From f0d3669437bb7f198ada9c0fead64d50a6e7a972 Mon Sep 17 00:00:00 2001 From: Remita Amine Date: Fri, 12 Aug 2016 18:05:49 +0100 Subject: [PATCH] [hgtv] Add new extractor(closes #3999) --- youtube_dl/extractor/extractors.py | 1 + youtube_dl/extractor/hgtv.py | 48 ++++++++++++++++++++++++++++++ 2 files changed, 49 insertions(+) create mode 100644 youtube_dl/extractor/hgtv.py diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py index acf4e5d62..6420167f2 100644 --- a/youtube_dl/extractor/extractors.py +++ b/youtube_dl/extractor/extractors.py @@ -325,6 +325,7 @@ from .heise import HeiseIE from .hellporno import HellPornoIE from .helsinki import HelsinkiIE from .hentaistigma import HentaiStigmaIE +from .hgtv import HGTVIE from .historicfilms import HistoricFilmsIE from .hitbox import HitboxIE, HitboxLiveIE from .hornbunny import HornBunnyIE diff --git a/youtube_dl/extractor/hgtv.py b/youtube_dl/extractor/hgtv.py new file mode 100644 index 000000000..c3f0733cf --- /dev/null +++ b/youtube_dl/extractor/hgtv.py @@ -0,0 +1,48 @@ +# coding: utf-8 +from __future__ import unicode_literals + +from .common import InfoExtractor +from ..utils import ( + int_or_none, + js_to_json, + smuggle_url, +) + + +class HGTVIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?hgtv\.ca/[^/]+/video/(?P[^/]+)/video.html' + _TEST = { + 'url': 'http://www.hgtv.ca/homefree/video/overnight-success/video.html?v=738081859718&p=1&s=da#video', + 'md5': '', + 'info_dict': { + 'id': 'aFH__I_5FBOX', + 'ext': 'mp4', + 'title': 'Overnight Success', + 'description': 'After weeks of hard work, high stakes, breakdowns and pep talks, the final 2 contestants compete to win the ultimate dream.', + 'uploader': 'SHWM-NEW', + 'timestamp': 1470320034, + 'upload_date': '20160804', + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + } + + def _real_extract(self, url): + display_id = self._match_id(url) + webpage = self._download_webpage(url, display_id) + embed_vars = self._parse_json(self._search_regex( + r'(?s)embed_vars\s*=\s*({.*?});', + webpage, 'embed vars'), display_id, js_to_json) + return { + '_type': 'url_transparent', + 'url': smuggle_url( + 'http://link.theplatform.com/s/dtjsEC/%s?mbr=true&manifest=m3u' % embed_vars['pid'], { + 'force_smil_url': True + }), + 'series': embed_vars.get('show'), + 'season_number': int_or_none(embed_vars.get('season')), + 'episode_number': int_or_none(embed_vars.get('episode')), + 'ie_key': 'ThePlatform', + }