atom_parser.py

#!/usr/bin/env python

# https://stackoverflow.com/a/18552833
# ISO/IEC 14496-12 - ISO base media file format
#   from: https://standards.iso.org/ittf/PubliclyAvailableStandards/index.html

from sys import argv
import heapq
import os

if len(argv) < 2:
    print("usage: <file.mp4>")
    exit(0)

to_enter = ['moov', 'trak', 'mdia', 'minf', 'stbl', 'edts']
#target = 'stsc'
#target = 'stts'
#target = 'stsz'
#target = 'elst'
target = 'ctts'
#trak_idx = 1
trak_idx = 1

if len(argv) > 2:
    trak_idx = int(argv[2])
print(f"using track_idx = {trak_idx}")
print()

to_name = {
    'stts' : 'Time-to-sample',
    'stsc' : 'Sample-to-chunk',
    'stco' : 'Chunk offset',
    'stss' : 'Sync sample',
    'stsz' : 'Sample size',
    'elst' : 'Edit List',
    'ctts' : 'Composition Offset',
}


f = open(argv[1], 'rb')

def read_utf(subj, n):
    try:
        name = f.read(n).decode('utf-8')
    except UnicodeDecodeError as e:
        print(f"bad {subj}: {e}")
        exit(1)
    return name

def read_len_and_name():
    length = int.from_bytes(f.read(4), byteorder='big') - 8
    name = read_utf('atom name', 4)
    if length == -7:
        length = int.from_bytes(f.read(8), byteorder='big') - 8
    return length, name

def skip(n):
    f.seek(n, os.SEEK_CUR)

cn = "?"
def parse_stsd(n):
    global cn
    skip(12)
    cn = read_utf('codec name', 4)
    skip(n-16)

def find_cn(cur_len):
    orig_off = f.tell()
    skip(cur_len)
    while cn == "?":
        proceed_atom()
    f.seek(orig_off)

lends = parent_ends = []
level = 0
def proceed_atom():
    global trak_idx, level

    if lends and lends[0] <= f.tell():
        heapq.heappop(lends)
        level -= 2

    l, n = read_len_and_name()
    print(f"{' '*level}{n} {l}")
    if l < 0:
        print(f"bad length: {l}")
        exit(1)

    if n == 'trak' and trak_idx:
        trak_idx -= 1
        skip(l)
    elif n == 'stsd':
        parse_stsd(l)
    elif n in to_enter:
        level += 2
        heapq.heappush(lends, f.tell() + l)
    elif n != target:
        skip(l)

    return l, n

while True:
    l, n = proceed_atom()
    if n == target:
        find_cn(l)
        break

def dump_stsc():
    skip(4)
    num_entries = int.from_bytes(f.read(4), byteorder='big')
    print(f"num_entries: {num_entries}")
    print(f"  i   first_chunk  sample_per_chunk  sample_descr_idx")
    for i in range(min(num_entries, 50)):
        first_chunk = int.from_bytes(f.read(4), byteorder='big')
        sample_per_chunk = int.from_bytes(f.read(4), byteorder='big')
        sample_descr_idx = int.from_bytes(f.read(4), byteorder='big')
        print(f"{i:>4} {first_chunk:>9} {sample_per_chunk:>15} {sample_descr_idx:>15}")

def dump_stts():
    skip(4)
    num_entries = int.from_bytes(f.read(4), byteorder='big')
    print(f"num_entries: {num_entries}")
    print(f"  i     n_samples       time/duration")
    for i in range(min(num_entries, 50)):
        n_samples = int.from_bytes(f.read(4), byteorder='big')
        time = int.from_bytes(f.read(4), byteorder='big')
        print(f"{i:>4} {n_samples:>9} {time:>17}")

def dump_stsz():
    skip(4)
    const_sz = int.from_bytes(f.read(4), byteorder='big')
    num_entries = int.from_bytes(f.read(4), byteorder='big')
    print(f"const_sz: {const_sz}")
    print(f"num_entries: {num_entries}")
    if const_sz:
        return
    print(f"  i     size")
    for i in range(min(num_entries, 50)):
        size = int.from_bytes(f.read(4), byteorder='big')
        print(f"{i:>4} {size:>9}")

def dump_elst():
    skip(4)
    num_entries = int.from_bytes(f.read(4), byteorder='big')
    print(f"num_entries: {num_entries}")
    print(f" i   track-dur    media-time    media-rate")
    for i in range(min(num_entries, 50)):
        tr_dur = int.from_bytes(f.read(4), byteorder='big')
        me_time = int.from_bytes(f.read(4), byteorder='big')
        me_rate = int.from_bytes(f.read(4), byteorder='big')
        print(f"{i:>3}  {tr_dur:>9} {me_time:>13} {me_rate:>9}")

def dump_ctts():
    ver = int.from_bytes(f.read(1), byteorder='big')
    print(f"ver: {ver}")
    skip(3)
    num_entries = int.from_bytes(f.read(4), byteorder='big')
    print(f"num_entries: {num_entries}")
    print(f" i  SampleIdx  SampleCount    compositionOffset")
    total_cnt = 0
    pairs = []
    for i in range(min(num_entries, 500)):
        sample_cnt = int.from_bytes(f.read(4), byteorder='big')
        comp_off = int.from_bytes(f.read(4), byteorder='big')
        total_cnt += sample_cnt
        pairs.append((sample_cnt, comp_off))
        print(f"{i:>3} {total_cnt:>9} {sample_cnt:>9} {comp_off:>13}")
    print(f"{total_cnt = }") 
    print(pairs)

print(f"found '{n}' (= {to_name[n]}) in '{cn}'-track  // at {f.tell()}\n")

if target == "stsc": dump_stsc()
elif target == "stts": dump_stts()
elif target == "stsz": dump_stsz()
elif target == "elst": dump_elst()
elif target == "ctts": dump_ctts()