[peekvids] Add extractor (#2414)

Authored by: schn0sch

yt_dlp/extractor/caltrans.py

@@ -0,0 +1,41 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+
+
+class CaltransIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:[^/]+\.)?ca\.gov/vm/loc/[^/]+/(?P<id>[a-z0-9_]+)\.htm'
+    _TEST = {
+        'url': 'https://cwwp2.dot.ca.gov/vm/loc/d3/hwy50at24th.htm',
+        'info_dict': {
+            'id': 'hwy50at24th',
+            'ext': 'ts',
+            'title': 'US-50 : Sacramento : Hwy 50 at 24th',
+            'live_status': 'is_live',
+            'thumbnail': 'https://cwwp2.dot.ca.gov/data/d3/cctv/image/hwy50at24th/hwy50at24th.jpg',
+        }
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+
+        global_vars = self._search_regex(
+            r'<script[^<]+?([^<]+\.m3u8[^<]+)</script>',
+            webpage, 'Global Vars')
+        route_place = self._search_regex(r'routePlace\s*=\s*"([^"]+)"', global_vars, 'Route Place', fatal=False)
+        location_name = self._search_regex(r'locationName\s*=\s*"([^"]+)"', global_vars, 'Location Name', fatal=False)
+        poster_url = self._search_regex(r'posterURL\s*=\s*"([^"]+)"', global_vars, 'Poster Url', fatal=False)
+        video_stream = self._search_regex(r'videoStreamURL\s*=\s*"([^"]+)"', global_vars, 'Video Stream URL', fatal=False)
+
+        formats = self._extract_m3u8_formats(video_stream, video_id, 'ts', live=True)
+        self._sort_formats(formats)
+
+        return {
+            'id': video_id,
+            'title': f'{route_place} : {location_name}',
+            'is_live': True,
+            'formats': formats,
+            'thumbnail': poster_url,
+        }

yt_dlp/extractor/extractors.py

@@ -195,6 +195,7 @@ from .byutv import BYUtvIE
 from .c56 import C56IE
 from .cableav import CableAVIE
 from .callin import CallinIE
+from .caltrans import CaltransIE
 from .cam4 import CAM4IE
 from .camdemy import (
     CamdemyIE,
@@ -1140,6 +1141,7 @@ from .patreon import (
 )
 from .pbs import PBSIE
 from .pearvideo import PearVideoIE
+from .peekvids import PeekVidsIE, PlayVidsIE
 from .peertube import (
     PeerTubeIE,
     PeerTubePlaylistIE,

yt_dlp/extractor/peekvids.py

@@ -0,0 +1,85 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+from ..utils import remove_end
+
+
+class PeekVidsIE(InfoExtractor):
+    _VALID_URL = r'''(?x)
+        https?://(?:www\.)?peekvids\.com/
+        (?:(?:[^/?#]+/){2}|embed/?\?(?:[^#]*&)?v=)
+        (?P<id>[^/?&#]*)
+    '''
+    _TESTS = [{
+        'url': 'https://peekvids.com/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp/BSyLMbN0YCd',
+        'md5': '2ff6a357a9717dc9dc9894b51307e9a2',
+        'info_dict': {
+            'id': 'BSyLMbN0YCd',
+            'ext': 'mp4',
+            'title': 'Dane Jones - Cute redhead with perfect tits with Mini Vamp',
+            'age_limit': 18,
+        },
+    }]
+    _DOMAIN = 'www.peekvids.com'
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+
+        short_video_id = self._html_search_regex(r'<video [^>]*data-id="(.+?)"', webpage, 'short video ID')
+        srcs = self._download_json(
+            f'https://{self._DOMAIN}/v-alt/{short_video_id}', video_id,
+            note='Downloading list of source files')
+        formats = [{
+            'url': url,
+            'ext': 'mp4',
+            'format_id': name[8:],
+        } for name, url in srcs.items() if len(name) > 8 and name.startswith('data-src')]
+        if not formats:
+            formats = [{'url': url} for url in srcs.values()]
+        self._sort_formats(formats)
+
+        title = remove_end(self._html_search_regex(
+            (r'<h1.*?>\s*(.+?)\s*</h1>', r'<title>\s*(.+?)\s*</title>'),
+            webpage, 'video title', default=None), ' - PeekVids')
+
+        return {
+            'id': video_id,
+            'title': title,
+            'age_limit': 18,
+            'formats': formats,
+        }
+
+
+class PlayVidsIE(PeekVidsIE):
+    _VALID_URL = r'https?://(?:www\.)?playvids\.com/(?:embed/|[^/]{2}/)?(?P<id>[^/?#]*)'
+    _TESTS = [{
+        'url': 'https://www.playvids.com/U3pBrYhsjXM/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp',
+        'md5': '2f12e50213dd65f142175da633c4564c',
+        'info_dict': {
+            'id': 'U3pBrYhsjXM',
+            'ext': 'mp4',
+            'title': 'Dane Jones - Cute redhead with perfect tits with Mini Vamp',
+            'age_limit': 18,
+        },
+    }, {
+        'url': 'https://www.playvids.com/es/U3pBrYhsjXM/pc/dane-jones-cute-redhead-with-perfect-tits-with-mini-vamp',
+        'md5': '2f12e50213dd65f142175da633c4564c',
+        'info_dict': {
+            'id': 'U3pBrYhsjXM',
+            'ext': 'mp4',
+            'title': 'Dane Jones - Cute redhead with perfect tits with Mini Vamp',
+            'age_limit': 18,
+        },
+    }, {
+        'url': 'https://www.playvids.com/embed/U3pBrYhsjXM',
+        'md5': '2f12e50213dd65f142175da633c4564c',
+        'info_dict': {
+            'id': 'U3pBrYhsjXM',
+            'ext': 'mp4',
+            'title': 'U3pBrYhsjXM',
+            'age_limit': 18,
+        },
+    }]
+    _DOMAIN = 'www.playvids.com'