       uint16_t index, total;
-                       char temp[32];
-                       if (subsize - header_size < min_body_size(parent))
-                               continue;
-                       read_int16(f);
-                       index = read_int16(f);
-                       total = read_int16(f);
-                       if (parent == ATOM_TRACK)
-                               read_int16(f);
-                       sprintf(temp, "%d", index);
-                       tag_add_field(&(f->tags), parent == ATOM_TRACK?
-                               "track" : "disc", temp, -1);
-                       if (total > 0) {
-                               sprintf(temp, "%d", total);
-                               tag_add_field(& (f-> tags),
-                                       parent == ATOM_TRACK?
-                                       "totaltracks" : "totaldiscs", temp, -1);
-                       }
-                       done = 1;
-               } else {
-                       free(data);
-                       data = read_string(f, subsize - (header_size + 8));
-                       len = subsize - (header_size + 8);
-               }
-       }
-       if (data) {
-               if (!done) {
-                       if (name == NULL)
-                               set_metadata_name(parent , &name);
-                       if (name)
-                               tag_add_field(&(f->tags), name, data, len);
-               }
-
                free(data);
+               data = read_string(f, subsize - (header_size + 8));
+               len = subsize - (header_size + 8);
        }
-       free(name);
-       return 1;
+       if (!data)
+               return;
+       tag_add_field(&f->tags, get_metadata_name(parent), data, len);
+       free(data);
 }
 
 static int32_t read_mdhd(struct mp4 *f)
@@ -1092,15 +732,25 @@ static int32_t read_mdhd(struct mp4 *f)
 
 static int32_t parse_metadata(struct mp4 *f, int32_t size)
 {
-       uint64_t subsize, sumsize = 0;
-       uint8_t atom_type;
-       uint8_t header_size = 0;
+       uint64_t sumsize = 0;
 
        while (sumsize < size) {
+               uint8_t atom_type;
+               uint64_t subsize, destpos;
+               uint8_t header_size = 0;
                subsize = atom_read_header(f, &atom_type, &header_size);
                if (subsize == 0)
                        break;
-               parse_tag(f, atom_type, (uint32_t)(subsize - header_size));
+               destpos = get_position(f) + subsize - header_size;
+               switch (atom_type) {
+               case ATOM_ARTIST:
+               case ATOM_TITLE:
+               case ATOM_ALBUM:
+               case ATOM_COMMENT:
+               case ATOM_DATE:
+                       parse_tag(f, atom_type, subsize - header_size);
+               }
+               set_position(f, destpos);
                sumsize += subsize;
        }
 
@@ -1140,9 +790,6 @@ static int32_t atom_read(struct mp4 *f, int32_t size, uint8_t atom_type)
        } else if (atom_type == ATOM_STTS) {
                /* time to sample box */
                read_stts(f);
-       } else if (atom_type == ATOM_CTTS) {
-               /* composition offset box */
-               read_ctts(f);
        } else if (atom_type == ATOM_STSC) {
                /* sample to chunk box */
                read_stsc(f);
@@ -1152,9 +799,6 @@ static int32_t atom_read(struct mp4 *f, int32_t size, uint8_t atom_type)
        } else if (atom_type == ATOM_STSD) {
                /* sample description box */
                read_stsd(f);
-       } else if (atom_type == ATOM_MVHD) {
-               /* movie header box */
-               read_mvhd(f);
        } else if (atom_type == ATOM_MDHD) {
                /* track header */
                read_mdhd(f);
@@ -1209,7 +853,7 @@ static int32_t parse_atoms(struct mp4 *f, int meta_only)
        uint8_t header_size = 0;
 
        f->file_size = 0;
-       f->stream->read_error = 0;
+       f->read_error = 0;
 
        while ((size =
                atom_read_header(f, &atom_type, &header_size)) != 0) {
@@ -1235,41 +879,17 @@ static int32_t parse_atoms(struct mp4 *f, int meta_only)
        return 0;
 }
 
