X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=sidebyside;f=youtube_dl%2Fdownloader%2Fdash.py;h=e2ddc369e03d03e46bde671c751018f290bf6e59;hb=f264c62334fdd31a7620b4fdefb822e1bae6bd77;hp=cd84e0b07200d5d93304757eac78f8e51d716845;hpb=93dfcb9357b400b4d7e353d0a9db0e0194135b19;p=youtube-dl diff --git a/youtube_dl/downloader/dash.py b/youtube_dl/downloader/dash.py index cd84e0b07..e2ddc369e 100644 --- a/youtube_dl/downloader/dash.py +++ b/youtube_dl/downloader/dash.py @@ -1,52 +1,84 @@ from __future__ import unicode_literals -import re +import os -from .common import FileDownloader -from ..compat import compat_urllib_request +from .fragment import FragmentFD +from ..compat import compat_urllib_error +from ..utils import ( + sanitize_open, + encodeFilename, +) -class DashSegmentsFD(FileDownloader): +class DashSegmentsFD(FragmentFD): """ Download segments in a DASH manifest """ + + FD_NAME = 'dashsegments' + def real_download(self, filename, info_dict): - self.report_destination(filename) - tmpfilename = self.temp_name(filename) - base_url = info_dict['url'] - segment_urls = info_dict['segment_urls'] - - byte_counter = 0 - - def append_url_to_file(outf, target_url, target_name): - self.to_screen('[DashSegments] %s: Downloading %s' % (info_dict['id'], target_name)) - req = compat_urllib_request.Request(target_url) - data = self.ydl.urlopen(req).read() - outf.write(data) - return len(data) - - def combine_url(base_url, target_url): - if re.match(r'^https?://', target_url): - return target_url - return '%s/%s' % (base_url, target_url) - - with open(tmpfilename, 'wb') as outf: - append_url_to_file( - outf, combine_url(base_url, info_dict['initialization_url']), - 'initialization segment') - for i, segment_url in enumerate(segment_urls): - segment_len = append_url_to_file( - outf, combine_url(base_url, segment_url), - 'segment %d / %d' % (i + 1, len(segment_urls))) - byte_counter += segment_len - - self.try_rename(tmpfilename, filename) - - self._hook_progress({ - 'downloaded_bytes': byte_counter, - 'total_bytes': byte_counter, + segments = info_dict['fragments'][:1] if self.params.get( + 'test', False) else info_dict['fragments'] + + ctx = { 'filename': filename, - 'status': 'finished', - }) + 'total_frags': len(segments), + } + + self._prepare_and_start_frag_download(ctx) + + segments_filenames = [] + + fragment_retries = self.params.get('fragment_retries', 0) + skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True) + + def process_segment(segment, tmp_filename, num): + segment_url = segment['url'] + segment_name = 'Frag%d' % num + target_filename = '%s-%s' % (tmp_filename, segment_name) + # In DASH, the first segment contains necessary headers to + # generate a valid MP4 file, so always abort for the first segment + fatal = num == 0 or not skip_unavailable_fragments + count = 0 + while count <= fragment_retries: + try: + success = ctx['dl'].download(target_filename, { + 'url': segment_url, + 'http_headers': info_dict.get('http_headers'), + }) + if not success: + return False + down, target_sanitized = sanitize_open(target_filename, 'rb') + ctx['dest_stream'].write(down.read()) + down.close() + segments_filenames.append(target_sanitized) + break + except compat_urllib_error.HTTPError as err: + # YouTube may often return 404 HTTP error for a fragment causing the + # whole download to fail. However if the same fragment is immediately + # retried with the same request data this usually succeeds (1-2 attemps + # is usually enough) thus allowing to download the whole file successfully. + # To be future-proof we will retry all fragments that fail with any + # HTTP error. + count += 1 + if count <= fragment_retries: + self.report_retry_fragment(err, segment_name, count, fragment_retries) + if count > fragment_retries: + if not fatal: + self.report_skip_fragment(segment_name) + return True + self.report_error('giving up after %s fragment retries' % fragment_retries) + return False + return True + + for i, segment in enumerate(segments): + if not process_segment(segment, ctx['tmpfilename'], i): + return False + + self._finish_frag_download(ctx) + + for segment_file in segments_filenames: + os.remove(encodeFilename(segment_file)) return True