depau/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-04 16:41:19 +00:00
Code Issues Releases Wiki Activity

[extractor/common] Use baseURL from f4m manifest for recursive manifest extraction

Browse source
This commit is contained in:
Sergey M․ 2015-11-21 18:01:39 +06:00
parent f52183a878
commit 019839faaa
1 changed files with 6 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
youtube_dl/extractor/common.py
View file

@ -891,6 +891,11 @@ class InfoExtractor(object):
if not media_nodes: if not media_nodes:
manifest_version = '2.0' manifest_version = '2.0'
media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media') media_nodes = manifest.findall('{http://ns.adobe.com/f4m/2.0}media')
base_url = xpath_text(
manifest, ['{http://ns.adobe.com/f4m/1.0}baseURL', '{http://ns.adobe.com/f4m/2.0}baseURL'],
'base URL', default=None)
if base_url:
base_url = base_url.strip()
for i, media_el in enumerate(media_nodes): for i, media_el in enumerate(media_nodes):
if manifest_version == '2.0': if manifest_version == '2.0':
media_url = media_el.attrib.get('href') or media_el.attrib.get('url') media_url = media_el.attrib.get('href') or media_el.attrib.get('url')
@ -898,7 +903,7 @@ class InfoExtractor(object):
continue continue
manifest_url = ( manifest_url = (
media_url if media_url.startswith('http://') or media_url.startswith('https://') media_url if media_url.startswith('http://') or media_url.startswith('https://')
else ('/'.join(manifest_url.split('/')[:-1]) + '/' + media_url)) else ((base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url))
# If media_url is itself a f4m manifest do the recursive extraction # If media_url is itself a f4m manifest do the recursive extraction
# since bitrates in parent manifest (this one) and media_url manifest # since bitrates in parent manifest (this one) and media_url manifest
# may differ leading to inability to resolve the format by requested # may differ leading to inability to resolve the format by requested