-void mp4_get_decoder_config(const struct mp4 *f, int track,
-               unsigned char **ppBuf, unsigned int *pBufSize)
+struct mp4 *mp4_open_read(const struct mp4_callback *cb)
 {
-       if (track >= f->total_tracks) {
-               *ppBuf = NULL;
-               *pBufSize = 0;
-               return;
-       }
+       struct mp4 *f = para_calloc(sizeof(struct mp4));
 
-       if (f->track[track]->decoderConfig == NULL
-               || f->track[track]->decoderConfigLen == 0) {
-               *ppBuf = NULL;
-               *pBufSize = 0;
-       } else {
-               *ppBuf = para_malloc(f->track[track]->decoderConfigLen);
-               memcpy(*ppBuf, f->track[track]->decoderConfig,
-                       f->track[track]->decoderConfigLen);
-               *pBufSize = f->track[track]->decoderConfigLen;
+       f->cb = cb;
+       parse_atoms(f, 0);
+       if (f->error) {
+               free(f);
+               f = NULL;
        }
-}
-
-struct mp4 *mp4_open_read(struct mp4_callback *f)
-{
-       struct mp4 *ff = para_calloc(sizeof(struct mp4));
-
-       ff->stream = f;
-
-       parse_atoms(ff, 0);
-
-       if (ff->error) {
-               free(ff);
-               ff = NULL;
-       }
-
-       return ff;
+       return f;
 }
 
 static int32_t tag_delete(struct mp4_metadata *tags)
@@ -1287,28 +907,25 @@ static int32_t tag_delete(struct mp4_metadata *tags)
        return 0;
 }
 
-void mp4_close(struct mp4 *ff)
+void mp4_close(struct mp4 *f)
 {
        int32_t i;
 
-       for (i = 0; i < ff->total_tracks; i++) {
-               if (ff->track[i]) {
-                       free(ff->track[i]->stsz_table);
-                       free(ff->track[i]->stts_sample_count);
-                       free(ff->track[i]->stts_sample_delta);
-                       free(ff->track[i]->stsc_first_chunk);
-                       free(ff->track[i]->stsc_samples_per_chunk);
-                       free(ff->track[i]->stsc_sample_desc_index);
-                       free(ff->track[i]->stco_chunk_offset);
-                       free(ff->track[i]->decoderConfig);
-                       free(ff->track[i]->ctts_sample_count);
-                       free(ff->track[i]->ctts_sample_offset);
-                       free(ff->track[i]);
+       for (i = 0; i < f->total_tracks; i++) {
+               if (f->track[i]) {
+                       free(f->track[i]->stsz_table);
+                       free(f->track[i]->stts_sample_count);
+                       free(f->track[i]->stts_sample_delta);
+                       free(f->track[i]->stsc_first_chunk);
+                       free(f->track[i]->stsc_samples_per_chunk);
+                       free(f->track[i]->stsc_sample_desc_index);
+                       free(f->track[i]->stco_chunk_offset);
+                       free(f->track[i]);
                }
        }
 
-       tag_delete(&(ff->tags));
-       free(ff);
+       tag_delete(&(f->tags));
+       free(f);
 }
 
 static int32_t chunk_of_sample(const struct mp4 *f, int32_t track,
@@ -1425,6 +1042,20 @@ uint64_t mp4_get_duration(const struct mp4 *f, int32_t track)
        return t->duration * 1000 / t->timeScale;
 }
 
+/**
+ * Check whether the given track number corresponds to an audio track.
+ *
+ * \param f See \ref mp4_get_duration().
+ * \param track See \ref mp4_get_duration().
+ *
+ * Besides audio tracks, an mp4 file may contain video and system tracks. For
+ * those the function returns false.
+ */
+bool mp4_is_audio_track(const struct mp4 *f, int32_t track)
+{
+       return f->track[track]->is_audio;
+}
+
 void mp4_set_sample_position(struct mp4 *f, int32_t track, int32_t sample)
 {
        int32_t offset = sample_to_offset(f, track, sample);
@@ -1461,20 +1092,17 @@ int32_t mp4_num_samples(const struct mp4 *f, int32_t track)
        return total;
 }
 
