1 from __future__ import division, unicode_literals
9 from .fragment import FragmentFD
10 from ..compat import (
11 compat_etree_fromstring,
14 compat_urllib_parse_urlparse,
26 class FlvReader(io.BytesIO):
29 The file format is documented in https://www.adobe.com/devnet/f4v.html
32 # Utility functions for reading numbers and strings
33 def read_unsigned_long_long(self):
34 return struct_unpack('!Q', self.read(8))[0]
36 def read_unsigned_int(self):
37 return struct_unpack('!I', self.read(4))[0]
39 def read_unsigned_char(self):
40 return struct_unpack('!B', self.read(1))[0]
42 def read_string(self):
51 def read_box_info(self):
53 Read a box and return the info as a tuple: (box_size, box_type, box_data)
55 real_size = size = self.read_unsigned_int()
56 box_type = self.read(4)
59 real_size = self.read_unsigned_long_long()
61 return real_size, box_type, self.read(real_size - header_end)
65 self.read_unsigned_char()
68 quality_entry_count = self.read_unsigned_char()
70 for i in range(quality_entry_count):
73 segment_run_count = self.read_unsigned_int()
75 for i in range(segment_run_count):
76 first_segment = self.read_unsigned_int()
77 fragments_per_segment = self.read_unsigned_int()
78 segments.append((first_segment, fragments_per_segment))
81 'segment_run': segments,
86 self.read_unsigned_char()
90 self.read_unsigned_int()
92 quality_entry_count = self.read_unsigned_char()
93 # QualitySegmentUrlModifiers
94 for i in range(quality_entry_count):
97 fragments_count = self.read_unsigned_int()
99 for i in range(fragments_count):
100 first = self.read_unsigned_int()
101 first_ts = self.read_unsigned_long_long()
102 duration = self.read_unsigned_int()
104 discontinuity_indicator = self.read_unsigned_char()
106 discontinuity_indicator = None
110 'duration': duration,
111 'discontinuity_indicator': discontinuity_indicator,
115 'fragments': fragments,
120 self.read_unsigned_char()
124 self.read_unsigned_int() # BootstrapinfoVersion
125 # Profile,Live,Update,Reserved
126 flags = self.read_unsigned_char()
127 live = flags & 0x20 != 0
129 self.read_unsigned_int()
131 self.read_unsigned_long_long()
132 # SmpteTimeCodeOffset
133 self.read_unsigned_long_long()
135 self.read_string() # MovieIdentifier
136 server_count = self.read_unsigned_char()
138 for i in range(server_count):
140 quality_count = self.read_unsigned_char()
142 for i in range(quality_count):
149 segments_count = self.read_unsigned_char()
151 for i in range(segments_count):
152 box_size, box_type, box_data = self.read_box_info()
153 assert box_type == b'asrt'
154 segment = FlvReader(box_data).read_asrt()
155 segments.append(segment)
156 fragments_run_count = self.read_unsigned_char()
158 for i in range(fragments_run_count):
159 box_size, box_type, box_data = self.read_box_info()
160 assert box_type == b'afrt'
161 fragments.append(FlvReader(box_data).read_afrt())
164 'segments': segments,
165 'fragments': fragments,
169 def read_bootstrap_info(self):
170 total_size, box_type, box_data = self.read_box_info()
171 assert box_type == b'abst'
172 return FlvReader(box_data).read_abst()
175 def read_bootstrap_info(bootstrap_bytes):
176 return FlvReader(bootstrap_bytes).read_bootstrap_info()
179 def build_fragments_list(boot_info):
180 """ Return a list of (segment, fragment) for each fragment in the video """
182 segment_run_table = boot_info['segments'][0]
183 fragment_run_entry_table = boot_info['fragments'][0]['fragments']
184 first_frag_number = fragment_run_entry_table[0]['first']
185 fragments_counter = itertools.count(first_frag_number)
186 for segment, fragments_count in segment_run_table['segment_run']:
187 for _ in range(fragments_count):
188 res.append((segment, next(fragments_counter)))
190 if boot_info['live']:
196 def write_unsigned_int(stream, val):
197 stream.write(struct_pack('!I', val))
200 def write_unsigned_int_24(stream, val):
201 stream.write(struct_pack('!I', val)[1:])
204 def write_flv_header(stream):
205 """Writes the FLV header to stream"""
207 stream.write(b'FLV\x01')
208 stream.write(b'\x05')
209 stream.write(b'\x00\x00\x00\x09')
210 stream.write(b'\x00\x00\x00\x00')
213 def write_metadata_tag(stream, metadata):
214 """Writes optional metadata tag to stream"""
216 FLV_TAG_HEADER_LEN = 11
219 stream.write(SCRIPT_TAG)
220 write_unsigned_int_24(stream, len(metadata))
221 stream.write(b'\x00\x00\x00\x00\x00\x00\x00')
222 stream.write(metadata)
223 write_unsigned_int(stream, FLV_TAG_HEADER_LEN + len(metadata))
226 def remove_encrypted_media(media):
227 return list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and
228 'drmAdditionalHeaderSetId' not in e.attrib,
233 return '{http://ns.adobe.com/f4m/1.0}%s' % prop
236 class F4mFD(FragmentFD):
