From c7bab7c39fd51c0812f70020172766303191bc01 Mon Sep 17 00:00:00 2001
From: Daniel Baumann <daniel.baumann@progress-linux.org>
Date: Sun, 5 May 2024 19:37:42 +0200
Subject: Adding upstream version 2023.03.04.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
---
 yt_dlp/extractor/karrierevideos.py | 96 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)
 create mode 100644 yt_dlp/extractor/karrierevideos.py

(limited to 'yt_dlp/extractor/karrierevideos.py')

diff --git a/yt_dlp/extractor/karrierevideos.py b/yt_dlp/extractor/karrierevideos.py
new file mode 100644
index 0000000..28d4841
--- /dev/null
+++ b/yt_dlp/extractor/karrierevideos.py
@@ -0,0 +1,96 @@
+from .common import InfoExtractor
+from ..compat import compat_urlparse
+from ..utils import (
+    fix_xml_ampersands,
+    float_or_none,
+    xpath_with_ns,
+    xpath_text,
+)
+
+
+class KarriereVideosIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?karrierevideos\.at(?:/[^/]+)+/(?P<id>[^/]+)'
+    _TESTS = [{
+        'url': 'http://www.karrierevideos.at/berufsvideos/mittlere-hoehere-schulen/altenpflegerin',
+        'info_dict': {
+            'id': '32c91',
+            'ext': 'flv',
+            'title': 'AltenpflegerIn',
+            'description': 'md5:dbadd1259fde2159a9b28667cb664ae2',
+            'thumbnail': r're:^http://.*\.png',
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,
+        }
+    }, {
+        # broken ampersands
+        'url': 'http://www.karrierevideos.at/orientierung/vaeterkarenz-und-neue-chancen-fuer-muetter-baby-was-nun',
+        'info_dict': {
+            'id': '5sniu',
+            'ext': 'flv',
+            'title': 'Väterkarenz und neue Chancen für Mütter - "Baby - was nun?"',
+            'description': 'md5:97092c6ad1fd7d38e9d6a5fdeb2bcc33',
+            'thumbnail': r're:^http://.*\.png',
+        },
+        'params': {
+            # rtmp download
+            'skip_download': True,
+        }
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, video_id)
+
+        title = (self._html_search_meta('title', webpage, default=None)
+                 or self._search_regex(r'<h1 class="title">([^<]+)</h1>', webpage, 'video title'))
+
+        video_id = self._search_regex(
+            r'/config/video/(.+?)\.xml', webpage, 'video id')
+        # Server returns malformed headers
+        # Force Accept-Encoding: * to prevent gzipped results
+        playlist = self._download_xml(
+            'http://www.karrierevideos.at/player-playlist.xml.php?p=%s' % video_id,
+            video_id, transform_source=fix_xml_ampersands,
+            headers={'Accept-Encoding': '*'})
+
+        NS_MAP = {
+            'jwplayer': 'http://developer.longtailvideo.com/trac/wiki/FlashFormats'
+        }
+
+        def ns(path):
+            return xpath_with_ns(path, NS_MAP)
+
+        item = playlist.find('./tracklist/item')
+        video_file = xpath_text(
+            item, ns('./jwplayer:file'), 'video url', fatal=True)
+        streamer = xpath_text(
+            item, ns('./jwplayer:streamer'), 'streamer', fatal=True)
+
+        uploader = xpath_text(
+            item, ns('./jwplayer:author'), 'uploader')
+        duration = float_or_none(
+            xpath_text(item, ns('./jwplayer:duration'), 'duration'))
+
+        description = self._html_search_regex(
+            r'(?s)<div class="leadtext">(.+?)</div>',
+            webpage, 'description')
+
+        thumbnail = self._html_search_meta(
+            'thumbnail', webpage, 'thumbnail')
+        if thumbnail:
+            thumbnail = compat_urlparse.urljoin(url, thumbnail)
+
+        return {
+            'id': video_id,
+            'url': streamer.replace('rtmpt', 'rtmp'),
+            'play_path': 'mp4:%s' % video_file,
+            'ext': 'flv',
+            'title': title,
+            'description': description,
+            'thumbnail': thumbnail,
+            'uploader': uploader,
+            'duration': duration,
+        }
-- 
cgit v1.2.3