X-Git-Url: http://git.tuebingen.mpg.de/?p=paraslash.git;a=blobdiff_plain;f=mp4.c;h=4b8607b185a9a87c845940d86412843654a13c7d;hp=23bd7934ac87db131331df06e78703d6ec5d0438;hb=HEAD;hpb=153f946f890f7de84d7e2c6e9db6daca2dd3f6ca diff --git a/mp4.c b/mp4.c index 23bd7934..fe9d4b37 100644 --- a/mp4.c +++ b/mp4.c @@ -5,1975 +5,1052 @@ * See file COPYING. */ +/** \file mp4.c Paraslash's internal mp4 parser. */ + +/* + * This is a stripped down version of the former mp4ff library which used to be + * part of the faad decoder project but was removed from the faad code base in + * 2017. The original code has been cleaned up substantially and the public API + * has been documented. See the git commit log for details. + */ + #include #include "para.h" +#include "error.h" #include "portable_io.h" #include "string.h" #include "mp4.h" +/** + * The three states of the mp4 parser. The parser only loads the audio specific + * values and tables when it is in the second state. + */ +enum audio_track_state { + /** We haven't encountered an mp4a atom so far. */ + ATS_INITIAL, + /** We have seen an mp4a atom but no subsequent trak atom yet. */ + ATS_SEEN_MP4A, + /** A trak atom was seen *after* the mp4a atom. */ + ATS_TRACK_CHANGE, +}; + struct mp4_track { - bool is_audio; - int32_t channelCount; - uint16_t sampleRate; + /* determines which atoms we still need to parse. */ + enum audio_track_state state; + + /* mp4a */ + uint16_t channel_count; + uint16_t sample_rate; /* stsz */ - int32_t stsz_sample_size; - int32_t stsz_sample_count; - int32_t *stsz_table; + uint32_t stsz_sample_size; + uint32_t stsz_sample_count; + uint32_t *stsz_table; /* stts */ - int32_t stts_entry_count; - int32_t *stts_sample_count; - int32_t *stts_sample_delta; + uint32_t stts_entry_count; + uint32_t *stts_sample_count; /* stsc */ - int32_t stsc_entry_count; - int32_t *stsc_first_chunk; - int32_t *stsc_samples_per_chunk; - int32_t *stsc_sample_desc_index; + uint32_t stsc_entry_count; + uint32_t *stsc_first_chunk; + uint32_t *stsc_samples_per_chunk; /* stsc */ - int32_t stco_entry_count; - int32_t *stco_chunk_offset; + uint32_t stco_entry_count; + uint32_t *stco_chunk_offset; - uint32_t timeScale; + /* mdhd */ + uint32_t time_scale; uint64_t duration; }; -#define MAX_TRACKS 1024 - struct mp4 { - struct mp4_callback *cb; - int64_t current_position; + const struct mp4_callback *cb; uint64_t moov_offset; uint64_t moov_size; - uint8_t last_atom; - uint64_t file_size; - - uint32_t read_error; - uint32_t error; - - /* incremental track index while reading the file */ - int32_t total_tracks; - - /* track data */ - struct mp4_track *track[MAX_TRACKS]; + uint64_t meta_offset; + uint32_t meta_size; + uint64_t ilst_offset; + uint32_t ilst_size; + uint64_t udta_offset; + uint32_t udta_size; - /* metadata */ - struct mp4_metadata tags; + uint8_t last_atom; + struct mp4_track track; + struct mp4_metadata meta; }; -int32_t mp4_total_tracks(const struct mp4 *f) -{ - return f->total_tracks; -} - -static int32_t read_data(struct mp4 *f, void *data, uint32_t size) +/* + * Returns -E_MP4_READ, 0, or 1 on errors/EOF/success. Partial reads followed + * by EOF or read errors are treated as errors. + */ +static int read_data(struct mp4 *f, void *data, size_t size) { - int32_t result = 1; - - result = f->cb->read(f->cb->user_data, data, size); - - if (result < size) - f->read_error++; - - f->current_position += size; - - return result; + while (size > 0) { + ssize_t ret = f->cb->read(f->cb->user_data, data, size); + if (ret < 0 && errno == EINTR) + continue; + /* regard EAGAIN as an error as reads should be blocking. */ + if (ret <= 0) + return ret < 0? -E_MP4_READ : 0; + size -= ret; + } + return 1; } -static uint64_t read_int64(struct mp4 *f) +static int read_int64(struct mp4 *f, uint64_t *result) { uint8_t data[8]; + int ret = read_data(f, data, 8); - read_data(f, data, 8); - return read_u64_be(data); + if (ret > 0) + *result = read_u64_be(data); + return ret; } -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) +static int read_int32(struct mp4 *f, uint32_t *result) { - return a1 == a2 && b1 == b2 && c1 == c2 && d1 == d2; -} - -enum atoms { - /* atoms with subatoms */ - ATOM_MOOV = 1, - ATOM_TRAK = 2, - ATOM_EDTS = 3, - ATOM_MDIA = 4, - ATOM_MINF = 5, - ATOM_STBL = 6, - ATOM_UDTA = 7, - ATOM_ILST = 8, /* iTunes Metadata list */ - ATOM_TITLE = 9, - ATOM_ARTIST = 10, - ATOM_WRITER = 11, - ATOM_ALBUM = 12, - ATOM_DATE = 13, - ATOM_TOOL = 14, - ATOM_COMMENT = 15, - ATOM_GENRE1 = 16, - ATOM_TRACK = 17, - ATOM_DISC = 18, - ATOM_COMPILATION = 19, - ATOM_GENRE2 = 20, - ATOM_TEMPO = 21, - ATOM_COVER = 22, - ATOM_DRMS = 23, - ATOM_SINF = 24, - ATOM_SCHI = 25, - - SUBATOMIC = 128, - - /* atoms without subatoms */ - ATOM_FTYP = 129, - ATOM_MDAT = 130, - ATOM_MVHD = 131, - ATOM_TKHD = 132, - ATOM_TREF = 133, - ATOM_MDHD = 134, - ATOM_VMHD = 135, - ATOM_SMHD = 136, - ATOM_HMHD = 137, - ATOM_STSD = 138, - ATOM_STTS = 139, - ATOM_STSZ = 140, - ATOM_STZ2 = 141, - ATOM_STCO = 142, - ATOM_STSC = 143, - ATOM_MP4A = 144, - ATOM_MP4V = 145, - ATOM_MP4S = 146, - ATOM_ESDS = 147, - ATOM_META = 148, /* iTunes Metadata box */ - ATOM_NAME = 149, /* iTunes Metadata name box */ - ATOM_DATA = 150, /* iTunes Metadata data box */ - ATOM_CTTS = 151, - ATOM_FRMA = 152, - ATOM_IVIV = 153, - ATOM_PRIV = 154, - ATOM_USER = 155, - ATOM_KEY = 156, - ATOM_ALBUM_ARTIST = 157, - ATOM_CONTENTGROUP = 158, - ATOM_LYRICS = 159, - ATOM_DESCRIPTION = 160, - ATOM_NETWORK = 161, - ATOM_SHOW = 162, - ATOM_EPISODENAME = 163, - ATOM_SORTTITLE = 164, - ATOM_SORTALBUM = 165, - ATOM_SORTARTIST = 166, - ATOM_SORTALBUMARTIST = 167, - ATOM_SORTWRITER = 168, - ATOM_SORTSHOW = 169, - ATOM_SEASON = 170, - ATOM_EPISODE = 171, - ATOM_PODCAST = 172, - - ATOM_UNKNOWN = 255 -}; - -#define ATOM_FREE ATOM_UNKNOWN -#define ATOM_SKIP ATOM_UNKNOWN + uint8_t data[4]; + int ret = read_data(f, data, 4); -#define COPYRIGHT_SYMBOL ((int8_t)0xA9) + if (ret > 0) + *result = read_u32_be(data); + return ret; +} -static uint8_t atom_name_to_type(int8_t a, int8_t b, int8_t c, int8_t d) +static int read_int16(struct mp4 *f, uint16_t *result) { - if (a == 'm') { - if (atom_compare(a, b, c, d, 'm', 'o', 'o', 'v')) - return ATOM_MOOV; - else if (atom_compare(a, b, c, d, 'm', 'i', 'n', 'f')) - return ATOM_MINF; - else if (atom_compare(a, b, c, d, 'm', 'd', 'i', 'a')) - return ATOM_MDIA; - else if (atom_compare(a, b, c, d, 'm', 'd', 'a', 't')) - return ATOM_MDAT; - else if (atom_compare(a, b, c, d, 'm', 'd', 'h', 'd')) - return ATOM_MDHD; - else if (atom_compare(a, b, c, d, 'm', 'v', 'h', 'd')) - return ATOM_MVHD; - else if (atom_compare(a, b, c, d, 'm', 'p', '4', 'a')) - return ATOM_MP4A; - else if (atom_compare(a, b, c, d, 'm', 'p', '4', 'v')) - return ATOM_MP4V; - else if (atom_compare(a, b, c, d, 'm', 'p', '4', 's')) - return ATOM_MP4S; - else if (atom_compare(a, b, c, d, 'm', 'e', 't', 'a')) - return ATOM_META; - } else if (a == 't') { - if (atom_compare(a, b, c, d, 't', 'r', 'a', 'k')) - return ATOM_TRAK; - else if (atom_compare(a, b, c, d, 't', 'k', 'h', 'd')) - return ATOM_TKHD; - else if (atom_compare(a, b, c, d, 't', 'r', 'e', 'f')) - return ATOM_TREF; - else if (atom_compare(a, b, c, d, 't', 'r', 'k', 'n')) - return ATOM_TRACK; - else if (atom_compare(a, b, c, d, 't', 'm', 'p', 'o')) - return ATOM_TEMPO; - else if (atom_compare(a, b, c, d, 't', 'v', 'n', 'n')) - return ATOM_NETWORK; - else if (atom_compare(a, b, c, d, 't', 'v', 's', 'h')) - return ATOM_SHOW; - else if (atom_compare(a, b, c, d, 't', 'v', 'e', 'n')) - return ATOM_EPISODENAME; - else if (atom_compare(a, b, c, d, 't', 'v', 's', 'n')) - return ATOM_SEASON; - else if (atom_compare(a, b, c, d, 't', 'v', 'e', 's')) - return ATOM_EPISODE; - } else if (a == 's') { - if (atom_compare(a, b, c, d, 's', 't', 'b', 'l')) - return ATOM_STBL; - else if (atom_compare(a, b, c, d, 's', 'm', 'h', 'd')) - return ATOM_SMHD; - else if (atom_compare(a, b, c, d, 's', 't', 's', 'd')) - return ATOM_STSD; - else if (atom_compare(a, b, c, d, 's', 't', 't', 's')) - return ATOM_STTS; - else if (atom_compare(a, b, c, d, 's', 't', 'c', 'o')) - return ATOM_STCO; - else if (atom_compare(a, b, c, d, 's', 't', 's', 'c')) - return ATOM_STSC; - else if (atom_compare(a, b, c, d, 's', 't', 's', 'z')) - return ATOM_STSZ; - else if (atom_compare(a, b, c, d, 's', 't', 'z', '2')) - return ATOM_STZ2; - else if (atom_compare(a, b, c, d, 's', 'k', 'i', 'p')) - return ATOM_SKIP; - else if (atom_compare(a, b, c, d, 's', 'i', 'n', 'f')) - return ATOM_SINF; - else if (atom_compare(a, b, c, d, 's', 'c', 'h', 'i')) - return ATOM_SCHI; - else if (atom_compare(a, b, c, d, 