From 315bdae00a906b0e07d400b63754abadec34dab5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sat, 16 Jan 2016 06:25:25 +0600 Subject: [PATCH] [zippcast] Improve (Closes #8198) --- youtube_dl/extractor/zippcast.py | 92 ++++++++++++++++++++++---------- 1 file changed, 64 insertions(+), 28 deletions(-) diff --git a/youtube_dl/extractor/zippcast.py b/youtube_dl/extractor/zippcast.py index afeab4114..de819376d 100644 --- a/youtube_dl/extractor/zippcast.py +++ b/youtube_dl/extractor/zippcast.py @@ -1,58 +1,94 @@ -# coding: utf-8 from __future__ import unicode_literals +import re + from .common import InfoExtractor +from ..utils import ( + determine_ext, + str_to_int, +) class ZippCastIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?zippcast\.com/video/(?P[0-9a-zA-Z]+)' + _VALID_URL = r'https?://(?:www\.)?zippcast\.com/(?:video/|videoview\.php\?.*\bvplay=)(?P[0-9a-zA-Z]+)' _TESTS = [{ + # m3u8, hq direct link 'url': 'http://www.zippcast.com/video/c9cfd5c7e44dbc29c81', - 'md5': 'f2aea8659962d9155031aaeac53f7c54', + 'md5': '5ea0263b5606866c4d6cda0fc5e8c6b6', 'info_dict': { 'id': 'c9cfd5c7e44dbc29c81', 'ext': 'mp4', 'title': '[Vinesauce] Vinny - Digital Space Traveler', + 'description': 'Muted on youtube, but now uploaded in it\'s original form.', 'thumbnail': 're:^https?://.*\.jpg$', 'uploader': 'vinesauce', - 'description': 'Muted on youtube, but now uploaded in it\'s original form.', - 'categories': ['Entertainment'], 'view_count': int, + 'categories': ['Entertainment'], + 'tags': list, }, }, { + # f4m, lq ipod direct link 'url': 'http://www.zippcast.com/video/b79c0a233e9c6581775', - 'md5': 'b8631f0cc48ed15387f9179988d0c97c', - 'info_dict': { - 'id': 'b79c0a233e9c6581775', - 'ext': 'mp4', - 'title': 'Battlefield Hardline Trailer', - 'thumbnail': 're:^https?://.*\.jpg$', - 'uploader': 'IGXGaming', - 'description': 'Battlefield Hardline Trailer', - 'categories': ['Gaming'], - 'view_count': int, - }, + 'only_matching': True, + }, { + 'url': 'http://www.zippcast.com/videoview.php?vplay=c9cfd5c7e44dbc29c81&auto=no', + 'only_matching': True, }] def _real_extract(self, url): video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) - title = self._html_search_regex(r'title="(.+?)"', webpage, 'title') - uploader = self._html_search_regex(r'http://www.zippcast.com/profile/(.+?)">', webpage, 'uploader') - url = self._html_search_regex(r'(.+?)', webpage, 'description') - thumbnail = self._html_search_regex(r'poster="(.+?)" controls>', webpage, 'thumbnail') - categories = self._html_search_regex(r'

(.+?) views!', webpage, 'view_count') + webpage = self._download_webpage( + 'http://www.zippcast.com/video/%s' % video_id, video_id) + + formats = [] + video_url = self._search_regex( + r']+src=(["\'])(?P.+?)\1', webpage, + 'video url', default=None, group='url') + if video_url: + formats.append({ + 'url': video_url, + 'format_id': 'http', + 'preference': 0, # direct link is almost always of worse quality + }) + src_url = self._search_regex( + r'src\s*:\s*(?:escape\()?(["\'])(?Phttp://.+?)\1', + webpage, 'src', default=None, group='url') + ext = determine_ext(src_url) + if ext == 'm3u8': + formats.extend(self._extract_m3u8_formats( + src_url, video_id, 'mp4', entry_protocol='m3u8_native', + m3u8_id='hls', fatal=False)) + elif ext == 'f4m': + formats.extend(self._extract_f4m_formats( + src_url, video_id, f4m_id='hds', fatal=False)) + self._sort_formats(formats) + + title = self._og_search_title(webpage) + description = self._og_search_description(webpage) or self._html_search_meta( + 'description', webpage) + uploader = self._search_regex( + r']+href="https?://[^/]+/profile/[^>]+>([^<]+)', + webpage, 'uploader', fatal=False) + thumbnail = self._og_search_thumbnail(webpage) + view_count = str_to_int(self._search_regex( + r'>([\d,.]+) views!', webpage, 'view count', fatal=False)) + + categories = re.findall( + r']+href="https?://[^/]+/categories/[^"]+">([^<]+),?<', + webpage) + tags = re.findall( + r']+href="https?://[^/]+/search/tags/[^"]+">([^<]+),?<', + webpage) return { 'id': video_id, 'title': title, - 'url': url, 'description': description, - 'uploader': uploader, 'thumbnail': thumbnail, - 'categories': [categories], - 'view_count': int(view_count.replace(',', '')), + 'uploader': uploader, + 'view_count': view_count, + 'categories': categories, + 'tags': tags, + 'formats': formats, }