From e94443de80d20a62fccda23fa25abcfa3798243a Mon Sep 17 00:00:00 2001 From: Yen Chi Hsuan Date: Tue, 21 Apr 2015 15:10:27 +0800 Subject: [PATCH] [Cinemassacre] Move to a standalone module --- youtube_dl/extractor/__init__.py | 3 +- youtube_dl/extractor/cinemassacre.py | 81 +++++++++++++++++++++++++ youtube_dl/extractor/screenwavemedia.py | 75 ----------------------- 3 files changed, 83 insertions(+), 76 deletions(-) create mode 100644 youtube_dl/extractor/cinemassacre.py diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 3bd6d16970..a64afa1da6 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -70,6 +70,7 @@ from .chirbit import ( ChirbitProfileIE, ) from .cinchcast import CinchcastIE +from .cinemassacre import CinemassacreIE from .clipfish import ClipfishIE from .cliphunter import CliphunterIE from .clipsyndicate import ClipsyndicateIE @@ -446,7 +447,7 @@ from .sbs import SBSIE from .scivee import SciVeeIE from .screencast import ScreencastIE from .screencastomatic import ScreencastOMaticIE -from .screenwavemedia import CinemassacreIE, ScreenwaveMediaIE, TeamFourIE +from .screenwavemedia import ScreenwaveMediaIE, TeamFourIE from .senateisvp import SenateISVPIE from .servingsys import ServingSysIE from .sexu import SexuIE diff --git a/youtube_dl/extractor/cinemassacre.py b/youtube_dl/extractor/cinemassacre.py new file mode 100644 index 0000000000..c8bbebe1a4 --- /dev/null +++ b/youtube_dl/extractor/cinemassacre.py @@ -0,0 +1,81 @@ +# encoding: utf-8 +from __future__ import unicode_literals + +import re + +from .common import InfoExtractor +from ..utils import ExtractorError +from .bliptv import BlipTVIE + + +class CinemassacreIE(InfoExtractor): + _VALID_URL = 'https?://(?:www\.)?cinemassacre\.com/(?P[0-9]{4})/(?P[0-9]{2})/(?P[0-9]{2})/(?P[^?#/]+)' + _TESTS = [ + { + 'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/', + 'md5': 'fde81fbafaee331785f58cd6c0d46190', + 'info_dict': { + 'id': 'Cinemassacre-19911', + 'ext': 'mp4', + 'upload_date': '20121110', + 'title': '“Angry Video Game Nerd: The Movie” – Trailer', + 'description': 'md5:fb87405fcb42a331742a0dce2708560b', + }, + }, + { + 'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940', + 'md5': 'd72f10cd39eac4215048f62ab477a511', + 'info_dict': { + 'id': 'Cinemassacre-521be8ef82b16', + 'ext': 'mp4', + 'upload_date': '20131002', + 'title': 'The Mummy’s Hand (1940)', + }, + }, + { + 'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/', + 'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de', + 'info_dict': { + 'id': '4065369', + 'ext': 'flv', + 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles', + 'upload_date': '20061207', + 'uploader': 'cinemassacre', + 'uploader_id': '250778', + 'timestamp': 1283233867, + 'description': 'md5:0a108c78d130676b207d0f6d029ecffd', + } + } + ] + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + display_id = mobj.group('display_id') + video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d') + + webpage = self._download_webpage(url, display_id) + + playerdata_url = self._search_regex( + r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', + webpage, 'player data URL', default=None) + if not playerdata_url: + playerdata_url = BlipTVIE._extract_url(webpage) + if not playerdata_url: + raise ExtractorError('Unable to find player data') + + video_title = self._html_search_regex( + r'(?P<title>.+?)\|', webpage, 'title') + video_description = self._html_search_regex( + r'<div class="entry-content">(?P<description>.+?)</div>', + webpage, 'description', flags=re.DOTALL, fatal=False) + video_thumbnail = self._og_search_thumbnail(webpage) + + return { + '_type': 'url_transparent', + 'display_id': display_id, + 'title': video_title, + 'description': video_description, + 'upload_date': video_date, + 'thumbnail': video_thumbnail, + 'url': playerdata_url, + } diff --git a/youtube_dl/extractor/screenwavemedia.py b/youtube_dl/extractor/screenwavemedia.py index b515b11b45..74fb1983ac 100644 --- a/youtube_dl/extractor/screenwavemedia.py +++ b/youtube_dl/extractor/screenwavemedia.py @@ -7,9 +7,7 @@ from .common import InfoExtractor from ..utils import ( int_or_none, unified_strdate, - ExtractorError ) -from .bliptv import BlipTVIE class ScreenwaveMediaIE(InfoExtractor): @@ -83,79 +81,6 @@ class ScreenwaveMediaIE(InfoExtractor): } -class CinemassacreIE(InfoExtractor): - _VALID_URL = 'https?://(?:www\.)?cinemassacre\.com/(?P<date_y>[0-9]{4})/(?P<date_m>[0-9]{2})/(?P<date_d>[0-9]{2})/(?P<display_id>[^?#/]+)' - _TESTS = [ - { - 'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/', - 'md5': 'fde81fbafaee331785f58cd6c0d46190', - 'info_dict': { - 'id': 'Cinemassacre-19911', - 'ext': 'mp4', - 'upload_date': '20121110', - 'title': '“Angry Video Game Nerd: The Movie” – Trailer', - 'description': 'md5:fb87405fcb42a331742a0dce2708560b', - }, - }, - { - 'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940', - 'md5': 'd72f10cd39eac4215048f62ab477a511', - 'info_dict': { - 'id': 'Cinemassacre-521be8ef82b16', - 'ext': 'mp4', - 'upload_date': '20131002', - 'title': 'The Mummy’s Hand (1940)', - }, - }, - { - 'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/', - 'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de', - 'info_dict': { - 'id': '4065369', - 'ext': 'flv', - 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles', - 'upload_date': '20061207', - 'uploader': 'cinemassacre', - 'uploader_id': '250778', - 'timestamp': 1283233867, - 'description': 'md5:0a108c78d130676b207d0f6d029ecffd', - } - } - ] - - def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - display_id = mobj.group('display_id') - video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d') - - webpage = self._download_webpage(url, display_id) - - playerdata_url = self._search_regex( - r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', - webpage, 'player data URL', default=None) - if not playerdata_url: - playerdata_url = BlipTVIE._extract_url(webpage) - if not playerdata_url: - raise ExtractorError('Unable to find player data') - - video_title = self._html_search_regex( - r'<title>(?P<title>.+?)\|', webpage, 'title') - video_description = self._html_search_regex( - r'<div class="entry-content">(?P<description>.+?)</div>', - webpage, 'description', flags=re.DOTALL, fatal=False) - video_thumbnail = self._og_search_thumbnail(webpage) - - return { - '_type': 'url_transparent', - 'display_id': display_id, - 'title': video_title, - 'description': video_description, - 'upload_date': video_date, - 'thumbnail': video_thumbnail, - 'url': playerdata_url, - } - - class TeamFourIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?teamfourstar\.com/video/(?P<id>[a-z0-9\-]+)/?' _TEST = {