1 from __future__ import division, unicode_literals
8 import xml.etree.ElementTree as etree
10 from .common import FileDownloader
11 from .http import HttpFD
12 from ..compat import (
25 class FlvReader(io.BytesIO):
28 The file format is documented in https://www.adobe.com/devnet/f4v.html
31 # Utility functions for reading numbers and strings
32 def read_unsigned_long_long(self):
33 return struct_unpack('!Q', self.read(8))[0]
35 def read_unsigned_int(self):
36 return struct_unpack('!I', self.read(4))[0]
38 def read_unsigned_char(self):
39 return struct_unpack('!B', self.read(1))[0]
41 def read_string(self):
50 def read_box_info(self):
52 Read a box and return the info as a tuple: (box_size, box_type, box_data)
54 real_size = size = self.read_unsigned_int()
55 box_type = self.read(4)
58 real_size = self.read_unsigned_long_long()
60 return real_size, box_type, self.read(real_size - header_end)
64 self.read_unsigned_char()
67 quality_entry_count = self.read_unsigned_char()
69 for i in range(quality_entry_count):
72 segment_run_count = self.read_unsigned_int()
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))
80 'segment_run': segments,
85 self.read_unsigned_char()
89 self.read_unsigned_int()
91 quality_entry_count = self.read_unsigned_char()
92 # QualitySegmentUrlModifiers
93 for i in range(quality_entry_count):
96 fragments_count = self.read_unsigned_int()
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()
103 discontinuity_indicator = self.read_unsigned_char()
105 discontinuity_indicator = None
109 'duration': duration,
110 'discontinuity_indicator': discontinuity_indicator,
114 'fragments': fragments,
119 self.read_unsigned_char()
123 self.read_unsigned_int() # BootstrapinfoVersion
124 # Profile,Live,Update,Reserved
127 self.read_unsigned_int()
129 self.read_unsigned_long_long()
130 # SmpteTimeCodeOffset
131 self.read_unsigned_long_long()
133 self.read_string() # MovieIdentifier
134 server_count = self.read_unsigned_char()
136 for i in range(server_count):
138 quality_count = self.read_unsigned_char()
140 for i in range(quality_count):
147 segments_count = self.read_unsigned_char()
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()
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())
162 'segments': segments,
163 'fragments': fragments,
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()
172 def read_bootstrap_info(bootstrap_bytes):
173 return FlvReader(bootstrap_bytes).read_bootstrap_info()
176 def build_fragments_list(boot_info):
177 """ Return a list of (segment, fragment) for each fragment in the video """
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)))
189 def write_unsigned_int(stream, val):
190 stream.write(struct_pack('!I', val))
193 def write_unsigned_int_24(stream, val):
194 stream.write(struct_pack('!I', val)[1:])
197 def write_flv_header(stream):
198 """Writes the FLV header to stream"""
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')
206 def write_metadata_tag(stream, metadata):
207 """Writes optional metadata tag to stream"""
209 FLV_TAG_HEADER_LEN = 11
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))
220 return '{http://ns.adobe.com/f4m/1.0}%s' % prop
223 class HttpQuietDownloader(HttpFD):
224 def to_screen(self, *args, **kargs):
228 class F4mFD(FileDownloader):
230 A downloader for f4m manifests or AdobeHDS.
233 def _get_unencrypted_media(self, doc):
234 media = doc.findall(_add_ns('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,
247 self.report_error('Unsupported DRM')
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()
256 doc = etree.fromstring(manifest)
257 formats = [(int(f.attrib.get('bitrate', -1)), f)
258 for f in self._get_unencrypted_media(doc)]
259 if requested_bitrate is None:
260 # get the best format
261 formats = sorted(formats, key=lambda f: f[0])
262 rate, media = formats[-1]
264 rate, media = list(filter(
265 lambda f: int(f[0]) == requested_bitrate, formats))[0]
267 base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
268 bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
269 if bootstrap_node.text is None:
270 bootstrap_url = compat_urlparse.urljoin(
271 base_url, bootstrap_node.attrib['url'])
272 bootstrap = self.ydl.urlopen(bootstrap_url).read()
274 bootstrap = base64.b64decode(bootstrap_node.text)
275 metadata_node = media.find(_add_ns('metadata'))
276 if metadata_node is not None:
277 metadata = base64.b64decode(metadata_node.text)
280 boot_info = read_bootstrap_info(bootstrap)
282 fragments_list = build_fragments_list(boot_info)
283 if self.params.get('test', False):
284 # We only download the first fragment
285 fragments_list = fragments_list[:1]
286 total_frags = len(fragments_list)
287 # For some akamai manifests we'll need to add a query to the fragment url
288 akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
290 self.report_destination(filename)
291 http_dl = HttpQuietDownloader(
297 'ratelimit': self.params.get('ratelimit', None),
298 'test': self.params.get('test', False),
301 tmpfilename = self.temp_name(filename)
302 (dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
304 write_flv_header(dest_stream)
305 write_metadata_tag(dest_stream, metadata)
307 # This dict stores the download progress, it's updated by the progress
310 'status': 'downloading',
311 'downloaded_bytes': 0,
313 'frag_count': total_frags,
314 'filename': filename,
315 'tmpfilename': tmpfilename,
319 def frag_progress_hook(s):
320 if s['status'] not in ('downloading', 'finished'):
323 frag_total_bytes = s.get('total_bytes', 0)
324 if s['status'] == 'finished':
325 state['downloaded_bytes'] += frag_total_bytes
326 state['frag_index'] += 1
329 (state['downloaded_bytes'] + frag_total_bytes)
330 / (state['frag_index'] + 1) * total_frags)
331 time_now = time.time()
332 state['total_bytes_estimate'] = estimated_size
333 state['elapsed'] = time_now - start
335 if s['status'] == 'finished':
336 progress = self.calc_percent(state['frag_index'], total_frags)
337 byte_counter = state['downloaded_bytes']
339 frag_downloaded_bytes = s['downloaded_bytes']
340 byte_counter = state['downloaded_bytes'] + frag_downloaded_bytes
341 frag_progress = self.calc_percent(frag_downloaded_bytes,
343 progress = self.calc_percent(state['frag_index'], total_frags)
344 progress += frag_progress / float(total_frags)
346 state['eta'] = self.calc_eta(
347 start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
348 state['speed'] = s.get('speed')
349 self._hook_progress(state)
351 http_dl.add_progress_hook(frag_progress_hook)
354 for (seg_i, frag_i) in fragments_list:
355 name = 'Seg%d-Frag%d' % (seg_i, frag_i)
356 url = base_url + name
358 url += '?' + akamai_pv.strip(';')
359 frag_filename = '%s-%s' % (tmpfilename, name)
360 success = http_dl.download(frag_filename, {'url': url})
363 with open(frag_filename, 'rb') as down:
364 down_data = down.read()
365 reader = FlvReader(down_data)
367 _, box_type, box_data = reader.read_box_info()
368 if box_type == b'mdat':
369 dest_stream.write(box_data)
371 frags_filenames.append(frag_filename)
375 elapsed = time.time() - start
376 self.try_rename(tmpfilename, filename)
377 for frag_file in frags_filenames:
380 fsize = os.path.getsize(encodeFilename(filename))
381 self._hook_progress({
382 'downloaded_bytes': fsize,
383 'total_bytes': fsize,
384 'filename': filename,
385 'status': 'finished',