From f870544302f75bee0d96f6a8623c8ff270beca89 Mon Sep 17 00:00:00 2001 From: fnord Date: Mon, 13 Jul 2015 07:41:38 -0500 Subject: [PATCH] Add support for democracynow.org Supports downloading clips or entire shows. Subtitle support --- youtube_dl/extractor/__init__.py | 1 + youtube_dl/extractor/democracynow.py | 100 +++++++++++++++++++++++++++ 2 files changed, 101 insertions(+) create mode 100644 youtube_dl/extractor/democracynow.py diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index cbaa07391..5cc03b875 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -112,6 +112,7 @@ from .daum import DaumIE from .dbtv import DBTVIE from .dctp import DctpTvIE from .deezer import DeezerPlaylistIE +from .democracynow import DemocracynowIE from .dfb import DFBIE from .dhm import DHMIE from .dotsub import DotsubIE diff --git a/youtube_dl/extractor/democracynow.py b/youtube_dl/extractor/democracynow.py new file mode 100644 index 000000000..1c9b36052 --- /dev/null +++ b/youtube_dl/extractor/democracynow.py @@ -0,0 +1,100 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import json +import time +import hmac +import hashlib +import itertools +import re +from ..utils import ( + ExtractorError, + int_or_none, + parse_age_limit, + parse_iso8601, +) +from ..compat import compat_urllib_request +from .common import InfoExtractor + + +class DemocracynowIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?democracynow.org/?(?P[^\?]*)' + IE_NAME = 'democracynow' + _TESTS = [{ + 'url': 'http://www.democracynow.org/shows/2015/7/3', + 'info_dict': { + 'id': '2015-0703-001', + 'ext': 'mp4', + 'title': 'July 03, 2015 - Democracy Now!', + 'description': 'A daily independent global news hour with Amy Goodman & Juan Gonz\xe1lez "What to the Slave is 4th of July?": James Earl Jones Reads Frederick Douglass\u2019 Historic Speech : "This Flag Comes Down Today": Bree Newsome Scales SC Capitol Flagpole, Takes Down Confederate Flag : "We Shall Overcome": Remembering Folk Icon, Activist Pete Seeger in His Own Words & Songs', + 'uploader': 'Democracy Now', + 'upload_date': None, + }, + },{ + 'url': 'http://www.democracynow.org/2015/7/3/this_flag_comes_down_today_bree', + 'info_dict': { + 'id': '2015-0703-001', + 'ext': 'mp4', + 'title': '"This Flag Comes Down Today": Bree Newsome Scales SC Capitol Flagpole, Takes Down Confederate Flag', + 'description': 'md5:4d2bc4f0d29f5553c2210a4bc7761a21', + 'uploader': 'Democracy Now', + 'upload_date': None, + }, + + }] + + def _real_extract(self, url): + display_id = self._match_id(url) + base_host = re.search(r'^(.+?://[^/]+)', url).group(1) + if display_id == '': + display_id = 'home' + webpage = self._download_webpage(url, display_id) + re_desc = re.search(r'[^/]+)/(?:dn)?(?P[^/]+?)\.(?P[^\.\?]+)(?P\?|$)',url) + if video_id == None: + video_id = purl.group('fn') + if js.get('start') != None: + url += '&' if purl.group('hasparams') == '?' else '?' + url = url + 'start='+str(js.get('start')) + formats.append({ + 'format_id': purl.group('dir'), + 'ext': purl.group('ext'), + 'url': url, + }) + self._sort_formats(formats) + ret = { + 'id': video_id, + 'title': js.get('title'), + 'description': description, + 'uploader': 'Democracy Now', +# 'thumbnails': thumbnails, + 'subtitles': subtitles, + 'formats': formats, + } + return ret +# \ No newline at end of file