Add initial version of feeds app

feeds/tests/__init__.py

@@ -0,0 +1 @@
+# This file marks the directory as a Python package.

feeds/tests/test_entry_id.py

@@ -0,0 +1,117 @@
+import os
+import threading
+from http.server import HTTPServer
+from http.server import SimpleHTTPRequestHandler
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+import pytest
+
+from feeds.models import Entry
+from feeds.models import Feed
+from feeds.services import fetch_and_archive_feed
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+
+@pytest.mark.django_db
+def test_entry_id_string_guid_dict(tmp_path: Path) -> None:
+    """Test that entry_id is always a string, even if guid is a dict."""
+    # Prepare a fake RSS feed with guid as dict (attributes)
+    feed_content = """
+    <rss version="2.0">
+      <channel>
+        <title>Test Feed</title>
+        <link>http://example.com/</link>
+        <description>Test feed description</description>
+        <item>
+          <title>Item 1</title>
+          <link>http://example.com/item1</link>
+          <guid isPermaLink="true">http://example.com/item1</guid>
+        </item>
+      </channel>
+    </rss>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+    assert entry.entry_id == "http://example.com/item1"
+    server.shutdown()
+
+
+@pytest.mark.django_db
+def test_entry_id_string_guid_string(tmp_path: Path) -> None:
+    """Test that entry_id is a string when guid is a plain string."""
+    feed_content = """
+    <rss version="2.0">
+      <channel>
+        <title>Test Feed</title>
+        <link>http://example.com/</link>
+        <description>Test feed description</description>
+        <item>
+          <title>Item 2</title>
+          <link>http://example.com/item2</link>
+          <guid>http://example.com/item2</guid>
+        </item>
+      </channel>
+    </rss>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+    assert entry.entry_id == "http://example.com/item2"
+    server.shutdown()
+
+
+@pytest.mark.django_db
+def test_entry_id_fallback_to_link(tmp_path: Path) -> None:
+    """Test that entry_id falls back to link if guid/id missing."""
+    feed_content = """
+    <rss version="2.0">
+      <channel>
+        <title>Test Feed</title>
+        <link>http://example.com/</link>
+        <description>Test feed description</description>
+        <item>
+          <title>Item 3</title>
+          <link>http://example.com/item3</link>
+        </item>
+      </channel>
+    </rss>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+    assert entry.entry_id == "http://example.com/item3"
+    server.shutdown()

feeds/tests/test_entry_id_extra.py

@@ -0,0 +1,111 @@
+import os
+import threading
+from http.server import HTTPServer
+from http.server import SimpleHTTPRequestHandler
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+import pytest
+
+from feeds.models import Entry
+from feeds.models import Feed
+from feeds.services import fetch_and_archive_feed
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+
+@pytest.mark.django_db
+def test_entry_id_id_dict(tmp_path: Path) -> None:
+    """Test that entry_id is a string when id is a dict."""
+    feed_content = """
+    <feed xmlns='http://www.w3.org/2005/Atom'>
+      <title>Test Atom Feed</title>
+      <id>http://example.com/feed</id>
+      <entry>
+        <title>Entry 1</title>
+        <id scheme='urn:uuid'>urn:uuid:1234</id>
+        <link href='http://example.com/entry1'/>
+      </entry>
+    </feed>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+    assert "urn:uuid:1234" in entry.entry_id
+    server.shutdown()
+
+
+@pytest.mark.django_db
+def test_entry_id_all_fields_missing(tmp_path: Path) -> None:
+    """Test that entry_id falls back to content_hash if guid/id/link missing."""
+    feed_content = """
+    <rss version='2.0'>
+      <channel>
+        <title>Test Feed</title>
+        <item>
+          <title>Item with no id</title>
+        </item>
+      </channel>
+    </rss>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+
+    # Should be a hash string (digits only)
+    assert entry.entry_id.isdigit() or entry.entry_id.lstrip("-").isdigit()
+    server.shutdown()
+
+
+@pytest.mark.django_db
+def test_entry_id_malformed_guid(tmp_path: Path) -> None:
+    """Test that entry_id handles malformed guid/id gracefully."""
+    feed_content = """
+    <rss version='2.0'>
+      <channel>
+        <title>Test Feed</title>
+        <item>
+          <title>Malformed guid</title>
+          <guid></guid>
+        </item>
+      </channel>
+    </rss>
+    """
+    feed_path: Path = tmp_path / "test_feed.xml"
+    feed_path.write_text(feed_content, encoding="utf-8")
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    fetch_and_archive_feed(feed)
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert isinstance(entry.entry_id, str)
+
+    # Should fallback to content_hash
+    assert entry.entry_id.isdigit() or entry.entry_id.lstrip("-").isdigit()
+    server.shutdown()

feeds/tests/test_services.py

@@ -0,0 +1,64 @@
+import os
+import threading
+from http.server import HTTPServer
+from http.server import SimpleHTTPRequestHandler
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+import pytest
+
+from feeds.models import Entry
+from feeds.models import Feed
+from feeds.services import fetch_and_archive_feed
+
+
+@pytest.mark.django_db
+def test_fetch_and_archive_feed_xml(tmp_path: Path) -> None:
+    """Test fetching and archiving a simple XML feed using a local HTTP server."""
+    # Use a local test XML file as a feed source
+    test_feed_path: Path = tmp_path / "test_feed.xml"
+    test_feed_path.write_text(
+        encoding="utf-8",
+        data="""
+        <rss version='2.0'>
+            <channel>
+                <title>Test Feed</title>
+                <link>http://example.com/</link>
+                <description>Test feed description</description>
+                <item>
+                    <title>Item 1</title>
+                    <link>http://example.com/item1</link>
+                    <description>Item 1 description</description>
+                </item>
+            </channel>
+        </rss>
+    """,
+    )
+
+    # Serve the file using a simple HTTP server
+    os.chdir(tmp_path)
+    server = HTTPServer(("localhost", 0), SimpleHTTPRequestHandler)
+    port: int = server.server_address[1]
+    thread = threading.Thread(target=server.serve_forever, daemon=True)
+    thread.start()
+    url: str = f"http://localhost:{port}/test_feed.xml"
+
+    feed: Feed = Feed.objects.create(url=url, domain="localhost")
+    new_entries: int = fetch_and_archive_feed(feed)
+    assert new_entries == 1
+
+    # Check that the entry was archived and contains the expected data
+    entry: Entry | None = Entry.objects.filter(feed=feed).first()
+    assert entry is not None
+    assert entry.data is not None
+    assert entry.data["title"] == "Item 1"
+    assert Entry.objects.filter(feed=feed).count() == 1
+
+    # Clean up: stop the server and wait for the thread to finish
+    server.shutdown()
+
+    # Wait until the thread terminates.
+    # This ensures the server is fully stopped before the test ends.
+    thread.join()