]> git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/downloader/f4m.py
[vk] Extend _VALID_URL to handle biqle.ru (Closes #6179)
[youtube-dl] / 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     compat_urllib_error,
15 )
16 from ..utils import (
17     struct_pack,
18     struct_unpack,
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         flags = self.read_unsigned_char()
126         live = flags & 0x20 != 0
127         # time scale
128         self.read_unsigned_int()
129         # CurrentMediaTime
130         self.read_unsigned_long_long()
131         # SmpteTimeCodeOffset
132         self.read_unsigned_long_long()
133
134         self.read_string()  # MovieIdentifier
135         server_count = self.read_unsigned_char()
136         # ServerEntryTable
137         for i in range(server_count):
138             self.read_string()
139         quality_count = self.read_unsigned_char()
140         # QualityEntryTable
141         for i in range(quality_count):
142             self.read_string()
143         # DrmData
144         self.read_string()
145         # MetaData
146         self.read_string()
147
148         segments_count = self.read_unsigned_char()
149         segments = []
150         for i in range(segments_count):
151             box_size, box_type, box_data = self.read_box_info()
152             assert box_type == b'asrt'
153             segment = FlvReader(box_data).read_asrt()
154             segments.append(segment)
155         fragments_run_count = self.read_unsigned_char()
156         fragments = []
157         for i in range(fragments_run_count):
158             box_size, box_type, box_data = self.read_box_info()
159             assert box_type == b'afrt'
160             fragments.append(FlvReader(box_data).read_afrt())
161
162         return {
163             'segments': segments,
164             'fragments': fragments,
165             'live': live,
166         }
167
168     def read_bootstrap_info(self):
169         total_size, box_type, box_data = self.read_box_info()
170         assert box_type == b'abst'
171         return FlvReader(box_data).read_abst()
172
173
174 def read_bootstrap_info(bootstrap_bytes):
175     return FlvReader(bootstrap_bytes).read_bootstrap_info()
176
177
178 def build_fragments_list(boot_info):
179     """ Return a list of (segment, fragment) for each fragment in the video """
180     res = []
181     segment_run_table = boot_info['segments'][0]
182     fragment_run_entry_table = boot_info['fragments'][0]['fragments']
183     first_frag_number = fragment_run_entry_table[0]['first']
184     fragments_counter = itertools.count(first_frag_number)
185     for segment, fragments_count in segment_run_table['segment_run']:
186         for _ in range(fragments_count):
187             res.append((segment, next(fragments_counter)))
188
189     if boot_info['live']:
190         res = res[-2:]
191
192     return res
193
194
195 def write_unsigned_int(stream, val):
196     stream.write(struct_pack('!I', val))
197
198
199 def write_unsigned_int_24(stream, val):
200     stream.write(struct_pack('!I', val)[1:])
201
202
203 def write_flv_header(stream):
204     """Writes the FLV header to stream"""
205     # FLV header
206     stream.write(b'FLV\x01')
207     stream.write(b'\x05')
208     stream.write(b'\x00\x00\x00\x09')
209     stream.write(b'\x00\x00\x00\x00')
210
211
212 def write_metadata_tag(stream, metadata):
213     """Writes optional metadata tag to stream"""
214     SCRIPT_TAG = b'\x12'
215     FLV_TAG_HEADER_LEN = 11
216
217     if metadata:
218         stream.write(SCRIPT_TAG)
219         write_unsigned_int_24(stream, len(metadata))
220         stream.write(b'\x00\x00\x00\x00\x00\x00\x00')
221         stream.write(metadata)
222         write_unsigned_int(stream, FLV_TAG_HEADER_LEN + len(metadata))
223
224
225 def _add_ns(prop):
226     return '{http://ns.adobe.com/f4m/1.0}%s' % prop
227
228
229 class HttpQuietDownloader(HttpFD):
230     def to_screen(self, *args, **kargs):
231         pass
232
233
234 class F4mFD(FileDownloader):
235     """