's', 'o', 'n', 'm')) - return ATOM_SORTTITLE; - else if (atom_compare(a, b, c, d, 's', 'o', 'a', 'l')) - return ATOM_SORTALBUM; - else if (atom_compare(a, b, c, d, 's', 'o', 'a', 'r')) - return ATOM_SORTARTIST; - else if (atom_compare(a, b, c, d, 's', 'o', 'a', 'a')) - return ATOM_SORTALBUMARTIST; - else if (atom_compare(a, b, c, d, 's', 'o', 'c', 'o')) - return ATOM_SORTWRITER; - else if (atom_compare(a, b, c, d, 's', 'o', 's', 'n')) - return ATOM_SORTSHOW; - } else if (a == COPYRIGHT_SYMBOL) { - if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'n', 'a', 'm')) - return ATOM_TITLE; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'A', 'R', 'T')) - return ATOM_ARTIST; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'w', 'r', 't')) - return ATOM_WRITER; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'a', 'l', 'b')) - return ATOM_ALBUM; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'd', 'a', 'y')) - return ATOM_DATE; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 't', 'o', 'o')) - return ATOM_TOOL; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'c', 'm', 't')) - return ATOM_COMMENT; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'g', 'e', 'n')) - return ATOM_GENRE1; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'g', 'r', 'p')) - return ATOM_CONTENTGROUP; - else if (atom_compare(a, b, c, d, COPYRIGHT_SYMBOL, 'l', 'y', 'r')) - return ATOM_LYRICS; - } + uint8_t data[2]; + int ret = read_data(f, data, 2); - if (atom_compare(a, b, c, d, 'e', 'd', 't', 's')) - return ATOM_EDTS; - else if (atom_compare(a, b, c, d, 'e', 's', 'd', 's')) - return ATOM_ESDS; - else if (atom_compare(a, b, c, d, 'f', 't', 'y', 'p')) - return ATOM_FTYP; - else if (atom_compare(a, b, c, d, 'f', 'r', 'e', 'e')) - return ATOM_FREE; - else if (atom_compare(a, b, c, d, 'h', 'm', 'h', 'd')) - return ATOM_HMHD; - else if (atom_compare(a, b, c, d, 'v', 'm', 'h', 'd')) - return ATOM_VMHD; - else if (atom_compare(a, b, c, d, 'u', 'd', 't', 'a')) - return ATOM_UDTA; - else if (atom_compare(a, b, c, d, 'i', 'l', 's', 't')) - return ATOM_ILST; - else if (atom_compare(a, b, c, d, 'n', 'a', 'm', 'e')) - return ATOM_NAME; - else if (atom_compare(a, b, c, d, 'd', 'a', 't', 'a')) - return ATOM_DATA; - else if (atom_compare(a, b, c, d, 'd', 'i', 's', 'k')) - return ATOM_DISC; - else if (atom_compare(a, b, c, d, 'g', 'n', 'r', 'e')) - return ATOM_GENRE2; - else if (atom_compare(a, b, c, d, 'c', 'o', 'v', 'r')) - return ATOM_COVER; - else if (atom_compare(a, b, c, d, 'c', 'p', 'i', 'l')) - return ATOM_COMPILATION; - else if (atom_compare(a, b, c, d, 'c', 't', 't', 's')) - return ATOM_CTTS; - else if (atom_compare(a, b, c, d, 'd', 'r', 'm', 's')) - return ATOM_DRMS; - else if (atom_compare(a, b, c, d, 'f', 'r', 'm', 'a')) - return ATOM_FRMA; - else if (atom_compare(a, b, c, d, 'p', 'r', 'i', 'v')) - return ATOM_PRIV; - else if (atom_compare(a, b, c, d, 'i', 'v', 'i', 'v')) - return ATOM_IVIV; - else if (atom_compare(a, b, c, d, 'u', 's', 'e', 'r')) - return ATOM_USER; - else if (atom_compare(a, b, c, d, 'k', 'e', 'y', ' ')) - return ATOM_KEY; - else if (atom_compare(a, b, c, d, 'a', 'A', 'R', 'T')) - return ATOM_ALBUM_ARTIST; - else if (atom_compare(a, b, c, d, 'd', 'e', 's', 'c')) - return ATOM_DESCRIPTION; - else if (atom_compare(a, b, c, d, 'p', 'c', 's', 't')) - return ATOM_PODCAST; - else - return ATOM_UNKNOWN; + if (ret > 0) + *result = read_u16_be(data); + return ret; } -/* read atom header, return atom size, atom size is with header included */ -static uint64_t atom_read_header(struct mp4 *f, uint8_t * atom_type, - uint8_t * header_size) -{ - uint64_t size; - int32_t ret; - int8_t atom_header[8]; +/** \cond atom_items */ +/* A macro defining the atoms we care about. It gets expanded twice. */ +#define ATOM_ITEMS \ + ATOM_ITEM(MOOV, 'm', 'o', 'o', 'v') /* movie (top-level container) */ \ + ATOM_ITEM(TRAK, 't', 'r', 'a', 'k') /* container for a single track */ \ + ATOM_ITEM(MDIA, 'm', 'd', 'i', 'a') /* media information */ \ + ATOM_ITEM(MINF, 'm', 'i', 'n', 'f') /* extends mdia */ \ + ATOM_ITEM(STBL, 's', 't', 'b', 'l') /* sample table container */ \ + ATOM_ITEM(UDTA, 'u', 'd', 't', 'a') /* user data */ \ + ATOM_ITEM(ILST, 'i', 'l', 's', 't') /* iTunes Metadata list */ \ + ATOM_ITEM(ARTIST, 0xa9, 'A', 'R', 'T') /* artist */ \ + ATOM_ITEM(TITLE, 0xa9, 'n', 'a', 'm') /* title */ \ + ATOM_ITEM(ALBUM, 0xa9, 'a', 'l', 'b') /* album */ \ + ATOM_ITEM(DATE, 0xa9, 'd', 'a', 'y') /* date */ \ + ATOM_ITEM(COMMENT, 0xa9, 'c', 'm', 't') /* comment */ \ + ATOM_ITEM(MDHD, 'm', 'd', 'h', 'd') /* track header */ \ + ATOM_ITEM(STSD, 's', 't', 's', 'd') /* sample description box */ \ + ATOM_ITEM(STTS, 's', 't', 't', 's') /* time to sample box */ \ + ATOM_ITEM(STSZ, 's', 't', 's', 'z') /* sample size box */ \ + ATOM_ITEM(STCO, 's', 't', 'c', 'o') /* chunk offset box */ \ + ATOM_ITEM(STSC, 's', 't', 's', 'c') /* sample to chunk box */ \ + ATOM_ITEM(MP4A, 'm', 'p', '4', 'a') /* mp4 audio */ \ + ATOM_ITEM(META, 'm', 'e', 't', 'a') /* iTunes Metadata box */ \ + ATOM_ITEM(DATA, 'd', 'a', 't', 'a') /* iTunes Metadata data box */ \ + +/** \endcond atom_items */ + +/** For the C enumeration we concatenate ATOM_ with the first argument. */ +#define ATOM_ITEM(_name, a, b, c, d) ATOM_ ## _name, +/** The enumeration of interesting atoms. */ +enum atom {ATOM_ITEMS}; +#undef ATOM_ITEM + +/** A cpp version of read_u32_be(). */ +#define ATOM_VALUE(a, b, c, d) ((a << 24) + (b << 16) + (c << 8) + d) + +static uint8_t atom_name_to_type(uint8_t *p) +{ + /** Expands to an instance of the following unnamed structure. */ + #define ATOM_ITEM(_name, a, b, c, d) \ + {.name = # _name, .val = ATOM_VALUE(a, b, c, d)}, + static const struct { + const char *name; + uint32_t val; + } atom_table[] = {ATOM_ITEMS}; + #undef ATOM_ITEM + uint32_t val = read_u32_be(p); + + for (uint8_t n = 0; n < ARRAY_SIZE(atom_table); n++) + if (val == atom_table[n].val) + return n; + return 255; +} + +/* read atom header, atom size is returned with header included. */ +static int atom_read_header(struct mp4 *f, uint8_t *atom_type, + uint8_t *header_size, uint64_t *atom_size) +{ + uint32_t size; + int ret; + uint8_t atom_header[8]; ret = read_data(f, atom_header, 8); - if (ret != 8) - return 0; - + if (ret <= 0) + return ret; size = read_u32_be(atom_header); - *header_size = 8; - - /* check for 64 bit atom size */ - if (size == 1) { - *header_size = 16; - size = read_int64(f); - } - *atom_type = atom_name_to_type(atom_header[4], atom_header[5], - atom_header[6], atom_header[7]); - return size; -} - -static int64_t get_position(const struct mp4 *f) -{ - return f->current_position; -} - -static int need_parse_when_meta_only(uint8_t atom_type) -{ - switch (atom_type) { - case ATOM_EDTS: - case ATOM_DRMS: - case ATOM_SINF: - case ATOM_SCHI: - case ATOM_STTS: - case ATOM_STSZ: - case ATOM_STZ2: - case ATOM_STCO: - case ATOM_STSC: - case ATOM_FRMA: - case ATOM_IVIV: - case ATOM_PRIV: - return 0; - default: - return 1; + if (size == 1) { /* 64 bit atom size */ + if (header_size) + *header_size = 16; + ret = read_int64(f, atom_size); + if (ret <= 0) + return ret; + } else { + if (header_size) + *header_size = 8; + *atom_size = size; } + *atom_type = atom_name_to_type(atom_header + 4); + return 1; } -static int32_t set_position(struct mp4 *f, int64_t position) -{ - f->cb->seek(f->cb->user_data, position); - f->current_position = position; - - return 0; -} - -static void track_add(struct mp4 *f) +static off_t get_position(const struct mp4 *f) { - f->total_tracks++; - - if (f->total_tracks > MAX_TRACKS) { - f->total_tracks = 0; - f->error++; - return; - } - f->track[f->total_tracks - 1] = para_calloc(sizeof(struct mp4_track)); + return f->cb->seek(f->cb->user_data, 0, SEEK_CUR); } -static uint8_t read_char(struct mp4 *f) +static void set_position(struct mp4 *f, off_t position) { - uint8_t output; - read_data(f, &output, 1); - return output; + f->cb->seek(f->cb->user_data, position, SEEK_SET); } -static uint32_t read_int24(struct mp4 *f) +static void skip_bytes(struct mp4 *f, off_t num_skip) { - int8_t data[4]; - - read_data(f, data, 3); - return read_u24_be(data); + f->cb->seek(f->cb->user_data, num_skip, SEEK_CUR); } -static uint32_t read_int32(struct mp4 *f) +static int read_stsz(struct mp4 *f) { - int8_t data[4]; - - read_data(f, data, 4); - return read_u32_be(data); -} + int ret; + struct mp4_track *t = &f->track; -static int32_t read_stsz(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]; - read_char(f); /* version */ - read_int24(f); /* flags */ - t->stsz_sample_size = read_int32(f); - t->stsz_sample_count = read_int32(f); + if (t->state != ATS_SEEN_MP4A || t->stsz_table) + return 1; + skip_bytes(f, 4); /* version (1), flags (3) */ + ret = read_int32(f, &t->stsz_sample_size); + if (ret <= 0) + return ret; + ret = read_int32(f, &t->stsz_sample_count); + if (ret <= 0) + return ret; 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->read_error; i++) - t->stsz_table[i] = read_int32(f); - return 0; -} - -static int32_t read_stts(struct mp4 *f) -{ - int32_t i; - struct mp4_track *t; - - /* CVE-2017-9223 */ - if (f->total_tracks == 0) - return f->error++; - t = f->track[f->total_tracks - 1]; - if (t->stts_entry_count) - return 0; - read_char(f); /* version */ - read_int24(f); /* flags */ - t->stts_entry_count = read_int32(f); - - t->stts_sample_count = para_malloc(t->stts_entry_count - * sizeof(int32_t)); - 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->read_error; i++) { - t->stts_sample_count[i] = read_int32(f); - t->stts_sample_delta[i] = read_int32(f); + return 1; + t->stsz_table = arr_alloc(t->stsz_sample_count, sizeof(int32_t)); + for (uint32_t n = 0; n < t->stsz_sample_count; n++) { + ret = read_int32(f, &t->stsz_table[n]); + if (ret <= 0) + return ret; } return 1; } -static int32_t read_stsc(struct mp4 *f) +static int read_stts(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]; - - read_char(f); /* version */ - read_int24(f); /* flags */ - t->stsc_entry_count = read_int32(f); - t->stsc_first_chunk = para_malloc(t->stsc_entry_count * sizeof(int32_t)); - t->stsc_samples_per_chunk = para_malloc(t->stsc_entry_count - * sizeof (int32_t)); - t->stsc_sample_desc_index = para_malloc(t->stsc_entry_count * - sizeof (int32_t)); + int ret; + struct mp4_track *t = &f->track; - /* CVE-2017-9255 */ - 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); + if (t->state != ATS_SEEN_MP4A || t->stts_sample_count) + return 1; + skip_bytes(f, 4); /* version (1), flags (3) */ + ret = read_int32(f, &t->stts_entry_count); + if (ret <= 0) + return ret; + t->stts_sample_count = arr_alloc(t->stts_entry_count, sizeof(int32_t)); + for (uint32_t n = 0; n < t->stts_entry_count; n++) { + ret = read_int32(f, &t->stts_sample_count[n]); + if (ret <= 0) + return ret; + skip_bytes(f, 4); /* sample delta */ } - return 0; -} - -static int32_t read_stco(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]; - - read_char(f); /* version */ - read_int24(f); /* flags */ - t->stco_entry_count = read_int32(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->read_error; i++) - t->stco_chunk_offset[i] = read_int32(f); - return 0; + return 1; } -static uint16_t read_int16(struct mp4 *f) +static int read_stsc(struct mp4 *f) { - int8_t data[2]; + int ret; + struct mp4_track *t = &f->track; - read_data(f, data, 2); - return read_u16_be(data); + if (t->state != ATS_SEEN_MP4A) + return 1; + if (t->stsc_first_chunk || t->stsc_samples_per_chunk) + return 1; + skip_bytes(f, 4); /* version (1), flags (3) */ + ret = read_int32(f, &t->stsc_entry_count); + if (ret <= 0) + return ret; + t->stsc_first_chunk = arr_alloc(t->stsc_entry_count, sizeof(int32_t)); + t->stsc_samples_per_chunk = arr_alloc(t->stsc_entry_count, + sizeof (int32_t)); + for (uint32_t n = 0; n < t->stsc_entry_count; n++) { + ret = read_int32(f, &t->stsc_first_chunk[n]); + if (ret <= 0) + return ret; + ret = read_int32(f, &t->stsc_samples_per_chunk[n]); + if (ret <= 0) + return ret; + skip_bytes(f, 4); /* sample desc index */ + } + return 1; } -static int32_t read_mp4a(struct mp4 *f) +static int read_stco(struct mp4 *f) { - int32_t i; - uint8_t atom_type = 0; - uint8_t header_size = 0; - struct mp4_track *t; - - if (f->total_tracks == 0) - return f->error++; - t = f->track[f->total_tracks - 1]; + int ret; + struct mp4_track *t = &f->track; - for (i = 0; i < 6; i++) { - read_char(f); /* reserved */ + if (t->state != ATS_SEEN_MP4A || t->stco_chunk_offset) + return 1; + skip_bytes(f, 4); /* version (1), flags (3) */ + ret = read_int32(f, &t->stco_entry_count); + if (ret <= 0) + return ret; + t->stco_chunk_offset = arr_alloc(t->stco_entry_count, sizeof(int32_t)); + for (uint32_t n = 0; n < t->stco_entry_count; n++) { + ret = read_int32(f, &t->stco_chunk_offset[n]); + if (ret <= 0) + return ret; } - /* data_reference_index */ read_int16(f); - - read_int32(f); /* reserved */ - read_int32(f); /* reserved */ - - t->channelCount = read_int16(f); - read_int16(f); - - read_int16(f); - read_int16(f); - - t->sampleRate = read_int16(f); - - read_int16(f); - - atom_read_header(f, &atom_type, &header_size); - return 0; + return 1; } -static int32_t read_stsd(struct mp4 *f) +static int read_stsd(struct mp4 *f) { - int32_t i, entry_count; - uint8_t header_size = 0; - struct mp4_track *t; - - /* CVE-2017-9218 */ - if (f->total_tracks == 0) - return f->error++; - t = f->track[f->total_tracks - 1]; + int ret; + uint32_t entry_count; - read_char(f); /* version */ - read_int24(f); /* flags */ - - entry_count = read_int32(f); - - /* CVE-2017-9253 */ - for (i = 0; i < entry_count && !f->read_error; i++) { + if (f->track.state != ATS_INITIAL) + return 1; + skip_bytes(f, 4); /* version (1), flags (3) */ + ret = read_int32(f, &entry_count); + if (ret <= 0) + return ret; + for (uint32_t n = 0; n < entry_count; n++) { uint64_t skip = get_position(f); uint64_t size; uint8_t atom_type = 0; - size = atom_read_header(f, &atom_type, &header_size); + ret = atom_read_header(f, &atom_type, NULL, &size); + if (ret <= 0) + return ret; skip += size; - t->is_audio = atom_type == ATOM_MP4A; - if (t->is_audio) - read_mp4a(f); + if (atom_type == ATOM_MP4A) { + f->track.state = ATS_SEEN_MP4A; + /* reserved (6), data reference index (2), reserved (8) */ + skip_bytes(f, 16); + ret = read_int16(f, &f->track.channel_count); + if (ret <= 0) + return ret; + skip_bytes(f, 6); + ret = read_int16(f, &f->track.sample_rate); + if (ret <= 0) + return ret; + } set_position(f, skip); } - - 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); - tags->tags[tags->count].len = len; - if (len >= 0) { - tags->tags[tags->count].value = para_malloc(len + 1); - memcpy(tags->tags[tags->count].value, value, len); - tags->tags[tags->count].value[len] = 0; - } else { - tags->tags[tags->count].value = para_strdup(value); - } - tags->count++; return 1; } -static const char *ID3v1GenreList[] = { - "Blues", "Classic Rock", "Country", "Dance", "Disco", "Funk", - "Grunge", "Hip-Hop", "Jazz", "Metal", "New Age", "Oldies", - "Other", "Pop", "R&B", "Rap", "Reggae", "Rock", - "Techno", "Industrial", "Alternative", "Ska", "Death Metal", "Pranks", - "Soundtrack", "Euro-Techno", "Ambient", "Trip-Hop", "Vocal", - "Jazz+Funk", "Fusion", "Trance", "Classical", "Instrumental", "Acid", - "House", "Game", "Sound Clip", "Gospel", "Noise", "AlternRock", "Bass", - "Soul", "Punk", "Space", "Meditative", "Instrumental Pop", - "Instrumental Rock", "Ethnic", "Gothic", "Darkwave", - "Techno-Industrial", "Electronic", "Pop-Folk", "Eurodance", "Dream", - "Southern Rock", "Comedy", "Cult", "Gangsta", "Top 40", - "Christian Rap", "Pop/Funk", "Jungle", "Native American", "Cabaret", - "New Wave", "Psychadelic", "Rave", "Showtunes", "Trailer", "Lo-Fi", - "Tribal", "Acid Punk", "Acid Jazz", "Polka", "Retro", "Musical", - "Rock & Roll", "Hard Rock", "Folk", "Folk/Rock", "National Folk", - "Swing", "Fast-Fusion", "Bebob", "Latin", "Revival", "Celtic", - "Bluegrass", "Avantgarde", "Gothic Rock", "Progressive Rock", - "Psychedelic Rock", "Symphonic Rock", "Slow Rock", "Big Band", - "Chorus", "Easy Listening", "Acoustic", "Humour", "Speech", "Chanson", - "Opera", "Chamber Music", "Sonata", "Symphony", "Booty Bass", "Primus", - "Porn Groove", "Satire", "Slow Jam", "Club", "Tango", "Samba", - "Folklore", "Ballad", "Power Ballad", "Rhythmic Soul", "Freestyle", - "Duet", "Punk Rock", "Drum Solo", "A capella", "Euro-House", - "Dance Hall", "Goa", "Drum & Bass", "Club House", "Hardcore", "Terror", - "Indie", "BritPop", "NegerPunk", "Polsk Punk", "Beat", - "Christian Gangsta", "Heavy Metal", "Black Metal", "Crossover", - "Contemporary C", "Christian Rock", "Merengue", "Salsa", "Thrash Metal", - "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); - if ((uint32_t)read_data(f, str, length) != length) { - free(str); - str = NULL; - } else - str[length] = 0; - 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 int parse_tag(struct mp4 *f, uint8_t parent, int32_t size) { - uint8_t atom_type; - uint8_t header_size = 0; + int ret; uint64_t subsize, sumsize; - char *name = NULL; - char *data = NULL; - uint32_t done = 0; + char *value = NULL; uint32_t len = 0; uint64_t destpos; + struct mp4_tag *tag; for ( sumsize = 0; - sumsize < size && !f->read_error; /* CVE-2017-9222 */ + sumsize < size; set_position(f, destpos), sumsize += subsize ) { - subsize = atom_read_header(f, &atom_type, &header_size); + uint8_t atom_type; + uint8_t header_size = 0; + ret = atom_read_header(f, &atom_type, &header_size, &subsize); + if (ret <= 0) + goto fail; 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); - } + skip_bytes(f, 8); /* version (1), flags (3), reserved (4) */ + ret = -E_MP4_CORRUPT; + if (subsize < header_size + 8 || subsize > UINT_MAX) + goto fail; + len = subsize - (header_size + 8); + free(value); + value = alloc(len + 1); + ret = read_data(f, value, len); + if (ret <= 0) + goto fail; + value[len] = '\0'; } - if (data) { - if (!done) { - if (name == NULL) - set_metadata_name(parent , &name); - if (name) - tag_add_field(&(f->tags), name, data, len); - } - - free(data); - } - free(name); + if (!value) + return -E_MP4_CORRUPT; + f->meta.tags = para_realloc(f->meta.tags, (f->meta.count + 1) + * sizeof(struct mp4_tag)); + tag = f->meta.tags + f->meta.count; + tag->item = para_strdup(get_metadata_name(parent)); + tag->value = value; + f->meta.count++; return 1; +fail: + free(value); + return ret; } -static int32_t read_mdhd(struct mp4 *f) +static int read_mdhd(struct mp4 *f) { + int ret; uint32_t version; - struct mp4_track *t; + struct mp4_track *t = &f->track; - /* CVE-2017-9221 */ - if (f->total_tracks == 0) - return f->error++; - t = f->track[f->total_tracks - 1]; - - version = read_int32(f); + if (t->state != ATS_INITIAL) + return 1; + ret = read_int32(f, &version); + if (ret <= 0) + return ret; if (version == 1) { - read_int64(f); //creation-time - read_int64(f); //modification-time - t->timeScale = read_int32(f); //timescale - t->duration = read_int64(f); //duration - } else { //version == 0 + skip_bytes(f, 16); /* creation time (8), modification time (8) */ + ret = read_int32(f, &t->time_scale); + if (ret <= 0) + return ret; + ret = read_int64(f, &t->duration); + if (ret <= 0) + return ret; + } else { /* version == 0 */ uint32_t temp; - read_int32(f); //creation-time - read_int32(f); //modification-time - t->timeScale = read_int32(f); //timescale - temp = read_int32(f); + skip_bytes(f, 8); /* creation time (4), modification time (4) */ + ret = read_int32(f, &t->time_scale); + if (ret <= 0) + return ret; + ret = read_int32(f, &temp); + if (ret <= 0) + return ret; t->duration = (temp == (uint32_t) (-1))? (uint64_t) (-1) : (uint64_t) (temp); } - read_int16(f); - read_int16(f); + skip_bytes(f, 4); return 1; } -static int32_t parse_metadata(struct mp4 *f, int32_t size) +static int read_ilst(struct mp4 *f, int32_t size) { - uint64_t subsize, sumsize = 0; - uint8_t atom_type; - uint8_t header_size = 0; + int ret; + uint64_t sumsize = 0; while (sumsize < size) { - subsize = atom_read_header(f, &atom_type, &header_size); - if (subsize == 0) - break; - parse_tag(f, atom_type, (uint32_t)(subsize - header_size)); + uint8_t atom_type; + uint64_t subsize, destpos; + uint8_t header_size = 0; + ret = atom_read_header(f, &atom_type, &header_size, &subsize); + if (ret <= 0) + return ret; + 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: + ret = parse_tag(f, atom_type, subsize - header_size); + if (ret <= 0) + return ret; + } + set_position(f, destpos); sumsize += subsize; } - - return 0; + return 1; } -static int32_t read_meta(struct mp4 *f, uint64_t size) +static int read_meta(struct mp4 *f, uint64_t size) { + int ret; uint64_t subsize, sumsize = 0; uint8_t atom_type; uint8_t header_size = 0; - read_char(f); /* version */ - read_int24(f); /* flags */ - + skip_bytes(f, 4); /* version (1), flags (3) */ while (sumsize < (size - (header_size + 4))) { - subsize = atom_read_header(f, &atom_type, &header_size); + ret = atom_read_header(f, &atom_type, &header_size, &subsize); + if (ret <= 0) + return ret; if (subsize <= header_size + 4) return 1; if (atom_type == ATOM_ILST) { - parse_metadata(f, (uint32_t) (subsize - (header_size + 4))); - } else { + f->ilst_offset = get_position(f) - header_size; + f->ilst_size = subsize; + ret = read_ilst(f, subsize - (header_size + 4)); + if (ret <= 0) + return ret; + } else set_position(f, get_position(f) + subsize - header_size); - } sumsize += subsize; } - - return 0; + return 1; } -static int32_t atom_read(struct mp4 *f, int32_t size, uint8_t atom_type) +static bool need_atom(uint8_t atom_type, bool meta_only) { - uint64_t dest_position = get_position(f) + size - 8; - if (atom_type == ATOM_STSZ) { - /* sample size box */ - read_stsz(f); - } else if (atom_type == ATOM_STTS) { - /* time to sample box */ - read_stts(f); - } else if (atom_type == ATOM_STSC) { - /* sample to chunk box */ - read_stsc(f); - } else if (atom_type == ATOM_STCO) { - /* chunk offset box */ - read_stco(f); - } else if (atom_type == ATOM_STSD) { - /* sample description box */ - read_stsd(f); - } else if (atom_type == ATOM_MDHD) { - /* track header */ - read_mdhd(f); - } else if (atom_type == ATOM_META) { - /* iTunes Metadata box */ - read_meta(f, size); + /* these are needed in any case */ + switch (atom_type) { + case ATOM_STSD: + case ATOM_META: + case ATOM_TRAK: + case ATOM_MDIA: + case ATOM_MINF: + case ATOM_STBL: + case ATOM_UDTA: + return true; } - - set_position(f, dest_position); - return 0; + /* meta-only opens don't need anything else */ + if (meta_only) + return false; + /* these are only required for regular opens */ + switch (atom_type) { + case ATOM_STTS: + case ATOM_STSZ: + case ATOM_STCO: + case ATOM_STSC: + case ATOM_MDHD: + return true; + } + return false; } /* parse atoms that are sub atoms of other atoms */ -static int32_t parse_sub_atoms(struct mp4 *f, uint64_t total_size, int meta_only) +static int parse_sub_atoms(struct mp4 *f, uint64_t total_size, bool meta_only) { - uint64_t size; - uint8_t atom_type = 0; - uint64_t counted_size = 0; - uint8_t header_size = 0; + int ret; + uint64_t dest, size, end = get_position(f) + total_size; - while (counted_size < total_size) { - size = atom_read_header(f, &atom_type, &header_size); - counted_size += size; - - /* check for end of file */ + for (dest = get_position(f); dest < end; set_position(f, dest)) { + uint8_t header_size, atom_type; + ret = atom_read_header(f, &atom_type, &header_size, &size); + if (ret <= 0) + return ret; if (size == 0) - break; - - /* we're starting to read a new track, update index, - * so that all data and tables get written in the right place - */ - if (atom_type == ATOM_TRAK) - track_add(f); - /* parse subatoms */ - if (meta_only && !need_parse_when_meta_only(atom_type)) { - set_position(f, get_position(f) + size - header_size); - } else if (atom_type < SUBATOMIC) { - parse_sub_atoms(f, size - header_size, meta_only); - } else { - atom_read(f, (uint32_t) size, atom_type); + return -E_MP4_CORRUPT; + dest = get_position(f) + size - header_size; + if (atom_type == ATOM_TRAK && f->track.state == ATS_SEEN_MP4A) { + f->track.state = ATS_TRACK_CHANGE; + continue; } - } - - return 0; -} - -/* parse root atoms */ -static int32_t parse_atoms(struct mp4 *f, int meta_only) -{ - uint64_t size; - uint8_t atom_type = 0; - uint8_t header_size = 0; - - f->file_size = 0; - f->read_error = 0; - - while ((size = - atom_read_header(f, &atom_type, &header_size)) != 0) { - f->file_size += size; - f->last_atom = atom_type; - - if (atom_type == ATOM_MOOV && size > header_size) { - f->moov_offset = get_position(f) - header_size; - f->moov_size = size; + if (atom_type == ATOM_UDTA) { + f->udta_offset = get_position(f) - header_size; + f->udta_size = size; } - - /* parse subatoms */ - if (meta_only && !need_parse_when_meta_only(atom_type)) { - set_position(f, get_position(f) + size - header_size); - } else if (atom_type < SUBATOMIC) { - parse_sub_atoms(f, size - header_size, meta_only); - } else { - /* skip this atom */ - set_position(f, get_position(f) + size - header_size); + if (!need_atom(atom_type, meta_only)) + continue; + switch (atom_type) { + case ATOM_STSZ: ret = read_stsz(f); break; + case ATOM_STTS: ret = read_stts(f); break; + case ATOM_STSC: ret = read_stsc(f); break; + case ATOM_STCO: ret = read_stco(f); break; + case ATOM_STSD: ret = read_stsd(f); break; + case ATOM_MDHD: ret = read_mdhd(f); break; + case ATOM_META: + f->meta_offset = get_position(f) - header_size; + f->meta_size = size; + ret = read_meta(f, size); + break; + default: + ret = parse_sub_atoms(f, size - header_size, meta_only); } + if (ret <= 0) + return ret; } - - return 0; + return 1; } -struct mp4 *mp4_open_read(struct mp4_callback *cb) +/** + * Deallocate all resources associated with an mp4 file handle. + * + * \param