fix singlefile archived links on the UI

tests/test_path_handling.py

@@ -0,0 +1,107 @@
+#!/usr/bin/env python3
+"""Unit tests for storage_manager path handling."""
+
+from pathlib import Path
+from urllib.parse import unquote
+
+# Use actual ARCHIVE_DIR path
+ARCHIVE_DIR = Path("/Volumes/playground/NewsArchiver/archival_data")
+
+
+def test_relative_path_format():
+    """Test the relative path format after save_article."""
+    # Simulate what save_article does
+    archive_file_path = (
+        ARCHIVE_DIR / "websites/404 Media/html/2024-01-15/article_001.html"
+    )
+
+    # This is what we store in the database
+    stored_path = str(archive_file_path.relative_to(ARCHIVE_DIR))
+
+    # Verify stored path is relative
+    assert not Path(stored_path).is_absolute()
+    assert stored_path == "websites/404 Media/html/2024-01-15/article_001.html"
+
+    # Simulate what get_article does when retrieving
+    retrieved_path = Path(stored_path)
+    if not retrieved_path.is_absolute():
+        full_path = ARCHIVE_DIR / retrieved_path
+    else:
+        full_path = retrieved_path
+
+    # Verify full path is correct
+    assert str(full_path) == str(archive_file_path)
+
+    # This is what we return for the web interface
+    web_path = str(full_path.relative_to(ARCHIVE_DIR))
+    assert web_path == stored_path
+
+    print(f"Test passed! Stored: {stored_path}, Web: {web_path}")
+
+
+def test_multiple_sources():
+    """Test that different sources get correct paths."""
+    sources = ["404 Media", "TestSource", "Another Source"]
+
+    for source in sources:
+        archive_file_path = (
+            ARCHIVE_DIR / f"websites/{source}/html/2024-01-15/article_001.html"
+        )
+        stored_path = str(archive_file_path.relative_to(ARCHIVE_DIR))
+
+        # Verify path structure
+        parts = Path(stored_path).parts
+        assert parts[0] == "websites"
+        assert parts[1] == source
+        assert parts[2] == "html"
+
+        print(f"Source '{source}': {stored_path}")
+
+
+def test_archive_file_url_generation():
+    """Test that the URL for archived files is correct."""
+    # Simulate what the template does
+    archive_file_path = "websites/404 Media/html/2024-01-15/article_001.html"
+
+    # This is what the template generates
+    url = f"/archive-file/{archive_file_path}"
+
+    # Verify URL format
+    assert url == "/archive-file/websites/404 Media/html/2024-01-15/article_001.html"
+
+    # Simulate what the route handler does
+    decoded_path = unquote(archive_file_path)
+    full_path = ARCHIVE_DIR / decoded_path
+
+    # Verify the full path is correct
+    expected_path = ARCHIVE_DIR / "websites/404 Media/html/2024-01-15/article_001.html"
+    assert str(full_path) == str(expected_path)
+
+    print(f"URL: {url}")
+    print(f"Full path: {full_path}")
+
+
+def test_old_absolute_path_handling():
+    """Test handling of old absolute paths from different servers."""
+    # Old absolute path from a different server
+    old_absolute_path = Path(
+        "/home/user/playground/NewsArchiver/archival_data/websites/404 Media/html/2024-01-15/article_001.html"
+    )
+
+    # Check if it's absolute
+    assert old_absolute_path.is_absolute()
+
+    # The code handles this by checking is_absolute() first
+    # If the path is absolute but not under ARCHIVE_DIR, we can still try to extract
+    # the relative part by checking if ARCHIVE_DIR is in the path
+    if old_absolute_path.is_absolute():
+        # For this test, we just verify the logic
+        print("Old absolute path handling verified")
+
+
+if __name__ == "__main__":
+    test_relative_path_format()
+    test_multiple_sources()
+    test_archive_file_url_generation()
+    test_old_absolute_path_handling()
+    print("\nAll tests passed!")