Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/downloader/ism.py

   1 from __future__ import unicode_literals
   2
   3 import os
   4 import time
   5 import struct
   6 import binascii
   7 import io
   8
   9 from .fragment import FragmentFD
  10 from ..compat import compat_urllib_error
  11 from ..utils import (
  12     sanitize_open,
  13     encodeFilename,
  14 )
  15
  16
  17 u8 = struct.Struct(b'>B')
  18 u88 = struct.Struct(b'>Bx')
  19 u16 = struct.Struct(b'>H')
  20 u1616 = struct.Struct(b'>Hxx')
  21 u32 = struct.Struct(b'>I')
  22 u64 = struct.Struct(b'>Q')
  23
  24 s88 = struct.Struct(b'>bx')
  25 s16 = struct.Struct(b'>h')
  26 s1616 = struct.Struct(b'>hxx')
  27 s32 = struct.Struct(b'>i')
  28
  29 unity_matrix = (s32.pack(0x10000) + s32.pack(0) * 3) * 2 + s32.pack(0x40000000)
  30
  31 TRACK_ENABLED = 0x1
  32 TRACK_IN_MOVIE = 0x2
  33 TRACK_IN_PREVIEW = 0x4
  34
  35 SELF_CONTAINED = 0x1
  36
  37
  38 def box(box_type, payload):
  39     return u32.pack(8 + len(payload)) + box_type + payload
  40
  41
  42 def full_box(box_type, version, flags, payload):
  43     return box(box_type, u8.pack(version) + u32.pack(flags)[1:] + payload)
  44
  45
  46 def write_piff_header(stream, params):
  47     track_id = params['track_id']
  48     fourcc = params['fourcc']
  49     duration = params['duration']
  50     timescale = params.get('timescale', 10000000)
  51     language = params.get('language', 'und')
  52     height = params.get('height', 0)
  53     width = params.get('width', 0)
  54     is_audio = width == 0 and height == 0
  55     creation_time = modification_time = int(time.time())
  56
  57     ftyp_payload = b'isml'  # major brand
  58     ftyp_payload += u32.pack(1)  # minor version
  59     ftyp_payload += b'piff' + b'iso2'  # compatible brands
  60     stream.write(box(b'ftyp', ftyp_payload))  # File Type Box
  61
  62     mvhd_payload = u64.pack(creation_time)
  63     mvhd_payload += u64.pack(modification_time)
  64     mvhd_payload += u32.pack(timescale)
  65     mvhd_payload += u64.pack(duration)
  66     mvhd_payload += s1616.pack(1)  # rate
  67     mvhd_payload += s88.pack(1)  # volume
  68     mvhd_payload += u16.pack(0)  # reserved
  69     mvhd_payload += u32.pack(0) * 2  # reserved
  70     mvhd_payload += unity_matrix
  71     mvhd_payload += u32.pack(0) * 6  # pre defined
  72     mvhd_payload += u32.pack(0xffffffff)  # next track id
  73     moov_payload = full_box(b'mvhd', 1, 0, mvhd_payload)  # Movie Header Box
  74
  75     tkhd_payload = u64.pack(creation_time)
  76     tkhd_payload += u64.pack(modification_time)
  77     tkhd_payload += u32.pack(track_id)  # track id
  78     tkhd_payload += u32.pack(0)  # reserved
  79     tkhd_payload += u64.pack(duration)
  80     tkhd_payload += u32.pack(0) * 2  # reserved
  81     tkhd_payload += s16.pack(0)  # layer
  82     tkhd_payload += s16.pack(0)  # alternate group
  83     tkhd_payload += s88.pack(1 if is_audio else 0)  # volume
  84     tkhd_payload += u16.pack(0)  # reserved
  85     tkhd_payload += unity_matrix
  86     tkhd_payload += u1616.pack(width)
  87     tkhd_payload += u1616.pack(height)
  88     trak_payload = full_box(b'tkhd', 1, TRACK_ENABLED | TRACK_IN_MOVIE | TRACK_IN_PREVIEW, tkhd_payload)  # Track Header Box
  89
  90     mdhd_payload = u64.pack(creation_time)
  91     mdhd_payload += u64.pack(modification_time)
  92     mdhd_payload += u32.pack(timescale)
  93     mdhd_payload += u64.pack(duration)
  94     mdhd_payload += u16.pack(((ord(language[0]) - 0x60) << 10) | ((ord(language[1]) - 0x60) << 5) | (ord(language[2]) - 0x60))
  95     mdhd_payload += u16.pack(0)  # pre defined
  96     mdia_payload = full_box(b'mdhd', 1, 0, mdhd_payload)  # Media Header Box
  97
  98     hdlr_payload = u32.pack(0)  # pre defined
  99     hdlr_payload += b'soun' if is_audio else b'vide'  # handler type
 100     hdlr_payload += u32.pack(0) * 3  # reserved
 101     hdlr_payload += (b'Sound' if is_audio else b'Video') + b'Handler\0'  # name
 102     mdia_payload += full_box(b'hdlr', 0, 0, hdlr_payload)  # Handler Reference Box
 103
 104     if is_audio:
