feat: add external site parser worker

site_parser_worker/test_parser.py

@@ -0,0 +1,16 @@
+from app import parse_rss
+
+
+def test_parse_rss() -> None:
+    xml = """<rss><channel><item><guid>1</guid><title>Title</title>
+    <link>https://example.test/1</link>
+    <description>&lt;p&gt;Body&lt;/p&gt;&lt;img src="https://example.test/1.jpg"&gt;</description>
+    <pubDate>Mon, 10 Aug 2026 10:00:00 +0000</pubDate></item></channel></rss>"""
+    item = parse_rss(xml, 20)[0]
+    assert item.external_id == "1"
+    assert item.text == "Body"
+    assert item.media[0]["url"] == "https://example.test/1.jpg"
+
+
+if __name__ == "__main__":
+    test_parse_rss()