2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from ..utils import int_or_none
8 from datetime import datetime
11 class PolskieRadioIE(InfoExtractor):
12 _VALID_URL = r'https?://(?:www\.)?polskieradio\.pl/[0-9]+/[0-9]+/Artykul/(?P<id>[0-9]+),.+'
14 'url': 'http://www.polskieradio.pl/7/5102/Artykul/1587943,Prof-Andrzej-Nowak-o-historii-nie-da-sie-myslec-beznamietnie',
15 'md5': '2984ee6ce9046d91fc233bc1a864a09a',
19 'title': 'Prof. Andrzej Nowak: o historii nie da się myśleć beznamiętnie',
20 'description': 'md5:12f954edbf3120c5e7075e17bf9fc5c5',
21 'release_date': '20160227',
22 'upload_date': '20160227',
23 'timestamp': 1456594200,
27 'url': 'http://polskieradio.pl/9/305/Artykul/1632955,Bardzo-popularne-slowo-remis',
28 'md5': '68a393e25b942c1a76872f56d303a31a',
32 'title': 'Bardzo popularne słowo: remis',
33 'description': 'md5:3b58dfae614100abc0f175a0b26d5680',
34 'release_date': '20160617',
35 'upload_date': '20160617',
36 'timestamp': 1466184900,
41 def _real_extract(self, url):
42 video_id = self._match_id(url)
43 webpage = self._download_webpage(url, video_id)
44 metadata_string = self._html_search_regex(r'<span class="play pr-media-play" data-media=(\{.+\})>', webpage, 'metadata')
45 metadata = self._parse_json(metadata_string, video_id)
47 title = self._og_search_title(webpage)
51 description = self._og_search_description(webpage)
52 if description is not None:
53 description = description.strip()
55 release_date = self._html_search_regex(r'Data emisji:[^0-9]+([0-9]{1,2}\.[0-9]{2}\.[0-9]{4})', webpage, 'release date', fatal=False)
56 if release_date is not None:
57 release_date = datetime.strptime(release_date, '%d.%m.%Y').strftime('%Y%m%d')
59 upload_datetime = self._html_search_regex(r'<span id="datetime2" class="time">\s+(.+)\s+</span>', webpage, 'release time', fatal=False)
60 if upload_datetime is not None:
61 timestamp = calendar.timegm(datetime.strptime(upload_datetime, '%d.%m.%Y %H:%M').timetuple())
68 'description': description,
69 'display_id': metadata.get('id'),
70 'duration': int_or_none(metadata.get('length')),
71 'url': self._proto_relative_url(metadata.get('file'), 'http:'),
72 'release_date': release_date,
73 'timestamp': timestamp