Parse alternate enclosures for podcast rss

newswire.py

@@ -623,11 +623,61 @@ def xml_podcast_to_dict(base_dir: str, xml_item: str, xml_str: str) -> {}:
     return podcast_properties
 
 
-def get_link_from_rss_item(rss_item: str) -> (str, str):
+def get_link_from_rss_item(rss_item: str,
+                           preferred_mime_types: [] = None,
+                           proxy_type: str = None) -> (str, str):
     """Extracts rss link from rss item string
     """
     mime_type = None
 
+    if preferred_mime_types and '<podcast:alternateEnclosure ' in rss_item:
+        enclosures = rss_item.split('<podcast:alternateEnclosure ')
+        ctr = 0
+        for enclosure in enclosures:
+            if ctr == 0:
+                ctr += 1
+                continue
+            ctr += 1
+            if '</podcast:alternateEnclosure' not in enclosure:
+                continue
+            enclosure = enclosure.split('</podcast:alternateEnclosure')[0]
+            if 'type="' not in enclosure:
+                continue
+            mime_type = enclosure.split('type="')[1]
+            if '"' in mime_type:
+                mime_type = mime_type.split('"')[0]
+            if mime_type not in preferred_mime_types:
+                continue
+            if 'uri="' not in enclosure:
+                continue
+            uris = enclosure.split('uri="')
+            ctr2 = 0
+            for uri in uris:
+                if ctr2 == 0:
+                    ctr2 += 1
+                    continue
+                ctr2 += 1
+                if '"' not in uri:
+                    continue
+                link = uri.split('"')[0]
+                if '://' not in link:
+                    continue
+                if proxy_type:
+                    if proxy_type == 'tor' and \
+                       '.onion/' not in link:
+                        continue
+                    if proxy_type == 'onion' and \
+                       '.onion/' not in link:
+                        continue
+                    if proxy_type == 'i2p' and \
+                       '.i2p/' not in link:
+                        continue
+                    return link, mime_type
+                else:
+                    if '.onion/' not in link and \
+                       '.i2p/' not in link:
+                        return link, mime_type
+
     if '<enclosure ' in rss_item:
         # get link from audio or video enclosure
         enclosure = rss_item.split('<enclosure ')[1]

tests.py

@@ -6795,18 +6795,49 @@ def _test_get_link_from_rss_item() -> None:
         '<link>' + \
         'https://anchor.fm/creativecommons/episodes/' + \
         'Hessel-van-Oorschot-of-Tribe-of-Noise--Free-Music-Archive-e1crvce' + \
-        '</link>' + \
-        '<pubDate>Wed, 12 Jan 2022 14:28:46 GMT</pubDate>' + \
+        '</link>\n' + \
+        '<pubDate>Wed, 12 Jan 2022 14:28:46 GMT</pubDate>\n' + \
         '<enclosure url="https://anchor.fm/s/4d70d828/podcast/' + \
         'play/46054222/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net' + \
         '%2Fstaging%2F2022-0-12%2F7352f28c-a928-ea7a-65ae-' + \
-        'ccb5edffbac1.mp3" length="67247880" type="audio/mpeg"/>'
-    link, mime_type = get_link_from_rss_item(rss_item)
+        'ccb5edffbac1.mp3" length="67247880" type="audio/mpeg"/>\n' + \
+        '<podcast:alternateEnclosure type="audio/mpeg" ' + \
+        'length="27800000" bitrate="128000" default="true" ' + \
+        'title="Standard">\n' + \
+        '<podcast:source uri="https://whoframed.rodger/rabbit.mp3" />\n' + \
+        '<podcast:source uri="http://randomaddress.onion/rabbit.mp3" />\n' + \
+        '<podcast:source uri="http://randomaddress.i2p/rabbit.mp3" />\n' + \
+        '</podcast:alternateEnclosure>\n' + \
+        '<podcast:alternateEnclosure type="audio/opus" ' + \
+        'length="19200000" bitrate="128000" ' + \
+        'title="High Quality">\n' + \
+        '<podcast:source uri="https://whoframed.rodger/rabbit.opus" />\n' + \
+        '<podcast:source uri="http://randomaddress.onion/rabbit.opus" />\n' + \
+        '<podcast:source uri="http://randomaddress.i2p/rabbit.opus" />\n' + \
+        '</podcast:alternateEnclosure>\n'
+
+    link, mime_type = get_link_from_rss_item(rss_item, None, None)
     assert link
-    assert link.endswith('.mp3')
+    assert link.endswith('1.mp3')
     assert mime_type
     assert mime_type == 'audio/mpeg'
 
+    link, mime_type = get_link_from_rss_item(rss_item, ['audio/opus'], None)
+    assert mime_type
+    if mime_type != 'audio/opus':
+        print('mime_type: ' + mime_type)
+    assert mime_type == 'audio/opus'
+    assert link
+    assert link == 'https://whoframed.rodger/rabbit.opus'
+
+    link, mime_type = get_link_from_rss_item(rss_item, ['audio/opus'], 'tor')
+    assert mime_type
+    if mime_type != 'audio/opus':
+        print('mime_type: ' + mime_type)
+    assert mime_type == 'audio/opus'
+    assert link
+    assert link == 'http://randomaddress.onion/rabbit.opus'
+
     rss_item = \
         '<link>' + \
         'https://anchor.fm/creativecommons/episodes/' + \