236     A downloader for f4m manifests or AdobeHDS.
237     """
238
239     def _get_unencrypted_media(self, doc):
240         media = doc.findall(_add_ns('media'))
241         if not media:
242             self.report_error('No media found')
243         for e in (doc.findall(_add_ns('drmAdditionalHeader')) +
244                   doc.findall(_add_ns('drmAdditionalHeaderSet'))):
245             # If id attribute is missing it's valid for all media nodes
246             # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute
247             if 'id' not in e.attrib:
248                 self.report_error('Missing ID in f4m DRM')
249         media = list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and
250                                       'drmAdditionalHeaderSetId' not in e.attrib,
251                             media))
252         if not media:
253             self.report_error('Unsupported DRM')
254         return media
255
256     def _get_bootstrap_from_url(self, bootstrap_url):
257         bootstrap = self.ydl.urlopen(bootstrap_url).read()
258         return read_bootstrap_info(bootstrap)
259
260     def _update_live_fragments(self, bootstrap_url, latest_fragment):
261         fragments_list = []
262         retries = 30
263         while (not fragments_list) and (retries > 0):
264             boot_info = self._get_bootstrap_from_url(bootstrap_url)
265             fragments_list = build_fragments_list(boot_info)
266             fragments_list = [f for f in fragments_list if f[1] > latest_fragment]
267             if not fragments_list:
268                 # Retry after a while
269                 time.sleep(5.0)
270                 retries -= 1
271
272         if not fragments_list:
273             self.report_error('Failed to update fragments')
274
275         return fragments_list
276
277     def _parse_bootstrap_node(self, node, base_url):
278         if node.text is None:
279             bootstrap_url = compat_urlparse.urljoin(
280                 base_url, node.attrib['url'])
281             boot_info = self._get_bootstrap_from_url(bootstrap_url)
282         else:
283             bootstrap_url = None
284             bootstrap = base64.b64decode(node.text.encode('ascii'))
285             boot_info = read_bootstrap_info(bootstrap)
286         return (boot_info, bootstrap_url)
287
288     def real_download(self, filename, info_dict):
289         man_url = info_dict['url']
290         requested_bitrate = info_dict.get('tbr')
291         self.to_screen('[download] Downloading f4m manifest')
292         manifest = self.ydl.urlopen(man_url).read()
293
294         doc = etree.fromstring(manifest)
295         formats = [(int(f.attrib.get('bitrate', -1)), f)
296                    for f in self._get_unencrypted_media(doc)]
297         if requested_bitrate is None:
298             # get the best format
299             formats = sorted(formats, key=lambda f: f[0])
300             rate, media = formats[-1]
301         else:
302             rate, media = list(filter(
303                 lambda f: int(f[0]) == requested_bitrate, formats))[0]
304
305         base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
306         bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
307         boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, base_url)
308         live = boot_info['live']
309         metadata_node = media.find(_add_ns('metadata'))
310         if metadata_node is not None:
311             metadata = base64.b64decode(metadata_node.text.encode('ascii'))
312         else:
313             metadata = None
314
315         fragments_list = build_fragments_list(boot_info)
316         if self.params.get('test', False):
317             # We only download the first fragment
318             fragments_list = fragments_list[:1]
319         total_frags = len(fragments_list)
320         # For some akamai manifests we'll need to add a query to the fragment url
321         akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
322
323         self.report_destination(filename)
324         http_dl = HttpQuietDownloader(
325             self.ydl,
326             {
327                 'continuedl': True,
328                 'quiet': True,
329                 'noprogress': True,
330                 'ratelimit': self.params.get('ratelimit', None),
331                 'test': self.params.get('test', False),
332             }
333         )
334         tmpfilename = self.temp_name(filename)
335         (dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
336
337         write_flv_header(dest_stream)
338         if not live:
339             write_metadata_tag(dest_stream, metadata)
