Make RSS enclosure type and length optional

Author: NicolasLM

atoma/rss.py

@@ -23,8 +23,8 @@ class RSSImage:
 @attr.s
 class RSSEnclosure:
     url: str = attr.ib()
-    length: int = attr.ib()
-    type: str = attr.ib()
+    length: Optional[int] = attr.ib()
+    type: Optional[str] = attr.ib()
 
 
 @attr.s
@@ -102,10 +102,14 @@ def _get_source(element: Element, name,
 
 
 def _get_enclosure(element: Element) -> RSSEnclosure:
+    length = element.attrib.get('length')
+    if length is not None:
+        length = int(length)
+
     return RSSEnclosure(
         element.attrib['url'],
-        int(element.attrib['length']),
-        element.attrib['type'],
+        length,
+        element.attrib.get('type'),
     )
 
 

tests/rss/broken-enclosure.xml

@@ -0,0 +1,12 @@
+<?xml version="1.0"?>
+<rss version="2.0">
+<channel>
+    <title>Foo</title>
+    <link>http://foo.bar</link>
+    <description>Foo bar.</description>
+    <item>
+        <title>Baz item</title>
+        <enclosure url="https://foo.com/test.mp3" />
+    </item>
+</channel>
+</rss>

tests/rss/test_rss.py

@@ -58,6 +58,15 @@ def test_broken_missing_source_url():
     assert p.items[0].source.url is None
 
 
+def test_broken_enclosure():
+    # The length and type of an enclosure are mandatory by specs,
+    # but some feeds in the wild do not provide them
+    p = parse_rss_file('tests/rss/broken-enclosure.xml')
+    assert p.items[0].enclosures[0].url == 'https://foo.com/test.mp3'
+    assert p.items[0].enclosures[0].length is None
+    assert p.items[0].enclosures[0].type is None
+
+
 def test_broken_version():
     with pytest.raises(FeedParseError):
         parse_rss_file('tests/rss/broken-version.xml')