 105         smhd_payload = s88.pack(0)  # balance
 106         smhd_payload = u16.pack(0)  # reserved
 107         media_header_box = full_box(b'smhd', 0, 0, smhd_payload)  # Sound Media Header
 108     else:
 109         vmhd_payload = u16.pack(0)  # graphics mode
 110         vmhd_payload += u16.pack(0) * 3  # opcolor
 111         media_header_box = full_box(b'vmhd', 0, 1, vmhd_payload)  # Video Media Header
 112     minf_payload = media_header_box
 113
 114     dref_payload = u32.pack(1)  # entry count
 115     dref_payload += full_box(b'url ', 0, SELF_CONTAINED, b'')  # Data Entry URL Box
 116     dinf_payload = full_box(b'dref', 0, 0, dref_payload)  # Data Reference Box
 117     minf_payload += box(b'dinf', dinf_payload)  # Data Information Box
 118
 119     stsd_payload = u32.pack(1)  # entry count
 120
 121     sample_entry_payload = u8.pack(0) * 6  # reserved
 122     sample_entry_payload += u16.pack(1)  # data reference index
 123     if is_audio:
 124         sample_entry_payload += u32.pack(0) * 2  # reserved
 125         sample_entry_payload += u16.pack(params.get('channels', 2))
 126         sample_entry_payload += u16.pack(params.get('bits_per_sample', 16))
 127         sample_entry_payload += u16.pack(0)  # pre defined
 128         sample_entry_payload += u16.pack(0)  # reserved
 129         sample_entry_payload += u1616.pack(params['sampling_rate'])
 130
 131         if fourcc == 'AACL':
 132             sample_entry_box = box(b'mp4a', sample_entry_payload)
 133     else:
 134         sample_entry_payload = sample_entry_payload
 135         sample_entry_payload += u16.pack(0)  # pre defined
 136         sample_entry_payload += u16.pack(0)  # reserved
 137         sample_entry_payload += u32.pack(0) * 3  # pre defined
 138         sample_entry_payload += u16.pack(width)
 139         sample_entry_payload += u16.pack(height)
 140         sample_entry_payload += u1616.pack(0x48)  # horiz resolution 72 dpi
 141         sample_entry_payload += u1616.pack(0x48)  # vert resolution 72 dpi
 142         sample_entry_payload += u32.pack(0)  # reserved
 143         sample_entry_payload += u16.pack(1)  # frame count
 144         sample_entry_payload += u8.pack(0) * 32  # compressor name
 145         sample_entry_payload += u16.pack(0x18)  # depth
 146         sample_entry_payload += s16.pack(-1)  # pre defined
 147
 148         codec_private_data = binascii.unhexlify(params['codec_private_data'])
 149         if fourcc in ('H264', 'AVC1'):
 150             sps, pps = codec_private_data.split(u32.pack(1))[1:]
 151             avcc_payload = u8.pack(1)  # configuration version
 152             avcc_payload += sps[1:4]  # avc profile indication + profile compatibility + avc level indication
 153             avcc_payload += u8.pack(0xfc | (params.get('nal_unit_length_field', 4) - 1))  # complete represenation (1) + reserved (11111) + length size minus one
 154             avcc_payload += u8.pack(1)  # reserved (0) + number of sps (0000001)
 155             avcc_payload += u16.pack(len(sps))
 156             avcc_payload += sps
 157             avcc_payload += u8.pack(1)  # number of pps
 158             avcc_payload += u16.pack(len(pps))
 159             avcc_payload += pps
 160             sample_entry_payload += box(b'avcC', avcc_payload)  # AVC Decoder Configuration Record
 161             sample_entry_box = box(b'avc1', sample_entry_payload)  # AVC Simple Entry
 162     stsd_payload += sample_entry_box
 163
 164     stbl_payload = full_box(b'stsd', 0, 0, stsd_payload)  # Sample Description Box
 165
 166     stts_payload = u32.pack(0)  # entry count
 167     stbl_payload += full_box(b'stts', 0, 0, stts_payload)  # Decoding Time to Sample Box
 168
 169     stsc_payload = u32.pack(0)  # entry count
 170     stbl_payload += full_box(b'stsc', 0, 0, stsc_payload)  # Sample To Chunk Box
 171
 172     stco_payload = u32.pack(0)  # entry count
 173     stbl_payload += full_box(b'stco', 0, 0, stco_payload)  # Chunk Offset Box
 174
 175     minf_payload += box(b'stbl', stbl_payload)  # Sample Table Box
 176
 177     mdia_payload += box(b'minf', minf_payload)  # Media Information Box