340
341         # This dict stores the download progress, it's updated by the progress
342         # hook
343         state = {
344             'status': 'downloading',
345             'downloaded_bytes': 0,
346             'frag_index': 0,
347             'frag_count': total_frags,
348             'filename': filename,
349             'tmpfilename': tmpfilename,
350         }
351         start = time.time()
352
353         def frag_progress_hook(s):
354             if s['status'] not in ('downloading', 'finished'):
355                 return
356
357             frag_total_bytes = s.get('total_bytes', 0)
358             if s['status'] == 'finished':
359                 state['downloaded_bytes'] += frag_total_bytes
360                 state['frag_index'] += 1
361
362             estimated_size = (
363                 (state['downloaded_bytes'] + frag_total_bytes) /
364                 (state['frag_index'] + 1) * total_frags)
365             time_now = time.time()
366             state['total_bytes_estimate'] = estimated_size
367             state['elapsed'] = time_now - start
368
369             if s['status'] == 'finished':
370                 progress = self.calc_percent(state['frag_index'], total_frags)
371             else:
372                 frag_downloaded_bytes = s['downloaded_bytes']
373                 frag_progress = self.calc_percent(frag_downloaded_bytes,
374                                                   frag_total_bytes)
375                 progress = self.calc_percent(state['frag_index'], total_frags)
376                 progress += frag_progress / float(total_frags)
377
378                 state['eta'] = self.calc_eta(
379                     start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
380                 state['speed'] = s.get('speed')
381             self._hook_progress(state)
382
383         http_dl.add_progress_hook(frag_progress_hook)
384
385         frags_filenames = []
386         while fragments_list:
387             seg_i, frag_i = fragments_list.pop(0)
388             name = 'Seg%d-Frag%d' % (seg_i, frag_i)
389             url = base_url + name
390             if akamai_pv:
391                 url += '?' + akamai_pv.strip(';')
392             if info_dict.get('extra_param_to_segment_url'):
393                 url += info_dict.get('extra_param_to_segment_url')
394             frag_filename = '%s-%s' % (tmpfilename, name)
395             try:
396                 success = http_dl.download(frag_filename, {'url': url})
397                 if not success:
398                     return False
399                 with open(frag_filename, 'rb') as down:
400                     down_data = down.read()
401                     reader = FlvReader(down_data)
402                     while True:
403                         _, box_type, box_data = reader.read_box_info()
404                         if box_type == b'mdat':
405                             dest_stream.write(box_data)
406                             break
407                 if live:
408                     os.remove(frag_filename)
409                 else:
410                     frags_filenames.append(frag_filename)
411             except (compat_urllib_error.HTTPError, ) as err:
412                 if live and (err.code == 404 or err.code == 410):
413                     # We didn't keep up with the live window. Continue
414                     # with the next available fragment.
415                     msg = 'Fragment %d unavailable' % frag_i
416                     self.report_warning(msg)
417                     fragments_list = []
418                 else:
419                     raise
420
421             if not fragments_list and live and bootstrap_url:
422                 fragments_list = self._update_live_fragments(bootstrap_url, frag_i)
423                 total_frags += len(fragments_list)
424                 if fragments_list and (fragments_list[0][1] > frag_i + 1):
425                     msg = 'Missed %d fragments' % (fragments_list[0][1] - (frag_i + 1))
426                     self.report_warning(msg)
427
428         dest_stream.close()
429
430         elapsed = time.time() - start
431         self.try_rename(tmpfilename, filename)
432         for frag_file in frags_filenames:
433             os.remove(frag_file)
434
435         fsize = os.path.getsize(encodeFilename(filename))
436         self._hook_progress({
437             'downloaded_bytes': fsize,
438             'total_bytes': fsize,
439             'filename': filename,
440             'status': 'finished',
441             'elapsed': elapsed,
442         })
443
444         return True