X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyoutube.py;h=28355bf4607b76c9b75da617803ff58d248e4b82;hb=75d572e9fb8d3e26e4ab45e65cd5e23c6b1c6915;hp=1c2420a331481eed04ceb5317984b6d59b6dbf81;hpb=46b4070f3f8cbe2b66baadb2927eba03fbdd1129;p=youtube-dl.git diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 1c2420a33..28355bf46 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -6,6 +6,7 @@ from __future__ import unicode_literals import itertools import json import os.path +import random import re import time import traceback @@ -16,23 +17,25 @@ from ..swfinterp import SWFInterpreter from ..compat import ( compat_chr, compat_parse_qs, - compat_urllib_parse, compat_urllib_parse_unquote, compat_urllib_parse_unquote_plus, + compat_urllib_parse_urlencode, compat_urllib_parse_urlparse, compat_urlparse, compat_str, ) from ..utils import ( clean_html, - encode_dict, + error_to_compat_str, ExtractorError, float_or_none, get_element_by_attribute, get_element_by_id, int_or_none, + mimetype2ext, orderedSet, parse_duration, + remove_quotes, remove_start, sanitized_Request, smuggle_url, @@ -41,6 +44,7 @@ from ..utils import ( unified_strdate, unsmuggle_url, uppercase_escape, + urlencode_postdata, ISO3166Utils, ) @@ -112,7 +116,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'hl': 'en_US', } - login_data = compat_urllib_parse.urlencode(encode_dict(login_form_strs)).encode('ascii') + login_data = urlencode_postdata(login_form_strs) req = sanitized_Request(self._LOGIN_URL, login_data) login_results = self._download_webpage( @@ -145,7 +149,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'TrustDevice': 'on', }) - tfa_data = compat_urllib_parse.urlencode(encode_dict(tfa_form_strs)).encode('ascii') + tfa_data = urlencode_postdata(tfa_form_strs) tfa_req = sanitized_Request(self._TWOFACTOR_URL, tfa_data) tfa_results = self._download_webpage( @@ -178,7 +182,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor): return -class YoutubeEntryListBaseInfoExtractor(InfoExtractor): +class YoutubeEntryListBaseInfoExtractor(YoutubeBaseInfoExtractor): # Extract entries from page with "Load more" button def _entries(self, page, playlist_id): more_widget_html = content_html = page @@ -230,7 +234,9 @@ class YoutubePlaylistBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor): class YoutubePlaylistsBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor): def _process_page(self, content): - for playlist_id in re.findall(r'href="/?playlist\?list=(.+?)"', content): + for playlist_id in orderedSet(re.findall( + r'