f File handle returned by \ref mp4_open() or \ref mp4_open_meta(). + * + * This frees the metadata items and various tables which were allocated when + * the file was opened. The given file handle must not be NULL. + */ +void mp4_close(struct mp4 *f) { - struct mp4 *f = para_calloc(sizeof(struct mp4)); - - f->cb = cb; - parse_atoms(f, 0); - if (f->error) { - free(f); - f = NULL; + free(f->track.stsz_table); + free(f->track.stts_sample_count); + free(f->track.stsc_first_chunk); + free(f->track.stsc_samples_per_chunk); + free(f->track.stco_chunk_offset); + for (uint32_t n = 0; n < f->meta.count; n++) { + free(f->meta.tags[n].item); + free(f->meta.tags[n].value); } - return f; + free(f->meta.tags); + free(f); } -static int32_t tag_delete(struct mp4_metadata *tags) +static int open_file(const struct mp4_callback *cb, bool meta_only, struct mp4 **result) { - uint32_t i; - - for (i = 0; i < tags->count; i++) { - free(tags->tags[i].item); - free(tags->tags[i].value); - } - free(tags->tags); - tags->tags = NULL; - tags->count = 0; - - return 0; -} + int ret; + uint64_t size; + uint8_t atom_type, header_size; + struct mp4 *f = zalloc(sizeof(*f)); -void mp4_close(struct mp4 *f) -{ - int32_t 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]); + f->cb = cb; + while ((ret = atom_read_header(f, &atom_type, &header_size, &size)) > 0) { + f->last_atom = atom_type; + if (atom_type != ATOM_MOOV || size <= header_size) { /* skip */ + set_position(f, get_position(f) + size - header_size); + continue; } + f->moov_offset = get_position(f) - header_size; + f->moov_size = size; + ret = parse_sub_atoms(f, size - header_size, meta_only); + if (ret <= 0) + break; } + if (ret < 0) + goto fail; + ret = -E_MP4_TRACK; + if (f->track.channel_count == 0) + goto fail; + ret = -E_MP4_BAD_SAMPLERATE; + if (f->track.sample_rate == 0) + goto fail; + ret = -E_MP4_MISSING_ATOM; + if (f->udta_size == 0 || f->meta_size == 0 || f->ilst_size == 0) + goto fail; + *result = f; + return 1; +fail: + *result = NULL; + mp4_close(f); + return ret; +} - tag_delete(&(f->tags)); - free(f); +/** + * Read the audio track and the metadata of an mp4 file. + * + * \param cb Only the ->read() and ->seek() methods need to be supplied. + * \param result Initialized to a non-NULL pointer iff the function succeeds. + * + * This detects and parses the first audio track and the metadata information + * of the mp4 file. Various error checks are performed after the mp4 atoms have + * been parsed successfully. + * + * This function does not modify the file. However, if the caller intents to + * update the metadata later, the ->write() and ->truncate() methods must be + * supplied in the callback structure. + * + * \return Standard. Several errors are possible. + * + * \sa \ref mp4_open_meta(). + */ +int mp4_open(const struct mp4_callback *cb, struct mp4 **result) +{ + struct mp4 *f; + int ret; + + *result = NULL; + ret = open_file(cb, false, &f); + if (ret < 0) + return ret; + ret = -E_MP4_BAD_SAMPLE_COUNT; + if (f->track.stsz_sample_count == 0) + goto fail; + ret = -E_MP4_CORRUPT; + if (f->track.time_scale == 0) + goto fail; + *result = f; + return 1; +fail: + mp4_close(f); + return ret; } -static int32_t chunk_of_sample(const struct mp4 *f, int32_t track, - int32_t sample, int32_t *chunk_sample, int32_t *chunk) +static int32_t chunk_of_sample(const struct mp4 *f, int32_t sample, + int32_t *chunk) { - int32_t total_entries = 0; - int32_t chunk2entry; - int32_t chunk1, chunk2, chunk1samples, range_samples, total = 0; - - *chunk_sample = 0; - *chunk = 1; - if (f->track[track] == NULL) { - return -1; - } - - total_entries = f->track[track]->stsc_entry_count; + const struct mp4_track *t = &f->track; + uint32_t *fc = t->stsc_first_chunk, *spc = t->stsc_samples_per_chunk; + uint32_t chunk1, chunk1samples, n, total, k; - chunk1 = 1; - chunk1samples = 0; - chunk2entry = 0; - - do { - chunk2 = f->track[track]->stsc_first_chunk[chunk2entry]; - *chunk = chunk2 - chunk1; - range_samples = *chunk * chunk1samples; - - if (sample < total + range_samples) + for (k = 1, total = 0; k < t->stsc_entry_count; k++, total += n) { + n = (fc[k] - fc[k - 1]) * spc[k - 1]; /* number of samples */ + if (sample < total + n) break; - - chunk1samples = f->track[track]->stsc_samples_per_chunk[chunk2entry]; - chunk1 = chunk2; - - if (chunk2entry < total_entries) { - chunk2entry++; - total += range_samples; - } - } while (chunk2entry < total_entries); - - if (chunk1samples) + } + chunk1 = fc[k - 1]; + chunk1samples = spc[k - 1]; + if (chunk1samples != 0) *chunk = (sample - total) / chunk1samples + chunk1; else *chunk = 1; - - *chunk_sample = total + (*chunk - chunk1) * chunk1samples; - - return 0; -} - -static int32_t chunk_to_offset(const struct mp4 *f, int32_t track, - int32_t chunk) -{ - const struct mp4_track *p_track = f->track[track]; - - if (p_track->stco_entry_count && (chunk > p_track->stco_entry_count)) { - return p_track->stco_chunk_offset[p_track->stco_entry_count - - 1]; - } else if (p_track->stco_entry_count) { - return p_track->stco_chunk_offset[chunk - 1]; - } else { - return 8; - } - - return 0; -} - -static int32_t sample_range_size(const struct mp4 *f, int32_t track, - int32_t chunk_sample, int32_t sample) -{ - int32_t i, total; - const struct mp4_track *p_track = f->track[track]; - - if (p_track->stsz_sample_size) { - return (sample - chunk_sample) * p_track->stsz_sample_size; - } else { - if (sample >= p_track->stsz_sample_count) - return 0; //error - - for (i = chunk_sample, total = 0; i < sample; i++) { - total += p_track->stsz_table[i]; - } - } - - return total; -} - -static int32_t sample_to_offset(const struct mp4 *f, int32_t track, - int32_t sample) -{ - int32_t chunk, chunk_sample, chunk_offset1, chunk_offset2; - - chunk_of_sample(f, track, sample, &chunk_sample, &chunk); - - chunk_offset1 = chunk_to_offset(f, track, chunk); - chunk_offset2 = chunk_offset1 + sample_range_size(f, - track, chunk_sample, sample); - return chunk_offset2; + return total + (*chunk - chunk1) * chunk1samples; } /** - * Return the number of milliseconds of the given track. + * Compute the duration of an mp4 file. * - * \param f As returned by \ref mp4_open_read(), must not be NULL. - * \param track Between zero and the value returned by \ref mp4_total_tracks(). + * \param f See \ref mp4_close(). * - * The function returns zero if the audio file is of zero length or contains a - * corrupt track header. + * \return The number of milliseconds of the audio track. This function never + * fails. */ -uint64_t mp4_get_duration(const struct mp4 *f, int32_t track) +uint64_t mp4_get_duration(const struct mp4 *f) { - const struct mp4_track *t = f->track[track]; + const struct mp4_track *t = &f->track; - if (t->timeScale == 0) - return 0; - return t->duration * 1000 / t->timeScale; + return t->duration * 1000 / t->time_scale; } /** - * Check whether the given track number corresponds to an audio track. + * Reposition the read/write file offset. + * + * \param f See \ref mp4_close(). + * \param sample The number of the sample to reposition to. * - * \param f See \ref mp4_get_duration(). - * \param track See \ref mp4_get_duration(). + * The given sample number must be within range, i.e., strictly less than the + * value returned by \ref mp4_num_samples(). * - * Besides audio tracks, an mp4 file may contain video and system tracks. For - * those the function returns false. + * \return Standard. The only possible error is an invalid sample number. */ -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); - set_position(f, offset); -} - -int32_t mp4_get_sample_size(const struct mp4 *f, int track, int sample) -{ - const struct mp4_track *t = f->track[track]; - - if (t->stsz_sample_size != 0) - return t->stsz_sample_size; - return t->stsz_table[sample]; -} - -uint32_t mp4_get_sample_rate(const struct mp4 *f, int32_t track) -{ - return f->track[track]->sampleRate; -} - -uint32_t mp4_get_channel_count(const struct mp4 *f, int32_t track) -{ - return f->track[track]->channelCount; -} - -int32_t mp4_num_samples(const struct mp4 *f, int32_t track) -{ - int32_t i; - int32_t total = 0; - - for (i = 0; i < f->track[track]->stts_entry_count; i++) { - total += f->track[track]->stts_sample_count[i]; - } - return total; -} - -struct mp4 *mp4_open_meta(struct mp4_callback *cb) -{ - struct mp4 *f = para_calloc(sizeof(struct mp4)); - - f->cb = cb; - parse_atoms(f, 1); - if (f->error) { - free(f); - f = NULL; - } - return f; -} - -int32_t mp4_meta_get_num_items(const struct mp4 *f) -{ - return f->tags.count; -} - -int32_t mp4_meta_get_by_index(const struct mp4 *f, uint32_t index, - char **item, char **value) -{ - if (index >= f->tags.count) { - *item = NULL; - *value = NULL; - return 0; - } else { - *item = para_strdup(f->tags.tags[index].item); - *value = para_strdup(f->tags.tags[index].value); - return 1; - } -} - -static uint32_t find_atom(struct mp4 *f, uint64_t base, uint32_t size, - const char *name) -{ - uint32_t remaining = size; - uint64_t atom_offset = base; - for (;;) { - char atom_name[4]; - uint32_t atom_size; - - set_position(f, atom_offset); - - if (remaining < 8) - break; - atom_size = read_int32(f); - if (atom_size > remaining || atom_size < 8) - break; - read_data(f, atom_name, 4); - - if (!memcmp(atom_name, name, 4)) { - set_position(f, atom_offset); - return 1; - } - - remaining -= atom_size; - atom_offset += atom_size; - } - return 0; -} - -static uint32_t find_atom_v2(struct mp4 *f, uint64_t base, uint32_t size, - const char *name, uint32_t extraheaders, const char *name_inside) -{ - uint64_t first_base = (uint64_t) (-1); - while (find_atom(f, base, size, name)) //try to find atom with atom in it - { - uint64_t mybase = get_position(f); - uint32_t mysize = read_int32(f); - - if (first_base == (uint64_t) (-1)) - first_base = mybase; - - if (mysize < 8 + extraheaders) - break; - - if (find_atom (f, mybase + (8 + extraheaders), - mysize - (8 + extraheaders), name_inside)) { - set_position(f, mybase); - return 2; - } - base += mysize; - if (size <= mysize) { - size = 0; - break; - } - size -= mysize; - } - - if (first_base != (uint64_t) (-1)) //wanted atom inside not found - { - set_position(f, first_base); - return 1; - } else - return 0; -} - -struct membuffer { - void *data; - unsigned written; - unsigned allocated; - unsigned error; -}; - -static struct membuffer *membuffer_create(void) -{ - const unsigned initial_size = 256; - - struct membuffer *buf = para_malloc(sizeof(*buf)); - buf->data = para_malloc(initial_size); - buf->written = 0; - buf->allocated = initial_size; - buf->error = buf->data == 0 ? 1 : 0; - - return buf; -} - -static unsigned membuffer_write(struct membuffer *buf, const void *ptr, unsigned bytes) -{ - unsigned dest_size = buf->written + bytes; - - if (buf->error) - return 0; - if (dest_size > buf->allocated) { - do { - buf->allocated <<= 1; - } while (dest_size > buf->allocated); - buf->data = para_realloc(buf->data, buf->allocated); - } - - if (ptr) - memcpy((char *) buf->data + buf->written, ptr, bytes); - buf->written += bytes; - return bytes; -} - -static unsigned membuffer_write_atom_name(struct membuffer *buf, const char *data) -{ - return membuffer_write(buf, data, 4) == 4 ? 1 : 0; -} - -static unsigned membuffer_write_int16(struct membuffer *buf, uint16_t data) -{ - uint8_t temp[2]; - - write_u16_be(temp, data); - return membuffer_write(buf, temp, 2); -} - -static unsigned membuffer_write_int32(struct membuffer *buf, uint32_t data) -{ - uint8_t temp[4]; - write_u32_be(temp, data); - return membuffer_write(buf, temp, 4); -} - -static void membuffer_write_track_tag(struct membuffer *buf, const char *name, - uint32_t index, uint32_t total) -{ - membuffer_write_int32(buf, - 8 /*atom header */ + 8 /*data atom header */ + - 8 /*flags + reserved */ + 8 /*actual data */ ); - membuffer_write_atom_name(buf, name); - membuffer_write_int32(buf, - 8 /*data atom header */ + - 8 /*flags + reserved */ + 8 /*actual data */ ); - membuffer_write_atom_name(buf, "data"); - membuffer_write_int32(buf, 0); //flags - membuffer_write_int32(buf, 0); //reserved - membuffer_write_int16(buf, 0); - membuffer_write_int16(buf, (uint16_t) index); //track number - membuffer_write_int16(buf, (uint16_t) total); //total tracks - membuffer_write_int16(buf, 0); -} - -static void membuffer_write_int16_tag(struct membuffer *buf, const char *name, - uint16_t value) -{ - membuffer_write_int32(buf, - 8 /*atom header */ + 8 /*data atom header */ + - 8 /*flags + reserved */ + 2 /*actual data */ ); - membuffer_write_atom_name(buf, name); - membuffer_write_int32(buf, - 8 /*data atom header */ + - 8 /*flags + reserved */ + 2 /*actual data */ ); - membuffer_write_atom_name(buf, "data"); - membuffer_write_int32(buf, 0); //flags - membuffer_write_int32(buf, 0); //reserved - membuffer_write_int16(buf, value); //value -} - -static uint32_t myatoi(const char *param) -{ - return param ? atoi(param) : 0; -} - -static uint32_t meta_genre_to_index(const char *genrestr) -{ - unsigned n; - for (n = 0; n < sizeof (ID3v1GenreList) / sizeof (ID3v1GenreList[0]); n++) { - if (!strcasecmp(genrestr, ID3v1GenreList[n])) - return n + 1; +int mp4_set_sample_position(struct mp4 *f, uint32_t sample) +{ + const struct mp4_track *t = &f->track; + int32_t offset, chunk, chunk_sample; + uint32_t n, srs; /* sample range size */ + + if (sample >= t->stsz_sample_count) + return -ERRNO_TO_PARA_ERROR(EINVAL); + chunk_sample = chunk_of_sample(f, sample, &chunk); + if (t->stsz_sample_size > 0) + srs = (sample - chunk_sample) * t->stsz_sample_size; + else { + for (srs = 0, n = chunk_sample; n < sample; n++) + srs += t->stsz_table[n]; } - return 0; -} - -struct stdmeta_entry { - const char *atom; - const char *name; -}; - -struct stdmeta_entry stdmetas[] = { - {"\xA9" "nam", "title"}, - {"\xA9" "ART", "artist"}, - {"\xA9" "wrt", "writer"}, - {"\xA9" "alb", "album"}, - {"\xA9" "day", "date"}, - {"\xA9" "too", "tool"}, - {"\xA9" "cmt", "comment"}, - {"cpil", "compilation"}, - {"covr", "cover"}, - {"aART", "album_artist"}, -}; - -static const char *find_standard_meta(const char *name) //returns atom name if found, 0 if not -{ - unsigned n; - for (n = 0; n < sizeof (stdmetas) / sizeof (stdmetas[0]); n++) { - if (!strcasecmp(name, stdmetas[n].name)) - return stdmetas[n].atom; - } - return 0; -} - -static void membuffer_write_std_tag(struct membuffer *buf, const char *name, - const char *value) -{ - uint32_t flags = 1; - - /* special check for compilation flag */ - if (strcmp(name, "cpil") == 0) { - flags = 21; - } - - membuffer_write_int32(buf, - 8 /*atom header */ + 8 /*data atom header */ + - 8 /*flags + reserved */ + strlen(value)); - membuffer_write_atom_name(buf, name); - membuffer_write_int32(buf, - 8 /*data atom header */ + - 8 /*flags + reserved */ + strlen(value)); - membuffer_write_atom_name(buf, "data"); - membuffer_write_int32(buf, flags); //flags - membuffer_write_int32(buf, 0); //reserved - membuffer_write(buf, value, strlen(value)); -} - -static void membuffer_write_custom_tag(struct membuffer *buf, const char *name, - const char *value) -{ - membuffer_write_int32(buf, - 8 /*atom header */ + - 0x1C /*weirdo itunes atom */ + - 12 /*name atom header */ + strlen(name) + - 16 /*data atom header + flags */ + strlen(value)); - membuffer_write_atom_name(buf, "----"); - membuffer_write_int32(buf, 0x1C); //weirdo itunes atom - membuffer_write_atom_name(buf, "mean"); - membuffer_write_int32(buf, 0); - membuffer_write(buf, "com.apple.iTunes", 16); - membuffer_write_int32(buf, 12 + strlen(name)); - membuffer_write_atom_name(buf, "name"); - membuffer_write_int32(buf, 0); - membuffer_write(buf, name, strlen(name)); - membuffer_write_int32(buf, - 8 /*data atom header */ + - 8 /*flags + reserved */ + strlen(value)); - membuffer_write_atom_name(buf, "data"); - membuffer_write_int32(buf, 1); //flags - membuffer_write_int32(buf, 0); //reserved - membuffer_write(buf, value, strlen(value)); -} - -static unsigned membuffer_error(const struct membuffer *buf) -{ - return buf->error; -} - -static void membuffer_free(struct membuffer *buf) -{ - free(buf->data); - free(buf); -} - -static unsigned membuffer_get_size(const struct membuffer *buf) -{ - return buf->written; -} - -static void *membuffer_detach(struct membuffer *buf) -{ - void *ret; - - if (buf->error) - return 0; - ret = para_realloc(buf->data, buf->written); - buf->data = 0; - buf->error = 1; - return ret; -} - -static uint32_t create_ilst(const struct mp4_metadata *data, void **out_buffer, - uint32_t * out_size) -{ - struct membuffer *buf = membuffer_create(); - unsigned metaptr; - char *mask = para_calloc(data->count); - const char *tracknumber_ptr = 0, *totaltracks_ptr = 0; - const char *discnumber_ptr = 0, *totaldiscs_ptr = 0; - const char *genre_ptr = 0, *tempo_ptr = 0; - - for (metaptr = 0; metaptr < data->count; metaptr++) { - struct mp4_tag *tag = &data->tags[metaptr]; - if (!strcasecmp(tag->item, "tracknumber") - || !strcasecmp(tag->item, "track")) { - if (tracknumber_ptr == 0) - tracknumber_ptr = tag->value; - mask[metaptr] = 1; - } else if (!strcasecmp(tag->item, "totaltracks")) { - if (totaltracks_ptr == 0) - totaltracks_ptr = tag->value; - mask[metaptr] = 1; - } else if (!strcasecmp(tag->item, "discnumber") - || !strcasecmp(tag->item, "disc")) { - if (discnumber_ptr == 0) - discnumber_ptr = tag->value; - mask[metaptr] = 1; - } else if (!strcasecmp(tag->item, "totaldiscs")) { - if (totaldiscs_ptr == 0) - totaldiscs_ptr = tag->value; - mask[metaptr] = 1; - } else if (!strcasecmp(tag->item, "genre")) { - if (genre_ptr == 0) - genre_ptr = tag->value; - mask[metaptr] = 1; - } else if (!strcasecmp(tag->item, "tempo")) { - if (tempo_ptr == 0) - tempo_ptr = tag->value; - mask[metaptr] = 1; - } - } - - if (tracknumber_ptr) - membuffer_write_track_tag(buf, "trkn", myatoi(tracknumber_ptr), - myatoi(totaltracks_ptr)); - if (discnumber_ptr) - membuffer_write_track_tag(buf, "disk", myatoi(discnumber_ptr), - myatoi(totaldiscs_ptr)); - if (tempo_ptr) - membuffer_write_int16_tag(buf, "tmpo", myatoi(tempo_ptr)); - - if (genre_ptr) { - uint32_t index = meta_genre_to_index(genre_ptr); - if (index == 0) - membuffer_write_std_tag(buf, "©gen", genre_ptr); - else - membuffer_write_int16_tag(buf, "gnre", index); - } - for (metaptr = 0; metaptr < data->count; metaptr++) { - struct mp4_tag *tag; - const char *std_meta_atom; - - if (mask[metaptr]) - continue; - tag = &data->tags[metaptr]; - std_meta_atom = find_standard_meta(tag->item); - if (std_meta_atom) - membuffer_write_std_tag(buf, std_meta_atom, tag->value); - else - membuffer_write_custom_tag(buf, tag->item, tag->value); - } - free(mask); - - if (membuffer_error(buf)) { - membuffer_free(buf); - return 0; - } - - *out_size = membuffer_get_size(buf); - *out_buffer = membuffer_detach(buf); - membuffer_free(buf); - + if (t->stco_entry_count > 0 && chunk > t->stco_entry_count) + offset = t->stco_chunk_offset[t->stco_entry_count - 1]; + else if (t->stco_entry_count > 0) + offset = t->stco_chunk_offset[chunk - 1]; + else + offset = 8; + set_position(f, offset + srs); return 1; } -static void membuffer_write_atom(struct membuffer *buf, const char *name, unsigned size, - const void *data) +/** + * Look up and return the size of the given sample in the stsz table. + * + * \param f See \ref mp4_close(). + * \param sample The sample number of interest. + * \param result Sample size is returned here. + * + * For the sample argument the restriction mentioned in the documentation of + * \ref mp4_set_sample_position() applies as well. + * + * \return Standard. Like for \ref mp4_set_sample_position(), EINVAL is the + * only possible error. + */ +int mp4_get_sample_size(const struct mp4 *f, uint32_t sample, uint32_t *result) { - membuffer_write_int32(buf, size + 8); - membuffer_write_atom_name(buf, name); - membuffer_write(buf, data, size); -} + const struct mp4_track *t = &f->track; -static void *membuffer_get_ptr(const struct membuffer *buf) -{ - return buf->data; + if (sample >= t->stsz_sample_count) + return -ERRNO_TO_PARA_ERROR(EINVAL); + if (t->stsz_sample_size != 0) + *result = t->stsz_sample_size; + else + *result = t->stsz_table[sample]; + return 1; } -static void membuffer_set_error(struct membuffer *buf) +/** + * Return the sample rate stored in the stsd atom. + * + * \param f See \ref mp4_close(). + * + * The sample rate is a property of the audio track of the mp4 file and is thus + * independent of the sample number. + * + * \return The function always returns a positive value because the open + * operation fails if the sample rate happens to be zero. A typical value is + * 44100. + */ +uint16_t mp4_get_sample_rate(const struct mp4 *f) { - buf->error = 1; + return f->track.sample_rate; } -static unsigned membuffer_transfer_from_file(struct membuffer *buf, struct mp4 *src, - unsigned bytes) +/** + * Return the number of channels of the audio track. + * + * \param f See \ref mp4_close(). + * + * \return The returned channel count is guaranteed to be positive because the + * open operation fails if the mp4a atom is missing or contains a zero channel + * count. + */ +uint16_t mp4_get_channel_count(const struct mp4 *f) { - unsigned oldsize; - void *bufptr; - - oldsize = membuffer_get_size(buf); - if (membuffer_write(buf, 0, bytes) != bytes) - return 0; - - bufptr = membuffer_get_ptr(buf); - if (bufptr == 0) - return 0; - - if ((unsigned)read_data(src, (char *) bufptr + oldsize, bytes) != - bytes) { - membuffer_set_error(buf); - return 0; - } - - return bytes; + return f->track.channel_count; } -static uint32_t create_meta(const struct mp4_metadata *data, void **out_buffer, - uint32_t * out_size) +/** + * Return the number of samples of the audio track. + * + * \param f See \ref mp4_close(). + * + * \return The sample count is read from the stsz atom during open. + */ +uint32_t mp4_num_samples(const struct mp4 *f) { - struct membuffer *buf; - uint32_t ilst_size; - void *ilst_buffer; - - if (!create_ilst(data, &ilst_buffer, &ilst_size)) - return 0; - - buf = membuffer_create(); - - membuffer_write_int32(buf, 0); - membuffer_write_atom(buf, "ilst", ilst_size, ilst_buffer); - free(ilst_buffer); - - *out_size = membuffer_get_size(buf); - *out_buffer = membuffer_detach(buf); - membuffer_free(buf); - return 1; + return f->track.stsz_sample_count; } -static uint32_t create_udta(const struct mp4_metadata *data, void **out_buffer, -uint32_t * out_size) +/** + * Open an mp4 file in metadata-only mode. + * + * \param cb See \ref mp4_open(). + * \param result See \ref mp4_open(). + * + * This is similar to \ref mp4_open() but is cheaper because it only parses the + * metadata of the mp4 file. The only functions that can subsequently be called + * with the file handle returned here are \ref mp4_get_meta() and \ref + * mp4_update_meta(). + * + * \return Standard. + * + * \sa \ref mp4_open(). The comment about ->write() and ->truncate() applies to + * this function as well. + */ +int mp4_open_meta(const struct mp4_callback *cb, struct mp4 **result) { - struct membuffer *buf; - uint32_t meta_size; - void *meta_buffer; + struct mp4 *f; + int ret = open_file(cb, true, &f); - if (!create_meta(data, &meta_buffer, &meta_size)) - return 0; - - buf = membuffer_create(); - - membuffer_write_atom(buf, "meta", meta_size, meta_buffer); - - free(meta_buffer); - - *out_size = membuffer_get_size(buf); - *out_buffer = membuffer_detach(buf); - membuffer_free(buf); + if (ret < 0) + return ret; + *result = f; return 1; } -static uint32_t fix_byte_order_32(uint32_t src) -{ - return read_u32_be(&src); +/** + * Return the metadata of an mp4 file. + * + * \param f See \ref mp4_close(). + * + * The caller is allowed to add, delete or modify the entries of the returned + * structure with the intention to pass the modified version to \ref + * mp4_update_meta(). + * + * \return This never returns NULL, even if the file contains no metadata tag + * items. However, the meta count will be zero and the ->tags pointer NULL in + * this case. + */ +struct mp4_metadata *mp4_get_meta(struct mp4 *f) +{ + return &f->meta; +} + +/** Total length of an on-disk metadata tag. */ +#define TAG_LEN(_len) (24 + (_len)) +static void create_ilst(const struct mp4_metadata *meta, uint8_t *out) +{ + for (unsigned n = 0; n < meta->count; n++) { + struct mp4_tag *tag = meta->tags + n; + unsigned len = strlen(tag->value); + const char *atom_name; + + if (!strcasecmp(tag->item, "title")) + atom_name = "\xA9" "nam"; + else if (!strcasecmp(tag->item, "artist")) + atom_name = "\xA9" "ART"; + else if (!strcasecmp(tag->item, "album")) + atom_name = "\xA9" "alb"; + else if (!strcasecmp(tag->item, "date")) + atom_name = "\xA9" "day"; + else if (!strcasecmp(tag->item, "comment")) + atom_name = "\xA9" "cmt"; + else + assert(false); + write_u32_be(out, TAG_LEN(len)); + memcpy(out + 4, atom_name, 4); + write_u32_be(out + 8, 8 /* data atom header */ + + 8 /* flags + reserved */ + + len); + memcpy(out + 12, "data", 4); + write_u32_be(out + 16, 1); /* flags */ + write_u32_be(out + 20, 0); /* reserved */ + memcpy(out + 24, tag->value, len); + out += TAG_LEN(len); + } } -static uint32_t modify_moov(struct mp4 *f, const struct mp4_metadata *data, - void **out_buffer, uint32_t * out_size) +static void *modify_moov(struct mp4 *f, uint32_t *out_size) { + int ret; uint64_t total_base = f->moov_offset + 8; - uint32_t total_size = (uint32_t) (f->moov_size - 8); - - uint64_t udta_offset, meta_offset, ilst_offset; - uint32_t udta_size, meta_size, ilst_size; - - uint32_t new_ilst_size; - void *new_ilst_buffer; - + uint32_t total_size = f->moov_size - 8; + uint32_t new_ilst_size = 0; + void *out_buffer; uint8_t *p_out; int32_t size_delta; - - if (!find_atom_v2(f, total_base, total_size, "udta", 0, "meta")) { - struct membuffer *buf; - void *new_udta_buffer; - uint32_t new_udta_size; - if (!create_udta(data, &new_udta_buffer, &new_udta_size)) - return 0; - - buf = membuffer_create(); - set_position(f, total_base); - membuffer_transfer_from_file(buf, f, total_size); - - membuffer_write_atom(buf, "udta", new_udta_size, - new_udta_buffer); - - free(new_udta_buffer); - - *out_size = membuffer_get_size(buf); - *out_buffer = membuffer_detach(buf); - membuffer_free(buf); - return 1; - } else { - udta_offset = get_position(f); - udta_size = read_int32(f); - if (!find_atom_v2 (f, udta_offset + 8, udta_size - 8, "meta", 4, "ilst")) { - struct membuffer *buf; - void *new_meta_buffer; - uint32_t new_meta_size; - if (!create_meta(data, &new_meta_buffer, &new_meta_size)) - return 0; - - buf = membuffer_create(); - set_position(f, total_base); - membuffer_transfer_from_file(buf, f, - (uint32_t)(udta_offset - total_base)); - - membuffer_write_int32(buf, udta_size + 8 + new_meta_size); - membuffer_write_atom_name(buf, "udta"); - membuffer_transfer_from_file(buf, f, udta_size); - - membuffer_write_atom(buf, "meta", new_meta_size, - new_meta_buffer); - free(new_meta_buffer); - - *out_size = membuffer_get_size(buf); - *out_buffer = membuffer_detach(buf); - membuffer_free(buf); - return 1; + uint32_t tmp; + + for (unsigned n = 0; n < f->meta.count; n++) + new_ilst_size += TAG_LEN(strlen(f->meta.tags[n].value)); + size_delta = new_ilst_size - (f->ilst_size - 8); + *out_size = total_size + size_delta; + out_buffer = alloc(*out_size); + p_out = out_buffer; + set_position(f, total_base); + ret = read_data(f, p_out, f->udta_offset - total_base); + if (ret <= 0) + return NULL; + p_out += f->udta_offset - total_base; + ret = read_int32(f, &tmp); + if (ret <= 0) + return NULL; + write_u32_be(p_out, tmp + size_delta); + p_out += 4; + ret = read_data(f, p_out, 4); + if (ret <= 0) + return NULL; + p_out += 4; + ret = read_data(f, p_out, f->meta_offset - f->udta_offset - 8); + if (ret <= 0) + return NULL; + p_out += f->meta_offset - f->udta_offset - 8; + ret = read_int32(f, &tmp); + if (ret <= 0) + return NULL; + write_u32_be(p_out, tmp + size_delta); + p_out += 4; + ret = read_data(f, p_out, 4); + if (ret <= 0) + return NULL; + p_out += 4; + ret = read_data(f, p_out, f->ilst_offset - f->meta_offset - 8); + if (ret <= 0) + return NULL; + p_out += f->ilst_offset - f->meta_offset - 8; + ret = read_int32(f, &tmp); + if (ret <= 0) + return NULL; + write_u32_be(p_out, tmp + size_delta); + p_out += 4; + ret = read_data(f, p_out, 4); + if (ret <= 0) + return NULL; + p_out += 4; + create_ilst(&f->meta, p_out); + p_out += new_ilst_size; + set_position(f, f->ilst_offset + f->ilst_size); + ret = read_data(f, p_out, total_size - (f->ilst_offset - total_base) + - f->ilst_size); + if (ret <= 0) + return NULL; + return out_buffer; +} + +static int write_data(struct mp4 *f, void *data, size_t size) +{ + while (size > 0) { + ssize_t ret = f->cb->write(f->cb->user_data, data, size); + if (ret < 0) { + if (errno == EINTR) + continue; + return -ERRNO_TO_PARA_ERROR(errno); } - meta_offset = get_position(f); - meta_size = read_int32(f); - if (!find_atom(f, meta_offset + 12, meta_size - 12, "ilst")) - return 0; //shouldn't happen, find_atom_v2 above takes care of it - ilst_offset = get_position(f); - ilst_size = read_int32(f); - - if (!create_ilst(data, &new_ilst_buffer, &new_ilst_size)) - return 0; - - size_delta = new_ilst_size - (ilst_size - 8); - - *out_size = total_size + size_delta; - *out_buffer = para_malloc(*out_size); - p_out = (uint8_t *) * out_buffer; - - set_position(f, total_base); - read_data(f, p_out, - (uint32_t) (udta_offset - total_base)); - p_out += (uint32_t) (udta_offset - total_base); - *(uint32_t *) p_out = fix_byte_order_32(read_int32(f) + size_delta); - p_out += 4; - read_data(f, p_out, 4); - p_out += 4; - read_data(f, p_out, - (uint32_t) (meta_offset - udta_offset - 8)); - p_out += (uint32_t) (meta_offset - udta_offset - 8); - *(uint32_t *) p_out = fix_byte_order_32(read_int32(f) + size_delta); - p_out += 4; - read_data(f, p_out, 4); - p_out += 4; - read_data(f, p_out, - (uint32_t) (ilst_offset - meta_offset - 8)); - p_out += (uint32_t) (ilst_offset - meta_offset - 8); - *(uint32_t *) p_out = fix_byte_order_32(read_int32(f) + size_delta); - p_out += 4; - read_data(f, p_out, 4); - p_out += 4; - - memcpy(p_out, new_ilst_buffer, new_ilst_size); - p_out += new_ilst_size; - - set_position(f, ilst_offset + ilst_size); - read_data(f, p_out, (uint32_t) (total_size - - (ilst_offset - total_base) - ilst_size)); - - free(new_ilst_buffer); + size -= ret; } return 1; } -static int32_t write_data(struct mp4 *f, void *data, uint32_t size) -{ - int32_t result = 1; - - result = f->cb->write(f->cb->user_data, data, size); - - f->current_position += size; - - return result; -} - -static int32_t write_int32(struct mp4 *f, uint32_t data) -{ - int8_t temp[4]; - write_u32_be(temp, data); - return write_data(f, temp, sizeof(temp)); -} - -int32_t mp4_meta_update(struct mp4_callback *cb, const struct mp4_metadata *data) +/** + * Write back the modified metadata items to the mp4 file. + * + * This is the only public function which modifies the contents of an mp4 file. + * This is achieved by calling the ->write() and ->truncate() methods of the + * callback structure passed to \ref mp4_open() or \ref mp4_open_meta(). + * + * \param f See \ref mp4_close(). + * + * The modified metadata structure does not need to be supplied to this + * function because it is part of the mp4 structure. + * + * \return Standard. + */ +int mp4_update_meta(struct mp4 *f) { void *new_moov_data; uint32_t new_moov_size; + uint8_t buf[8] = "----moov"; + int ret; - 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)) { + new_moov_data = modify_moov(f, &new_moov_size); + if (!new_moov_data ) { mp4_close(f); return 0; } - - /* copy moov atom to end of the file */ if (f->last_atom != ATOM_MOOV) { - char *free_data = "free"; - - /* rename old moov to free */ set_position(f, f->moov_offset + 4); - write_data(f, free_data, 4); - - set_position(f, f->file_size); - write_int32(f, new_moov_size + 8); - write_data(f, "moov", 4); - write_data(f, new_moov_data, new_moov_size); - } else { + ret = write_data(f, "free", 4); /* rename old moov to free */ + if (ret < 0) + goto free_moov; + /* write new moov atom at EOF */ + f->cb->seek(f->cb->user_data, 0, SEEK_END); + } else /* overwrite old moov atom */ set_position(f, f->moov_offset); - write_int32(f, new_moov_size + 8); - write_data(f, "moov", 4); - write_data(f, new_moov_data, new_moov_size); - } - cb->truncate(cb->user_data); - mp4_close(f); - return 1; -} - -/* find a metadata item by name */ -/* returns 0 if item found, 1 if no such item */ -static int32_t meta_find_by_name(const struct mp4 *f, const char *item, - char **value) -{ - uint32_t i; - - for (i = 0; i < f->tags.count; i++) { - if (!strcasecmp(f->tags.tags[i].item, item)) { - *value = para_strdup(f->tags.tags[i].value); - return 1; - } - } - - *value = NULL; - - /* not found */ - return 0; -} - -int32_t mp4_meta_get_artist(const struct mp4 *f, char **value) -{ - return meta_find_by_name(f, "artist", value); -} - -int32_t mp4_meta_get_title(const struct mp4 *f, char **value) -{ - return meta_find_by_name(f, "title", value); -} - -int32_t mp4_meta_get_date(const struct mp4 *f, char **value) -{ - return meta_find_by_name(f, "date", value); -} - -int32_t mp4_meta_get_album(const struct mp4 *f, char **value) -{ - return meta_find_by_name(f, "album", value); + write_u32_be(buf, new_moov_size + 8); + ret = write_data(f, buf, sizeof(buf)); + if (ret < 0) + goto free_moov; + ret = write_data(f, new_moov_data, new_moov_size); + if (ret < 0) + goto free_moov; + ret = f->cb->truncate(f->cb->user_data); + if (ret < 0) + ret = -ERRNO_TO_PARA_ERROR(errno); +free_moov: + free(new_moov_data); + return ret; } -int32_t mp4_meta_get_comment(const struct mp4 *f, char **value) +/** + * Return the value of the given tag item. + * + * \param f See \ref mp4_close(). + * \param item "artist", "title", "album", "comment", or "date". + * + * \return The function returns NULL if the given item is not in the above + * list. Otherwise, if the file does not contain a tag for the given item, the + * function also returns NULL. Otherwise a copy of the tag value is returned + * and the caller should free this memory when it is no longer needed. + */ +__malloc char *mp4_get_tag_value(const struct mp4 *f, const char *item) { - return meta_find_by_name(f, "comment", value); + for (unsigned n = 0; n < f->meta.count; n++) + if (!strcasecmp(f->meta.tags[n].item, item)) + return para_strdup(f->meta.tags[n].value); + return NULL; }