 178
 179     trak_payload += box(b'mdia', mdia_payload)  # Media Box
 180
 181     moov_payload += box(b'trak', trak_payload)  # Track Box
 182
 183     mehd_payload = u64.pack(duration)
 184     mvex_payload = full_box(b'mehd', 1, 0, mehd_payload)  # Movie Extends Header Box
 185
 186     trex_payload = u32.pack(track_id)  # track id
 187     trex_payload += u32.pack(1)  # default sample description index
 188     trex_payload += u32.pack(0)  # default sample duration
 189     trex_payload += u32.pack(0)  # default sample size
 190     trex_payload += u32.pack(0)  # default sample flags
 191     mvex_payload += full_box(b'trex', 0, 0, trex_payload)  # Track Extends Box
 192
 193     moov_payload += box(b'mvex', mvex_payload)  # Movie Extends Box
 194     stream.write(box(b'moov', moov_payload))  # Movie Box
 195
 196
 197 def extract_box_data(data, box_sequence):
 198     data_reader = io.BytesIO(data)
 199     while True:
 200         box_size = u32.unpack(data_reader.read(4))[0]
 201         box_type = data_reader.read(4)
 202         if box_type == box_sequence[0]:
 203             box_data = data_reader.read(box_size - 8)
 204             if len(box_sequence) == 1:
 205                 return box_data
 206             return extract_box_data(box_data, box_sequence[1:])
 207         data_reader.seek(box_size - 8, 1)
 208
 209
 210 class IsmFD(FragmentFD):
 211     """
 212     Download segments in a ISM manifest
 213     """
 214
 215     FD_NAME = 'ism'
 216
 217     def real_download(self, filename, info_dict):
 218         segments = info_dict['fragments'][:1] if self.params.get(
 219             'test', False) else info_dict['fragments']
 220
 221         ctx = {
 222             'filename': filename,
 223             'total_frags': len(segments),
 224         }
 225
 226         self._prepare_and_start_frag_download(ctx)
 227
 228         segments_filenames = []
 229
 230         fragment_retries = self.params.get('fragment_retries', 0)
 231         skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
 232
 233         track_written = False
 234         for i, segment in enumerate(segments):
 235             segment_url = segment['url']
 236             segment_name = 'Frag%d' % i
 237             target_filename = '%s-%s' % (ctx['tmpfilename'], segment_name)
 238             count = 0
 239             while count <= fragment_retries:
 240                 try:
 241                     success = ctx['dl'].download(target_filename, {'url': segment_url})
 242                     if not success:
 243                         return False
 244                     down, target_sanitized = sanitize_open(target_filename, 'rb')
 245                     down_data = down.read()
 246                     if not track_written:
 247                         tfhd_data = extract_box_data(down_data, [b'moof', b'traf', b'tfhd'])
 248                         info_dict['_download_params']['track_id'] = u32.unpack(tfhd_data[4:8])[0]
 249                         write_piff_header(ctx['dest_stream'], info_dict['_download_params'])
 250                         track_written = True
 251                     ctx['dest_stream'].write(down_data)
 252                     down.close()
 253                     segments_filenames.append(target_sanitized)
 254                     break
 255                 except compat_urllib_error.HTTPError as err:
 256                     count += 1
 257                     if count <= fragment_retries:
 258                         self.report_retry_fragment(err, segment_name, count, fragment_retries)
 259             if count > fragment_retries:
 260                 if skip_unavailable_fragments:
 261                     self.report_skip_fragment(segment_name)
 262                     continue
 263                 self.report_error('giving up after %s fragment retries' % fragment_retries)
 264                 return False
 265
 266         self._finish_frag_download(ctx)
 267
 268         for segment_file in segments_filenames:
 269             os.remove(encodeFilename(segment_file))
 270
 271         return True