From 9dac2cec2d7e31b65cf063164b3a99f257a86a63 Mon Sep 17 00:00:00 2001 From: slocum Date: Sun, 9 Apr 2017 12:02:44 +0200 Subject: [PATCH] [zaq1] Add new extractor --- youtube_dl/extractor/extractors.py | 1 + youtube_dl/extractor/zaq1.py | 67 ++++++++++++++++++++++++++++++ 2 files changed, 68 insertions(+) create mode 100644 youtube_dl/extractor/zaq1.py diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py index 612007a46..c0020dd7d 100644 --- a/youtube_dl/extractor/extractors.py +++ b/youtube_dl/extractor/extractors.py @@ -1300,5 +1300,6 @@ from .youtube import ( YoutubeWatchLaterIE, ) from .zapiks import ZapiksIE +from .zaq1 import Zaq1IE from .zdf import ZDFIE, ZDFChannelIE from .zingmp3 import ZingMp3IE diff --git a/youtube_dl/extractor/zaq1.py b/youtube_dl/extractor/zaq1.py new file mode 100644 index 000000000..33068f6b5 --- /dev/null +++ b/youtube_dl/extractor/zaq1.py @@ -0,0 +1,67 @@ +# coding: utf-8 +from __future__ import unicode_literals + +from .common import InfoExtractor +from ..utils import ( + unified_strdate, + int_or_none +) + + +class Zaq1IE(InfoExtractor): + _VALID_URL = r'http://(?:www\.)?zaq1\.pl/video/(?P[^/?#&]+)' + _TESTS = [{ + 'url': 'http://zaq1.pl/video/xev0e', + 'md5': '24a5eb3f052e604ae597c4d0d19b351e', + 'info_dict': { + 'id': 'xev0e', + 'title': 'DJ NA WESELE. TANIEC Z FIGURAMI.węgrów/sokołów podlaski/siedlce/mińsk mazowiecki/warszawa', + 'ext': 'mp4', + 'duration': 511, + 'uploader': 'Anonim', + 'upload_date': '20170330', + } + }, { + 'url': 'http://zaq1.pl/video/x80nc', + 'md5': '1245973520adc78139928a820959d9c5', + 'info_dict': { + 'id': 'x80nc', + 'title': 'DIY Inspiration Challenge #86 | koraliki | gwiazdka na choinkę z koralików i drutu', + 'ext': 'mp4', + 'duration': 438, + 'uploader': 'Anonim', + 'upload_date': '20170404', + } + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + title = self._html_search_regex( + r'(?s)

\s*\1\s*\s*

', webpage, 'title') + + div = self._search_regex(r'(?s)(?P
)', webpage, 'video url', group='div') + video_url = self._search_regex(r'data-video-url="(http[^"]+)"', div, 'video url') + + ext = self._search_regex(r'data-file-extension="([^"]+)"', div, 'ext', None, False) + duration = int_or_none(self._search_regex(r'data-duration="([^"]+)"', div, 'duration', None, False)) + thumbnail = self._search_regex(r'data-photo-url="([^"]+)"', div, 'thumbnail', None, False) + + upload_date = unified_strdate(self._search_regex(r'\s*Opublikowany\s+([0-9]{4}-[0-9]{2}-[0-9]{2})', webpage, 'upload date')) + uploader = self._search_regex(r'.*Wideo dodał:\s*]*>\s*([^<]+)\s*', webpage, 'uploader') + + return { + 'id': video_id, + 'title': title, + 'formats': [{ + 'url': video_url, + 'ext': ext, + 'http_headers': {'Referer': url}, + }], + 'thumbnail': thumbnail, + 'uploader': uploader, + 'upload_date': upload_date, + 'duration': duration, + } -- 2.39.5