238 A downloader for f4m manifests or AdobeHDS.
243 def _get_unencrypted_media(self, doc):
244 media = doc.findall(_add_ns('media'))
246 self.report_error('No media found')
247 for e in (doc.findall(_add_ns('drmAdditionalHeader')) +
248 doc.findall(_add_ns('drmAdditionalHeaderSet'))):
249 # If id attribute is missing it's valid for all media nodes
250 # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute
251 if 'id' not in e.attrib:
252 self.report_error('Missing ID in f4m DRM')
253 media = remove_encrypted_media(media)
255 self.report_error('Unsupported DRM')
258 def _get_bootstrap_from_url(self, bootstrap_url):
259 bootstrap = self.ydl.urlopen(bootstrap_url).read()
260 return read_bootstrap_info(bootstrap)
262 def _update_live_fragments(self, bootstrap_url, latest_fragment):
265 while (not fragments_list) and (retries > 0):
266 boot_info = self._get_bootstrap_from_url(bootstrap_url)
267 fragments_list = build_fragments_list(boot_info)
268 fragments_list = [f for f in fragments_list if f[1] > latest_fragment]
269 if not fragments_list:
270 # Retry after a while
274 if not fragments_list:
275 self.report_error('Failed to update fragments')
277 return fragments_list
279 def _parse_bootstrap_node(self, node, base_url):
280 # Sometimes non empty inline bootstrap info can be specified along
281 # with bootstrap url attribute (e.g. dummy inline bootstrap info
282 # contains whitespace characters in [1]). We will prefer bootstrap
283 # url over inline bootstrap info when present.
284 # 1. http://live-1-1.rutube.ru/stream/1024/HDS/SD/C2NKsS85HQNckgn5HdEmOQ/1454167650/S-s604419906/move/four/dirs/upper/1024-576p.f4m
285 bootstrap_url = node.get('url')
287 bootstrap_url = compat_urlparse.urljoin(
288 base_url, bootstrap_url)
289 boot_info = self._get_bootstrap_from_url(bootstrap_url)
292 bootstrap = base64.b64decode(node.text.encode('ascii'))
293 boot_info = read_bootstrap_info(bootstrap)
294 return boot_info, bootstrap_url
296 def real_download(self, filename, info_dict):
297 man_url = info_dict['url']
298 requested_bitrate = info_dict.get('tbr')
299 self.to_screen('[%s] Downloading f4m manifest' % self.FD_NAME)
300 urlh = self.ydl.urlopen(man_url)
301 man_url = urlh.geturl()
302 # Some manifests may be malformed, e.g. prosiebensat1 generated manifests
303 # (see https://github.com/rg3/youtube-dl/issues/6215#issuecomment-121704244
304 # and https://github.com/rg3/youtube-dl/issues/7823)
305 manifest = fix_xml_ampersands(urlh.read().decode('utf-8', 'ignore')).strip()
307 doc = compat_etree_fromstring(manifest)
308 formats = [(int(f.attrib.get('bitrate', -1)), f)
309 for f in self._get_unencrypted_media(doc)]
310 if requested_bitrate is None:
311 # get the best format
312 formats = sorted(formats, key=lambda f: f[0])
313 rate, media = formats[-1]
315 rate, media = list(filter(
316 lambda f: int(f[0]) == requested_bitrate, formats))[0]
318 base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
319 bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
320 boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, base_url)
321 live = boot_info['live']
322 metadata_node = media.find(_add_ns('metadata'))
323 if metadata_node is not None:
324 metadata = base64.b64decode(metadata_node.text.encode('ascii'))
328 fragments_list = build_fragments_list(boot_info)
329 test = self.params.get('test', False)
331 # We only download the first fragment
332 fragments_list = fragments_list[:1]
333 total_frags = len(fragments_list)
334 # For some akamai manifests we'll need to add a query to the fragment url
335 akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
338 'filename': filename,
339 'total_frags': total_frags,
343 self._prepare_frag_download(ctx)
345 dest_stream = ctx['dest_stream']
347 write_flv_header(dest_stream)
349 write_metadata_tag(dest_stream, metadata)
351 base_url_parsed = compat_urllib_parse_urlparse(base_url)
353 self._start_frag_download(ctx)
356 while fragments_list:
357 seg_i, frag_i = fragments_list.pop(0)
358 name = 'Seg%d-Frag%d' % (seg_i, frag_i)
360 if base_url_parsed.query:
361 query.append(base_url_parsed.query)
363 query.append(akamai_pv.strip(';'))
364 if info_dict.get('extra_param_to_segment_url'):
365 query.append(info_dict['extra_param_to_segment_url'])
366 url_parsed = base_url_parsed._replace(path=base_url_parsed.path + name, query='&'.join(query))
367 frag_filename = '%s-%s' % (ctx['tmpfilename'], name)
369 success = ctx['dl'].download(frag_filename, {'url': url_parsed.geturl()})
372 (down, frag_sanitized) = sanitize_open(frag_filename, 'rb')
373 down_data = down.read()
375 reader = FlvReader(down_data)
377 _, box_type, box_data = reader.read_box_info()
378 if box_type == b'mdat':
379 dest_stream.write(box_data)
382 os.remove(encodeFilename(frag_sanitized))
384 frags_filenames.append(frag_sanitized)
385 except (compat_urllib_error.HTTPError, ) as err:
386 if live and (err.code == 404 or err.code == 410):
387 # We didn't keep up with the live window. Continue
388 # with the next available fragment.
389 msg = 'Fragment %d unavailable' % frag_i
390 self.report_warning(msg)
395 if not fragments_list and not test and live and bootstrap_url:
396 fragments_list = self._update_live_fragments(bootstrap_url, frag_i)
397 total_frags += len(fragments_list)
398 if fragments_list and (fragments_list[0][1] > frag_i + 1):
399 msg = 'Missed %d fragments' % (fragments_list[0][1] - (frag_i + 1))
400 self.report_warning(msg)
402 self._finish_frag_download(ctx)
404 for frag_file in frags_filenames:
405 os.remove(encodeFilename(frag_file))