1 from __future__ import division, unicode_literals
8 import xml.etree.ElementTree as etree
10 from .fragment import FragmentFD
11 from ..compat import (
24 class FlvReader(io.BytesIO):
27 The file format is documented in https://www.adobe.com/devnet/f4v.html
30 # Utility functions for reading numbers and strings
31 def read_unsigned_long_long(self):
32 return struct_unpack('!Q', self.read(8))[0]
34 def read_unsigned_int(self):
35 return struct_unpack('!I', self.read(4))[0]
37 def read_unsigned_char(self):
38 return struct_unpack('!B', self.read(1))[0]
40 def read_string(self):
49 def read_box_info(self):
51 Read a box and return the info as a tuple: (box_size, box_type, box_data)
53 real_size = size = self.read_unsigned_int()
54 box_type = self.read(4)
57 real_size = self.read_unsigned_long_long()
59 return real_size, box_type, self.read(real_size - header_end)
63 self.read_unsigned_char()
66 quality_entry_count = self.read_unsigned_char()
68 for i in range(quality_entry_count):
71 segment_run_count = self.read_unsigned_int()
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))
79 'segment_run': segments,
84 self.read_unsigned_char()
88 self.read_unsigned_int()
90 quality_entry_count = self.read_unsigned_char()
91 # QualitySegmentUrlModifiers
92 for i in range(quality_entry_count):
95 fragments_count = self.read_unsigned_int()
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()
102 discontinuity_indicator = self.read_unsigned_char()
104 discontinuity_indicator = None
108 'duration': duration,
109 'discontinuity_indicator': discontinuity_indicator,
113 'fragments': fragments,
118 self.read_unsigned_char()
122 self.read_unsigned_int() # BootstrapinfoVersion
123 # Profile,Live,Update,Reserved
124 flags = self.read_unsigned_char()
125 live = flags & 0x20 != 0
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,
167 def read_bootstrap_info(self):
168 total_size, box_type, box_data = self.read_box_info()
169 assert box_type == b'abst'
170 return FlvReader(box_data).read_abst()
173 def read_bootstrap_info(bootstrap_bytes):
174 return FlvReader(bootstrap_bytes).read_bootstrap_info()
177 def build_fragments_list(boot_info):
178 """ Return a list of (segment, fragment) for each fragment in the video """
180 segment_run_table = boot_info['segments'][0]
181 fragment_run_entry_table = boot_info['fragments'][0]['fragments']
182 first_frag_number = fragment_run_entry_table[0]['first']
183 fragments_counter = itertools.count(first_frag_number)
184 for segment, fragments_count in segment_run_table['segment_run']:
185 for _ in range(fragments_count):
186 res.append((segment, next(fragments_counter)))
188 if boot_info['live']:
194 def write_unsigned_int(stream, val):
195 stream.write(struct_pack('!I', val))
198 def write_unsigned_int_24(stream, val):
199 stream.write(struct_pack('!I', val)[1:])
202 def write_flv_header(stream):
203 """Writes the FLV header to stream"""
205 stream.write(b'FLV\x01')
206 stream.write(b'\x05')
207 stream.write(b'\x00\x00\x00\x09')
208 stream.write(b'\x00\x00\x00\x00')
211 def write_metadata_tag(stream, metadata):
212 """Writes optional metadata tag to stream"""
214 FLV_TAG_HEADER_LEN = 11
217 stream.write(SCRIPT_TAG)
218 write_unsigned_int_24(stream, len(metadata))
219 stream.write(b'\x00\x00\x00\x00\x00\x00\x00')
220 stream.write(metadata)
221 write_unsigned_int(stream, FLV_TAG_HEADER_LEN + len(metadata))
225 return '{http://ns.adobe.com/f4m/1.0}%s' % prop
228 class F4mFD(FragmentFD):
