From f50365e91cfccf33d0b5696c7f989944bcf748e7 Mon Sep 17 00:00:00 2001 From: Remita Amine Date: Sat, 13 Aug 2016 09:08:57 +0100 Subject: [PATCH] [pbs] add test for videos with undocumented http formats and remove unused import --- youtube_dl/extractor/pbs.py | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) diff --git a/youtube_dl/extractor/pbs.py b/youtube_dl/extractor/pbs.py index 6e2ef0fba..335e44bdc 100644 --- a/youtube_dl/extractor/pbs.py +++ b/youtube_dl/extractor/pbs.py @@ -4,7 +4,6 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import compat_HTTPError from ..utils import ( ExtractorError, determine_ext, @@ -334,6 +333,16 @@ class PBSIE(InfoExtractor): 'formats': 'mincount:8', }, }, + { + # has undocumented http formats(4500k and 6500k) + 'url': 'http://www.pbs.org/video/2365815229/', + 'md5': '94635cd06b7133688e23f4b94e6637a5', + 'info_dict': { + 'id': '2365815229', + 'ext': 'mp4', + 'title': 'FRONTLINE - Mosquito Hunter', + }, + }, { 'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true', 'only_matching': True, @@ -501,14 +510,18 @@ class PBSIE(InfoExtractor): if http_url: for m3u8_format in m3u8_formats: bitrate = self._search_regex(r'(\d+)k', m3u8_format['url'], 'bitrate', default=None) - # extract only the formats that we know that they will be available as http format. + # lower qualities(150k and 192k) are not available as http formats + # https://github.com/rg3/youtube-dl/commit/cbc032c8b70a038a69259378c92b4ba97b42d491#commitcomment-17313656 + # we will try to extract any http format higher than than the lowest quality documented in # https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications + # as there also undocumented http formats formats(4500k and 6500k) + # http://www.pbs.org/video/2365815229/ if not bitrate or int(bitrate) < 400: continue f_url = re.sub(r'\d+k|baseline', bitrate + 'k', http_url) # This may produce invalid links sometimes (e.g. # http://www.pbs.org/wgbh/frontline/film/suicide-plan) - if not self._is_valid_url(f_url, display_id, 'http-%s video' % bitrate): + if not self._is_valid_url(f_url, display_id, 'http-%sk video' % bitrate): continue f = m3u8_format.copy() f.update({