_ Git - youtube-dl/blob - youtube_dl/downloader/f4m.py

   1 from __future__ import division, unicode_literals
   2
   3 import base64
   4 import io
   5 import itertools
   6 import os
   7 import time
   8 import xml.etree.ElementTree as etree
   9
  10 from .common import FileDownloader
  11 from .http import HttpFD
  12 from ..compat import (
  13     compat_urlparse,
  14 )
  15 from ..utils import (
  16     struct_pack,
  17     struct_unpack,
  18     encodeFilename,
  19     sanitize_open,
  20     xpath_text,
  21 )
  22
  23
  24 class FlvReader(io.BytesIO):
  25     """
  26     Reader for Flv files
  27     The file format is documented in https://www.adobe.com/devnet/f4v.html
  28     """
  29
  30     # Utility functions for reading numbers and strings
  31     def read_unsigned_long_long(self):
  32         return struct_unpack('!Q', self.read(8))[0]
  33
  34     def read_unsigned_int(self):
  35         return struct_unpack('!I', self.read(4))[0]
  36
  37     def read_unsigned_char(self):
  38         return struct_unpack('!B', self.read(1))[0]
  39
  40     def read_string(self):
  41         res = b''
  42         while True:
  43             char = self.read(1)
  44             if char == b'\x00':
  45                 break
  46             res += char
  47         return res
  48
  49     def read_box_info(self):
  50         """
  51         Read a box and return the info as a tuple: (box_size, box_type, box_data)
  52         """
  53         real_size = size = self.read_unsigned_int()
  54         box_type = self.read(4)
  55         header_end = 8
  56         if size == 1:
  57             real_size = self.read_unsigned_long_long()
  58             header_end = 16
  59         return real_size, box_type, self.read(real_size - header_end)
  60
  61     def read_asrt(self):
  62         # version
  63         self.read_unsigned_char()
  64         # flags
  65         self.read(3)
  66         quality_entry_count = self.read_unsigned_char()
  67         # QualityEntryCount
  68         for i in range(quality_entry_count):
  69             self.read_string()
  70
  71         segment_run_count = self.read_unsigned_int()
  72         segments = []
  73         for i in range(segment_run_count):
  74             first_segment = self.read_unsigned_int()
  75             fragments_per_segment = self.read_unsigned_int()
  76             segments.append((first_segment, fragments_per_segment))
  77
  78         return {
  79             'segment_run': segments,
  80         }
  81
  82     def read_afrt(self):
  83         # version
  84         self.read_unsigned_char()
  85         # flags
  86         self.read(3)
  87         # time scale
  88         self.read_unsigned_int()
  89
  90         quality_entry_count = self.read_unsigned_char()
  91         # QualitySegmentUrlModifiers
  92         for i in range(quality_entry_count):
  93             self.read_string()
  94
  95         fragments_count = self.read_unsigned_int()
  96         fragments = []
  97         for i in range(fragments_count):
  98             first = self.read_unsigned_int()
  99             first_ts = self.read_unsigned_long_long()
 100             duration = self.read_unsigned_int()
 101             if duration == 0:
 102                 discontinuity_indicator = self.read_unsigned_char()
 103             else:
 104                 discontinuity_indicator = None
 105             fragments.append({
 106                 'first': first,
 107                 'ts': first_ts,
 108                 'duration': duration,
 109                 'discontinuity_indicator': discontinuity_indicator,
 110             })
 111
 112         return {
 113             'fragments': fragments,
 114         }
 115
 116     def read_abst(self):
 117         # version
 118         self.read_unsigned_char()
 119         # flags
 120         self.read(3)
 121
 122         self.read_unsigned_int()  # BootstrapinfoVersion
 123         # Profile,Live,Update,Reserved
 124         self.read(1)
 125         # time scale
 126         self.read_unsigned_int()
 127         # CurrentMediaTime
 128         self.read_unsigned_long_long()
 129         # SmpteTimeCodeOffset
 130         self.read_unsigned_long_long()
 131
 132         self.read_string()  # MovieIdentifier
 133         server_count = self.read_unsigned_char()
 134         # ServerEntryTable
 135         for i in range(server_count):
 136             self.read_string()
 137         quality_count = self.read_unsigned_char()
 138         # QualityEntryTable
 139         for i in range(quality_count):
 140             self.read_string()
 141         # DrmData
 142         self.read_string()
 143         # MetaData
 144         self.read_string()
 145
 146         segments_count = self.read_unsigned_char()
 147         segments = []
 148         for i in range(segments_count):