230 A downloader for f4m manifests or AdobeHDS.
235 def _get_unencrypted_media(self, doc):
236 media = doc.findall(_add_ns('media'))
238 self.report_error('No media found')
239 for e in (doc.findall(_add_ns('drmAdditionalHeader')) +
240 doc.findall(_add_ns('drmAdditionalHeaderSet'))):
241 # If id attribute is missing it's valid for all media nodes
242 # without drmAdditionalHeaderId or drmAdditionalHeaderSetId attribute
243 if 'id' not in e.attrib:
244 self.report_error('Missing ID in f4m DRM')
245 media = list(filter(lambda e: 'drmAdditionalHeaderId' not in e.attrib and
246 'drmAdditionalHeaderSetId' not in e.attrib,
249 self.report_error('Unsupported DRM')
252 def _get_bootstrap_from_url(self, bootstrap_url):
253 bootstrap = self.ydl.urlopen(bootstrap_url).read()
254 return read_bootstrap_info(bootstrap)
256 def _update_live_fragments(self, bootstrap_url, latest_fragment):
259 while (not fragments_list) and (retries > 0):
260 boot_info = self._get_bootstrap_from_url(bootstrap_url)
261 fragments_list = build_fragments_list(boot_info)
262 fragments_list = [f for f in fragments_list if f[1] > latest_fragment]
263 if not fragments_list:
264 # Retry after a while
268 if not fragments_list:
269 self.report_error('Failed to update fragments')
271 return fragments_list
273 def _parse_bootstrap_node(self, node, base_url):
274 if node.text is None:
275 bootstrap_url = compat_urlparse.urljoin(
276 base_url, node.attrib['url'])
277 boot_info = self._get_bootstrap_from_url(bootstrap_url)
280 bootstrap = base64.b64decode(node.text.encode('ascii'))
281 boot_info = read_bootstrap_info(bootstrap)
282 return (boot_info, bootstrap_url)
284 def real_download(self, filename, info_dict):
285 man_url = info_dict['url']
286 requested_bitrate = info_dict.get('tbr')
287 self.to_screen('[%s] Downloading f4m manifest' % self.FD_NAME)
288 manifest = self.ydl.urlopen(man_url).read()
290 doc = etree.fromstring(manifest)
291 formats = [(int(f.attrib.get('bitrate', -1)), f)
292 for f in self._get_unencrypted_media(doc)]
293 if requested_bitrate is None:
294 # get the best format
295 formats = sorted(formats, key=lambda f: f[0])
296 rate, media = formats[-1]
298 rate, media = list(filter(
299 lambda f: int(f[0]) == requested_bitrate, formats))[0]
301 base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
302 bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
303 boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, base_url)
304 live = boot_info['live']
305 metadata_node = media.find(_add_ns('metadata'))
306 if metadata_node is not None:
307 metadata = base64.b64decode(metadata_node.text.encode('ascii'))
311 fragments_list = build_fragments_list(boot_info)
312 if self.params.get('test', False):
313 # We only download the first fragment
314 fragments_list = fragments_list[:1]
315 total_frags = len(fragments_list)
316 # For some akamai manifests we'll need to add a query to the fragment url
317 akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
320 'filename': filename,
321 'total_frags': total_frags,
324 self._prepare_frag_download(ctx)
326 dest_stream = ctx['dest_stream']
328 write_flv_header(dest_stream)
330 write_metadata_tag(dest_stream, metadata)
332 self._start_frag_download(ctx)
335 while fragments_list:
336 seg_i, frag_i = fragments_list.pop(0)
337 name = 'Seg%d-Frag%d' % (seg_i, frag_i)
338 url = base_url + name
340 url += '?' + akamai_pv.strip(';')
341 if info_dict.get('extra_param_to_segment_url'):
342 url += info_dict.get('extra_param_to_segment_url')
343 frag_filename = '%s-%s' % (ctx['tmpfilename'], name)
345 success = ctx['dl'].download(frag_filename, {'url': url})
348 (down, frag_sanitized) = sanitize_open(frag_filename, 'rb')
349 down_data = down.read()
351 reader = FlvReader(down_data)
353 _, box_type, box_data = reader.read_box_info()
354 if box_type == b'mdat':
355 dest_stream.write(box_data)
358 os.remove(encodeFilename(frag_sanitized))
360 frags_filenames.append(frag_sanitized)
361 except (compat_urllib_error.HTTPError, ) as err:
362 if live and (err.code == 404 or err.code == 410):
363 # We didn't keep up with the live window. Continue
364 # with the next available fragment.
365 msg = 'Fragment %d unavailable' % frag_i
366 self.report_warning(msg)
371 if not fragments_list and live and bootstrap_url:
372 fragments_list = self._update_live_fragments(bootstrap_url, frag_i)
373 total_frags += len(fragments_list)
374 if fragments_list and (fragments_list[0][1] > frag_i + 1):
375 msg = 'Missed %d fragments' % (fragments_list[0][1] - (frag_i + 1))
376 self.report_warning(msg)
378 self._finish_frag_download(ctx)
380 for frag_file in frags_filenames:
381 os.remove(encodeFilename(frag_file))