Filter rss enclosures by application/x-nzb mimetype (#3093)

* Filter rss enclosures by application/x-nzb mimetype

* Add tests and fix the failing one

* Change empty RSS entry logging to info

sabnzbd/rss.py

@@ -649,19 +649,26 @@ def _get_link(entry):
     """Retrieve the post link from this entry
     Returns (link, category, size)
     """
+    link = None
     size = 0
     age = datetime.datetime.now()
 
     # Try standard link and enclosures first
-    link = entry.link
-    if not link:
-        link = entry.links[0].href
-    if "enclosures" in entry:
+    if "enclosures" in entry and entry["enclosures"]:
         try:
-            link = entry.enclosures[0]["href"]
-            size = int(entry.enclosures[0]["length"])
+            for enclosure in entry["enclosures"]:
+                if "type" in enclosure and enclosure["type"] != "application/x-nzb":
+                    continue
+
+                link = enclosure["href"]
+                size = int(enclosure["length"])
+                break
         except Exception:
             pass
+    else:
+        link = entry.link
+        if not link:
+            link = entry.links[0].href
 
     # GUID usually has URL to result on page
     infourl = None
@@ -717,7 +724,7 @@ def _get_link(entry):
 
         return link, infourl, category, size, age, season, episode
     else:
-        logging.warning(T("Empty RSS entry found (%s)"), link)
+        logging.info(T("Empty RSS entry found (%s)"), link)
         return None, None, "", 0, None, 0, 0
 
 

tests/data/rss_enclosure_multiple.xml

@@ -0,0 +1,16 @@
+<?xml version="1.0" encoding="utf-8"?>
+<rss version="2.0">
+  <channel>
+    <title>My RSS Feed</title>
+    <link>https://sabnzbd.org/</link>
+    <item>
+      <title>TITLE</title>
+      <link>http://LINK</link>
+      <comments>COMMENTS</comments>
+      <enclosure url="http://TORRENT_LINK" type="application/x-bittorrent" length="100" />
+      <enclosure url="http://NZB_LINK" type="application/x-nzb" length="200" />
+      <pubDate>Tue, 20 May 2025 18:21:01 +0000</pubDate>
+      <guid isPermaLink="true">https://sabnzbd.org/rss_enclosure_multiple</guid>
+    </item>
+  </channel>
+</rss>

tests/data/rss_enclosure_no_nzb.xml

@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="utf-8"?>
+<rss version="2.0">
+  <channel>
+    <title>My RSS Feed</title>
+    <link>https://sabnzbd.org/</link>
+    <item>
+      <title>TITLE</title>
+      <link>http://LINK</link>
+      <comments>COMMENTS</comments>
+      <enclosure url="http://TORRENT_LINK" type="application/x-bittorrent" length="100" />
+      <pubDate>Tue, 20 May 2025 18:21:01 +0000</pubDate>
+      <guid isPermaLink="true">https://sabnzbd.org/rss_enclosure_no_nzb</guid>
+    </item>
+  </channel>
+</rss>

tests/data/rss_link.xml

@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="utf-8"?>
+<rss version="2.0">
+  <channel>
+    <title>My RSS Feed</title>
+    <link>https://sabnzbd.org/</link>
+    <item>
+      <title>TITLE</title>
+      <link>http://LINK</link>
+      <comments>COMMENTS</comments>
+      <description><![CDATA[<strong>Total Size</strong>: 200.0 B]]></description>
+      <pubDate>Tue, 20 May 2025 18:21:01 +0000</pubDate>
+      <guid isPermaLink="true">https://sabnzbd.org/rss_link</guid>
+    </item>
+  </channel>
+</rss>

tests/test_rss.py

@@ -21,9 +21,11 @@ tests.test_misc - Testing functions in misc.py
 import datetime
 import time
 import configobj
+from pytest_httpserver import HTTPServer
 
 import sabnzbd.rss as rss
 import sabnzbd.config
+from tests.testhelper import httpserver_handler_data_dir
 
