feat: RSS fetcher with dedup, image download, HTML rewriting

Made-with: Cursor

tests/test_fetcher.py

@@ -0,0 +1,83 @@
+import json
+from unittest.mock import patch, MagicMock
+
+from src.fetcher import fetch_and_cache_articles
+from src.models import Article, Image
+from tests.conftest import SAMPLE_RSS_XML
+
+
+def _mock_feed_response(xml_content):
+    mock = MagicMock()
+    mock.content = xml_content.encode("utf-8")
+    mock.text = xml_content
+    mock.status_code = 200
+    mock.raise_for_status = MagicMock()
+    return mock
+
+
+def test_fetch_creates_articles(app, db):
+    with app.app_context():
+        with patch("src.fetcher.requests.get") as mock_get:
+            mock_get.return_value = _mock_feed_response(SAMPLE_RSS_XML)
+            with patch("src.fetcher.process_image") as mock_img:
+                mock_img.return_value = ("/fake/path.jpg", 800, 450)
+                result = fetch_and_cache_articles()
+
+        assert result["new"] == 2
+        assert result["skipped"] == 0
+        articles = Article.query.order_by(Article.pub_date).all()
+        assert len(articles) == 2
+        assert articles[0].title == "Test Article One"
+        assert articles[1].title == "Test Article Two"
+
+
+def test_fetch_deduplicates(app, db):
+    with app.app_context():
+        with patch("src.fetcher.requests.get") as mock_get:
+            mock_get.return_value = _mock_feed_response(SAMPLE_RSS_XML)
+            with patch("src.fetcher.process_image") as mock_img:
+                mock_img.return_value = ("/fake/path.jpg", 800, 450)
+                fetch_and_cache_articles()
+                result = fetch_and_cache_articles()
+
+        assert result["new"] == 0
+        assert result["skipped"] == 2
+        assert Article.query.count() == 2
+
+
+def test_fetch_downloads_images(app, db):
+    with app.app_context():
+        with patch("src.fetcher.requests.get") as mock_get:
+            mock_get.return_value = _mock_feed_response(SAMPLE_RSS_XML)
+            with patch("src.fetcher.process_image") as mock_img:
+                mock_img.return_value = ("/fake/path.jpg", 800, 450)
+                fetch_and_cache_articles()
+
+        images = Image.query.all()
+        assert len(images) == 1
+        assert images[0].original_url == "https://example.com/image1.jpg"
+
+
+def test_fetch_rewrites_image_src(app, db):
+    with app.app_context():
+        with patch("src.fetcher.requests.get") as mock_get:
+            mock_get.return_value = _mock_feed_response(SAMPLE_RSS_XML)
+            with patch("src.fetcher.process_image") as mock_img:
+                mock_img.return_value = ("/fake/path.jpg", 800, 450)
+                fetch_and_cache_articles()
+
+        article = Article.query.filter_by(
+            guid="https://example.com/?p=1001"
+        ).first()
+        assert "https://example.com/image1.jpg" not in article.content_html
+        assert "/fake/path.jpg" in article.content_html
+
+
+def test_fetch_handles_feed_error(app, db):
+    with app.app_context():
+        with patch("src.fetcher.requests.get") as mock_get:
+            mock_get.side_effect = Exception("Network error")
+            result = fetch_and_cache_articles()
+
+        assert result["error"] is not None
+        assert Article.query.count() == 0