X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/15b1d10671b48df598afd70e17ba21e9e64ac766..ef2d059988ee523e76933953419ba32637860796:/youtube_dl/downloader/hls.py diff --git a/youtube_dl/downloader/hls.py b/youtube_dl/downloader/hls.py index 954beff..4989abc 100644 --- a/youtube_dl/downloader/hls.py +++ b/youtube_dl/downloader/hls.py @@ -1,103 +1,176 @@ from __future__ import unicode_literals -import os +import os.path import re -import subprocess +import binascii +try: + from Crypto.Cipher import AES + can_decrypt_frag = True +except ImportError: + can_decrypt_frag = False -from .common import FileDownloader -from ..utils import ( +from .fragment import FragmentFD +from .external import FFmpegFD + +from ..compat import ( + compat_urllib_error, compat_urlparse, - compat_urllib_request, - check_executable, + compat_struct_pack, +) +from ..utils import ( encodeFilename, + sanitize_open, + parse_m3u8_attributes, + update_url_query, ) -class HlsFD(FileDownloader): - def real_download(self, filename, info_dict): - url = info_dict['url'] - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - args = [ - '-y', '-i', url, '-f', 'mp4', '-c', 'copy', - '-bsf:a', 'aac_adtstoasc', - encodeFilename(tmpfilename, for_subprocess=True)] - - for program in ['avconv', 'ffmpeg']: - if check_executable(program, ['-version']): - break - else: - self.report_error('m3u8 download detected but ffmpeg or avconv could not be found. Please install one.') - return False - cmd = [program] + args - - retval = subprocess.call(cmd) - if retval == 0: - fsize = os.path.getsize(encodeFilename(tmpfilename)) - self.to_screen('\r[%s] %s bytes' % (cmd[0], fsize)) - self.try_rename(tmpfilename, filename) - self._hook_progress({ - 'downloaded_bytes': fsize, - 'total_bytes': fsize, - 'filename': filename, - 'status': 'finished', - }) - return True - else: - self.to_stderr('\n') - self.report_error('%s exited with code %d' % (program, retval)) - return False - - -class NativeHlsFD(FileDownloader): - """ A more limited implementation that does not require ffmpeg """ +class HlsFD(FragmentFD): + """ A limited implementation that does not require ffmpeg """ + + FD_NAME = 'hlsnative' + + @staticmethod + def can_download(manifest, info_dict): + UNSUPPORTED_FEATURES = ( + r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)', # encrypted streams [1] + r'#EXT-X-BYTERANGE', # playlists composed of byte ranges of media files [2] + + # Live streams heuristic does not always work (e.g. geo restricted to Germany + # http://hls-geo.daserste.de/i/videoportal/Film/c_620000/622873/format,716451,716457,716450,716458,716459,.mp4.csmil/index_4_av.m3u8?null=0) + # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)', # live streams [3] + + # This heuristic also is not correct since segments may not be appended as well. + # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite + # no segments will definitely be appended to the end of the playlist. + # r'#EXT-X-PLAYLIST-TYPE:EVENT', # media segments may be appended to the end of + # # event media playlists [4] + + # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4 + # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2 + # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2 + # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5 + ) + check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES] + check_results.append(can_decrypt_frag or '#EXT-X-KEY:METHOD=AES-128' not in manifest) + check_results.append(not info_dict.get('is_live')) + return all(check_results) def real_download(self, filename, info_dict): - url = info_dict['url'] - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - - self.to_screen( - '[hlsnative] %s: Downloading m3u8 manifest' % info_dict['id']) - data = self.ydl.urlopen(url).read() - s = data.decode('utf-8', 'ignore') - segment_urls = [] + man_url = info_dict['url'] + self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME) + + manifest = self.ydl.urlopen(self._prepare_url(info_dict, man_url)).read() + + s = manifest.decode('utf-8', 'ignore') + + if not self.can_download(s, info_dict): + if info_dict.get('extra_param_to_segment_url'): + self.report_error('pycrypto not found. Please install it.') + return False + self.report_warning( + 'hlsnative has detected features it does not support, ' + 