Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/downloader/f4m.py

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