 
 class TestRSS:
@@ -97,3 +99,67 @@ class TestRSS:
         # of the system, so now we have to return to UTC
         adjusted_date = datetime.datetime(2019, 3, 2, 17, 18, 7) - datetime.timedelta(seconds=time.timezone)
         assert job_data["age"] == adjusted_date
+
+    def test_rss_link(self, httpserver: HTTPServer):
+        httpserver.expect_request("/rss_link.xml").respond_with_handler(httpserver_handler_data_dir)
+
+        feed_name = "TestFeedLink"
+        self.setup_rss(feed_name, httpserver.url_for("/rss_link.xml"))
+
+        # Start the RSS reader
+        rss_obj = rss.RSSReader()
+        rss_obj.run_feed(feed_name)
+
+        # Is the feed processed?
+        assert feed_name in rss_obj.jobs
+        assert "http://LINK" in rss_obj.jobs[feed_name]
+
+        # Check some job-data
+        job_data = rss_obj.jobs[feed_name]["http://LINK"]
+        assert job_data["title"] == "TITLE"
+        assert job_data["infourl"] == "https://sabnzbd.org/rss_link"
+        assert job_data["size"] == 200
+
+        # feedparser returns UTC so SABnzbd converts to locale
+        # of the system, so now we have to return to UTC
+        adjusted_date = datetime.datetime(2025, 5, 20, 18, 21, 1) - datetime.timedelta(seconds=time.timezone)
+        assert job_data["age"] == adjusted_date
+
+    def test_rss_enclosure_no_nzb(self, httpserver: HTTPServer):
+        httpserver.expect_request("/rss_enclosure_no_nzb.xml").respond_with_handler(httpserver_handler_data_dir)
+
+        feed_name = "TestFeedEnclosureNoNZB"
+        self.setup_rss(feed_name, httpserver.url_for("/rss_enclosure_no_nzb.xml"))
+
+        # Start the RSS reader
+        rss_obj = rss.RSSReader()
+        rss_obj.run_feed(feed_name)
+
+        # Is the feed processed?
+        assert feed_name in rss_obj.jobs
+        assert not rss_obj.jobs[feed_name]
+
+    def test_rss_enclosure_multiple(self, httpserver: HTTPServer):
+        httpserver.expect_request("/rss_enclosure_multiple.xml").respond_with_handler(httpserver_handler_data_dir)
+
+        feed_name = "TestFeedEnclosureMultiple"
+        self.setup_rss(feed_name, httpserver.url_for("/rss_enclosure_multiple.xml"))
+
+        # Start the RSS reader
+        rss_obj = rss.RSSReader()
+        rss_obj.run_feed(feed_name)
+
+        # Is the feed processed?
+        assert feed_name in rss_obj.jobs
+        assert "http://NZB_LINK" in rss_obj.jobs[feed_name]
+
+        # Check some job-data
+        job_data = rss_obj.jobs[feed_name]["http://NZB_LINK"]
+        assert job_data["title"] == "TITLE"
+        assert job_data["infourl"] == "https://sabnzbd.org/rss_enclosure_multiple"
+        assert job_data["size"] == 200
+
+        # feedparser returns UTC so SABnzbd converts to locale
+        # of the system, so now we have to return to UTC
+        adjusted_date = datetime.datetime(2025, 5, 20, 18, 21, 1) - datetime.timedelta(seconds=time.timezone)
+        assert job_data["age"] == adjusted_date

tests/testhelper.py

@@ -36,6 +36,8 @@ from unittest import mock
 from urllib3.exceptions import ProtocolError
 import xmltodict
 import functools
+from werkzeug import Request
+from werkzeug.utils import send_from_directory
 
 import sabnzbd
 import sabnzbd.cfg as cfg
@@ -164,6 +166,11 @@ def create_and_read_nzb_fp(nzbdir: str, metadata: Optional[Dict[str, str]] = Non
     return io.BytesIO(nzb_data)
 
 
+def httpserver_handler_data_dir(request: Request):
+    """Respond to a httpserver request with a file in SAB_DATA_DIR"""
+    return send_from_directory(directory=SAB_DATA_DIR, path=request.path.lstrip("/"), environ=request.environ)
+
+
 def random_name(length: int = 16) -> str:
     """Shorthand to create a simple random string"""
     return "".join(choice(ascii_lowercase + digits) for _ in range(length))