 149             box_size, box_type, box_data = self.read_box_info()
 150             assert box_type == b'asrt'
 151             segment = FlvReader(box_data).read_asrt()
 152             segments.append(segment)
 153         fragments_run_count = self.read_unsigned_char()
 154         fragments = []
 155         for i in range(fragments_run_count):
 156             box_size, box_type, box_data = self.read_box_info()
 157             assert box_type == b'afrt'
 158             fragments.append(FlvReader(box_data).read_afrt())
 159
 160         return {
 161             'segments': segments,
 162             'fragments': fragments,
 163         }
 164
 165     def read_bootstrap_info(self):
 166         total_size, box_type, box_data = self.read_box_info()
 167         assert box_type == b'abst'
 168         return FlvReader(box_data).read_abst()
 169
 170
 171 def read_bootstrap_info(bootstrap_bytes):
 172     return FlvReader(bootstrap_bytes).read_bootstrap_info()
 173
 174
 175 def build_fragments_list(boot_info):
 176     """ Return a list of (segment, fragment) for each fragment in the video """
 177     res = []
 178     segment_run_table = boot_info['segments'][0]
 179     fragment_run_entry_table = boot_info['fragments'][0]['fragments']
 180     first_frag_number = fragment_run_entry_table[0]['first']
 181     fragments_counter = itertools.count(first_frag_number)
 182     for segment, fragments_count in segment_run_table['segment_run']:
 183         for _ in range(fragments_count):
 184             res.append((segment, next(fragments_counter)))
 185     return res
 186
 187
 188 def write_unsigned_int(stream, val):
 189     stream.write(struct_pack('!I', val))
 190
 191
 192 def write_unsigned_int_24(stream, val):
 193     stream.write(struct_pack('!I', val)[1:])
 194
 195
 196 def write_flv_header(stream):
 197     """Writes the FLV header to stream"""
 198     # FLV header
 199     stream.write(b'FLV\x01')
 200     stream.write(b'\x05')
 201     stream.write(b'\x00\x00\x00\x09')
 202     stream.write(b'\x00\x00\x00\x00')
 203
 204
 205 def write_metadata_tag(stream, metadata):
 206     """Writes optional metadata tag to stream"""
 207     SCRIPT_TAG = b'\x12'
 208     FLV_TAG_HEADER_LEN = 11
 209
 210     if metadata:
 211         stream.write(SCRIPT_TAG)
 212         write_unsigned_int_24(stream, len(metadata))
 213         stream.write(b'\x00\x00\x00\x00\x00\x00\x00')
 214         stream.write(metadata)
 215         write_unsigned_int(stream, FLV_TAG_HEADER_LEN + len(metadata))
 216
 217
 218 def _add_ns(prop):
 219     return '{http://ns.adobe.com/f4m/1.0}%s' % prop
 220
 221
 222 class HttpQuietDownloader(HttpFD):
 223     def to_screen(self, *args, **kargs):
 224         pass
 225
 226
 227 class F4mFD(FileDownloader):
 228     """
 229     A downloader for f4m manifests or AdobeHDS.
 230     """
 231
 232     def _get_unencrypted_media(self, doc):
 233         media = doc.findall(_add_ns('media'))
 234         if not media:
 235             self.report_error('No media found')
 236         for e in (doc.findall(_add_ns('drmAdditionalHeader')) +
 237                   doc.findall(_add_ns('drmAdditionalHeaderSet'))):
 238             # If id attribute is missing it's valid for all media nodes
 239             # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute
 240             if 'id' not in e.attrib:
 241                 self.report_error('Missing ID in f4m DRM')
 242         media = list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and
 243                                       'drmAdditionalHeaderSetId' not in e.attrib,
 244                             media))
 245         if not media:
 246             self.report_error('Unsupported DRM')
 247         return media
 248
 249     def real_download(self, filename, info_dict):
 250         man_url = info_dict['url']
 251         requested_bitrate = info_dict.get('tbr')
 252         self.to_screen('[download] Downloading f4m manifest')
 253         manifest = self.ydl.urlopen(man_url).read()
 254
 255         doc = etree.fromstring(manifest)
 256         formats = [(int(f.attrib.get('bitrate', -1)), f)
 257                    for f in self._get_unencrypted_media(doc)]
 258         if requested_bitrate is None:
 259             # get the best format
 260             formats = sorted(formats, key=lambda f: f[0])
 261             rate, media = formats[-1]
 262         else:
 263             rate, media = list(filter(
 264                 lambda f: int(f[0]) == requested_bitrate, formats))[0]
 265
 266         base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
 267         bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
 268         if bootstrap_node.text is None:
 269             bootstrap_url = compat_urlparse.urljoin(
 270                 base_url, bootstrap_node.attrib['url'])
 271             bootstrap = self.ydl.urlopen(bootstrap_url).read()