'extraction will be delegated to ffmpeg') + fd = FFmpegFD(self.ydl, self.params) + for ph in self._progress_hooks: + fd.add_progress_hook(ph) + return fd.real_download(filename, info_dict) + + total_frags = 0 for line in s.splitlines(): line = line.strip() if line and not line.startswith('#'): - segment_url = ( - line - if re.match(r'^https?://', line) - else compat_urlparse.urljoin(url, line)) - segment_urls.append(segment_url) - - is_test = self.params.get('test', False) - remaining_bytes = self._TEST_FILE_SIZE if is_test else None - byte_counter = 0 - with open(tmpfilename, 'wb') as outf: - for i, segurl in enumerate(segment_urls): - self.to_screen( - '[hlsnative] %s: Downloading segment %d / %d' % - (info_dict['id'], i + 1, len(segment_urls))) - seg_req = compat_urllib_request.Request(segurl) - if remaining_bytes is not None: - seg_req.add_header('Range', 'bytes=0-%d' % (remaining_bytes - 1)) - - segment = self.ydl.urlopen(seg_req).read() - if remaining_bytes is not None: - segment = segment[:remaining_bytes] - remaining_bytes -= len(segment) - outf.write(segment) - byte_counter += len(segment) - if remaining_bytes is not None and remaining_bytes <= 0: - break - - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': byte_counter, + total_frags += 1 + + ctx = { 'filename': filename, - 'status': 'finished', - }) - self.try_rename(tmpfilename, filename) + 'total_frags': total_frags, + } + + self._prepare_and_start_frag_download(ctx) + + fragment_retries = self.params.get('fragment_retries', 0) + skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True) + test = self.params.get('test', False) + + extra_query = None + extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url') + if extra_param_to_segment_url: + extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url) + i = 0 + media_sequence = 0 + decrypt_info = {'METHOD': 'NONE'} + frags_filenames = [] + for line in s.splitlines(): + line = line.strip() + if line: + if not line.startswith('#'): + frag_url = ( + line + if re.match(r'^https?://', line) + else compat_urlparse.urljoin(man_url, line)) + frag_name = 'Frag%d' % i + frag_filename = '%s-%s' % (ctx['tmpfilename'], frag_name) + if extra_query: + frag_url = update_url_query(frag_url, extra_query) + count = 0 + while count <= fragment_retries: + try: + success = ctx['dl'].download(frag_filename, { + 'url': frag_url, + 'http_headers': info_dict.get('http_headers'), + }) + if not success: + return False + down, frag_sanitized = sanitize_open(frag_filename, 'rb') + frag_content = down.read() + down.close() + break + except compat_urllib_error.HTTPError as err: + # Unavailable (possibly temporary) fragments may be served. + # First we try to retry then either skip or abort. + # See https://github.com/rg3/youtube-dl/issues/10165, + # https://github.com/rg3/youtube-dl/issues/10448). + count += 1 + if count <= fragment_retries: + self.report_retry_fragment(err, frag_name, count, fragment_retries) + if count > fragment_retries: + if skip_unavailable_fragments: + i += 1 + media_sequence += 1 + self.report_skip_fragment(frag_name) + continue + self.report_error( + 'giving up after %s fragment retries' % fragment_retries) + return False + if decrypt_info['METHOD'] == 'AES-128': + iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence) + frag_content = AES.new( + decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content) + ctx['dest_stream'].write(frag_content) + frags_filenames.append(frag_sanitized) + # We only download the first fragment during the test + if test: + break + i += 1 + media_sequence += 1 + elif line.startswith('#EXT-X-KEY'): + decrypt_info = parse_m3u8_attributes(line[11:]) + if decrypt_info['METHOD'] == 'AES-128': + if 'IV' in decrypt_info: + decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32)) + if not re.match(r'^https?://', decrypt_info['URI']): + decrypt_info['URI'] = compat_urlparse.urljoin( + man_url, decrypt_info['URI']) + if extra_query: + decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query) + decrypt_info['KEY'] = self.ydl.urlopen(decrypt_info['URI']).read() + elif line.startswith('#EXT-X-MEDIA-SEQUENCE'): + media_sequence = int(line[22:]) + + self._finish_frag_download(ctx) + + for frag_file in frags_filenames: + os.remove(encodeFilename(frag_file)) + return True