From 27fe903c511691c078942bef5ee9a05a43b15c8f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jes=C3=BAs?= <heckyel@hyperbola.info>
Date: Wed, 9 Jun 2021 17:54:27 -0500
Subject: initial

---
 hypervideo_dl/extractor/rozhlas.py | 50 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100644 hypervideo_dl/extractor/rozhlas.py

(limited to 'hypervideo_dl/extractor/rozhlas.py')

diff --git a/hypervideo_dl/extractor/rozhlas.py b/hypervideo_dl/extractor/rozhlas.py
new file mode 100644
index 0000000..fccf694
--- /dev/null
+++ b/hypervideo_dl/extractor/rozhlas.py
@@ -0,0 +1,50 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+from ..utils import (
+    int_or_none,
+    remove_start,
+)
+
+
+class RozhlasIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?prehravac\.rozhlas\.cz/audio/(?P<id>[0-9]+)'
+    _TESTS = [{
+        'url': 'http://prehravac.rozhlas.cz/audio/3421320',
+        'md5': '504c902dbc9e9a1fd50326eccf02a7e2',
+        'info_dict': {
+            'id': '3421320',
+            'ext': 'mp3',
+            'title': 'Echo Pavla Klusáka (30.06.2015 21:00)',
+            'description': 'Osmdesátiny Terryho Rileyho jsou skvělou příležitostí proletět se elektronickými i akustickými díly zakladatatele minimalismu, který je aktivní už přes padesát let'
+        }
+    }, {
+        'url': 'http://prehravac.rozhlas.cz/audio/3421320/embed',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        audio_id = self._match_id(url)
+
+        webpage = self._download_webpage(
+            'http://prehravac.rozhlas.cz/audio/%s' % audio_id, audio_id)
+
+        title = self._html_search_regex(
+            r'<h3>(.+?)</h3>\s*<p[^>]*>.*?</p>\s*<div[^>]+id=["\']player-track',
+            webpage, 'title', default=None) or remove_start(
+            self._og_search_title(webpage), 'Radio Wave - ')
+        description = self._html_search_regex(
+            r'<p[^>]+title=(["\'])(?P<url>(?:(?!\1).)+)\1[^>]*>.*?</p>\s*<div[^>]+id=["\']player-track',
+            webpage, 'description', fatal=False, group='url')
+        duration = int_or_none(self._search_regex(
+            r'data-duration=["\'](\d+)', webpage, 'duration', default=None))
+
+        return {
+            'id': audio_id,
+            'url': 'http://media.rozhlas.cz/_audio/%s.mp3' % audio_id,
+            'title': title,
+            'description': description,
+            'duration': duration,
+            'vcodec': 'none',
+        }
-- 
cgit v1.2.3