 272         else:
 273             bootstrap = base64.b64decode(bootstrap_node.text)
 274         metadata_node = media.find(_add_ns('metadata'))
 275         if metadata_node is not None:
 276             metadata = base64.b64decode(metadata_node.text)
 277         else:
 278             metadata = None
 279         boot_info = read_bootstrap_info(bootstrap)
 280
 281         fragments_list = build_fragments_list(boot_info)
 282         if self.params.get('test', False):
 283             # We only download the first fragment
 284             fragments_list = fragments_list[:1]
 285         total_frags = len(fragments_list)
 286         # For some akamai manifests we'll need to add a query to the fragment url
 287         akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
 288
 289         self.report_destination(filename)
 290         http_dl = HttpQuietDownloader(
 291             self.ydl,
 292             {
 293                 'continuedl': True,
 294                 'quiet': True,
 295                 'noprogress': True,
 296                 'ratelimit': self.params.get('ratelimit', None),
 297                 'test': self.params.get('test', False),
 298             }
 299         )
 300         tmpfilename = self.temp_name(filename)
 301         (dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
 302
 303         write_flv_header(dest_stream)
 304         write_metadata_tag(dest_stream, metadata)
 305
 306         # This dict stores the download progress, it's updated by the progress
 307         # hook
 308         state = {
 309             'status': 'downloading',
 310             'downloaded_bytes': 0,
 311             'frag_index': 0,
 312             'frag_count': total_frags,
 313             'filename': filename,
 314             'tmpfilename': tmpfilename,
 315         }
 316         start = time.time()
 317
 318         def frag_progress_hook(s):
 319             if s['status'] not in ('downloading', 'finished'):
 320                 return
 321
 322             frag_total_bytes = s.get('total_bytes', 0)
 323             if s['status'] == 'finished':
 324                 state['downloaded_bytes'] += frag_total_bytes
 325                 state['frag_index'] += 1
 326
 327             estimated_size = (
 328                 (state['downloaded_bytes'] + frag_total_bytes)
 329                 / (state['frag_index'] + 1) * total_frags)
 330             time_now = time.time()
 331             state['total_bytes_estimate'] = estimated_size
 332             state['elapsed'] = time_now - start
 333
 334             if s['status'] == 'finished':
 335                 progress = self.calc_percent(state['frag_index'], total_frags)
 336             else:
 337                 frag_downloaded_bytes = s['downloaded_bytes']
 338                 frag_progress = self.calc_percent(frag_downloaded_bytes,
 339                                                   frag_total_bytes)
 340                 progress = self.calc_percent(state['frag_index'], total_frags)
 341                 progress += frag_progress / float(total_frags)
 342
 343                 state['eta'] = self.calc_eta(
 344                     start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
 345                 state['speed'] = s.get('speed')
 346             self._hook_progress(state)
 347
 348         http_dl.add_progress_hook(frag_progress_hook)
 349
 350         frags_filenames = []
 351         for (seg_i, frag_i) in fragments_list:
 352             name = 'Seg%d-Frag%d' % (seg_i, frag_i)
 353             url = base_url + name
 354             if akamai_pv:
 355                 url += '?' + akamai_pv.strip(';')
 356             frag_filename = '%s-%s' % (tmpfilename, name)
 357             success = http_dl.download(frag_filename, {'url': url})
 358             if not success:
 359                 return False
 360             with open(frag_filename, 'rb') as down:
 361                 down_data = down.read()
 362                 reader = FlvReader(down_data)
 363                 while True:
 364                     _, box_type, box_data = reader.read_box_info()
 365                     if box_type == b'mdat':
 366                         dest_stream.write(box_data)
 367                         break
 368             frags_filenames.append(frag_filename)
 369
 370         dest_stream.close()
 371
 372         elapsed = time.time() - start
 373         self.try_rename(tmpfilename, filename)
 374         for frag_file in frags_filenames:
 375             os.remove(frag_file)
 376
 377         fsize = os.path.getsize(encodeFilename(filename))
 378         self._hook_progress({
 379             'downloaded_bytes': fsize,
 380             'total_bytes': fsize,
 381             'filename': filename,
 382             'status': 'finished',
 383             'elapsed': elapsed,
 384         })
 385
 386         return True