-struct mp4 *mp4_open_meta(struct mp4_callback *f)
+struct mp4 *mp4_open_meta(const struct mp4_callback *cb)
 {
-       struct mp4 *ff = para_calloc(sizeof(struct mp4));
-
-       ff->stream = f;
-
-       parse_atoms(ff, 1);
+       struct mp4 *f = para_calloc(sizeof(struct mp4));
 
-       if (ff->error) {
-               free(ff);
-               ff = NULL;
+       f->cb = cb;
+       parse_atoms(f, 1);
+       if (f->error) {
+               free(f);
+               f = NULL;
        }
-
-       return ff;
+       return f;
 }
 
 int32_t mp4_meta_get_num_items(const struct mp4 *f)
@@ -2067,7 +1695,7 @@ static int32_t write_data(struct mp4 *f, void *data, uint32_t size)
 {
        int32_t result = 1;
 
-       result = f->stream->write(f->stream->user_data, data, size);
+       result = f->cb->write(f->cb->user_data, data, size);
 
        f->current_position += size;
 
@@ -2081,49 +1709,41 @@ static int32_t write_int32(struct mp4 *f, uint32_t data)
        return write_data(f, temp, sizeof(temp));
 }
 
-static int32_t truncate_stream(struct mp4 *f)
-{
-       return f->stream->truncate(f->stream->user_data);
-}
-
-int32_t mp4_meta_update(struct mp4_callback *f, const struct mp4_metadata *data)
+int32_t mp4_meta_update(const struct mp4_callback *cb,
+               const struct mp4_metadata *data)
 {
        void *new_moov_data;
        uint32_t new_moov_size;
 
-       struct mp4 *ff = para_calloc(sizeof(struct mp4));
-       ff->stream = f;
-       set_position(ff, 0);
-
-       parse_atoms(ff, 1);
-
-       if (!modify_moov(ff, data, &new_moov_data, &new_moov_size)) {
-               mp4_close(ff);
+       struct mp4 *f = para_calloc(sizeof(struct mp4));
+       f->cb = cb;
+       set_position(f, 0);
+       parse_atoms(f, 1);
+       if (!modify_moov(f, data, &new_moov_data, &new_moov_size)) {
+               mp4_close(f);
                return 0;
        }
 
        /* copy moov atom to end of the file */
-       if (ff->last_atom != ATOM_MOOV) {
+       if (f->last_atom != ATOM_MOOV) {
                char *free_data = "free";
 
                /* rename old moov to free */
-               set_position(ff, ff->moov_offset + 4);
-               write_data(ff, free_data, 4);
+               set_position(ff->moov_offset + 4);
+               write_data(f, free_data, 4);
 
-               set_position(ff, ff->file_size);
-               write_int32(ff, new_moov_size + 8);
-               write_data(ff, "moov", 4);
-               write_data(ff, new_moov_data, new_moov_size);
+               set_position(ff->file_size);
+               write_int32(f, new_moov_size + 8);
+               write_data(f, "moov", 4);
+               write_data(f, new_moov_data, new_moov_size);
        } else {
-               set_position(ff, ff->moov_offset);
-               write_int32(ff, new_moov_size + 8);
-               write_data(ff, "moov", 4);
-               write_data(ff, new_moov_data, new_moov_size);
+               set_position(ff->moov_offset);
+               write_int32(f, new_moov_size + 8);
+               write_data(f, "moov", 4);
+               write_data(f, new_moov_data, new_moov_size);
        }
-
-       truncate_stream(ff);
-
-       mp4_close(ff);
+       cb->truncate(cb->user_data);
+       mp4_close(f);
        return 1;
 }