Add more tests

tests/web/test_admin_annotations.py

@@ -196,3 +196,121 @@ class TestAnnotationModel:
         assert 0 <= ann.y_center <= 1
         assert 0 <= ann.width <= 1
         assert 0 <= ann.height <= 1
+
+
+class TestAutoLabelFilePathResolution:
+    """Tests for auto-label file path resolution.
+
+    The auto-label endpoint needs to resolve the storage path (e.g., "raw_pdfs/uuid.pdf")
+    to an actual filesystem path via the storage helper.
+    """
+
+    def test_extracts_filename_from_storage_path(self):
+        """Test that filename is extracted from storage path correctly."""
+        # Storage paths are like "raw_pdfs/uuid.pdf"
+        storage_path = "raw_pdfs/550e8400-e29b-41d4-a716-446655440000.pdf"
+
+        # The annotation endpoint extracts filename
+        filename = storage_path.split("/")[-1] if "/" in storage_path else storage_path
+
+        assert filename == "550e8400-e29b-41d4-a716-446655440000.pdf"
+
+    def test_handles_path_without_prefix(self):
+        """Test that paths without prefix are handled."""
+        storage_path = "550e8400-e29b-41d4-a716-446655440000.pdf"
+
+        filename = storage_path.split("/")[-1] if "/" in storage_path else storage_path
+
+        assert filename == "550e8400-e29b-41d4-a716-446655440000.pdf"
+
+    def test_storage_helper_resolves_path(self):
+        """Test that storage helper can resolve the path."""
+        from pathlib import Path
+        from unittest.mock import MagicMock, patch
+
+        # Mock storage helper
+        mock_storage = MagicMock()
+        mock_path = Path("/storage/raw_pdfs/test.pdf")
+        mock_storage.get_raw_pdf_local_path.return_value = mock_path
+
+        with patch(
+            "inference.web.services.storage_helpers.get_storage_helper",
+            return_value=mock_storage,
+        ):
+            from inference.web.services.storage_helpers import get_storage_helper
+
+            storage = get_storage_helper()
+            result = storage.get_raw_pdf_local_path("test.pdf")
+
+            assert result == mock_path
+            mock_storage.get_raw_pdf_local_path.assert_called_once_with("test.pdf")
+
+    def test_auto_label_request_validation(self):
+        """Test AutoLabelRequest validates field_values."""
+        # Valid request
+        request = AutoLabelRequest(
+            field_values={"InvoiceNumber": "12345"},
+            replace_existing=False,
+        )
+        assert request.field_values == {"InvoiceNumber": "12345"}
+
+        # Empty field_values should be valid at schema level
+        # (endpoint validates non-empty)
+        request_empty = AutoLabelRequest(
+            field_values={},
+            replace_existing=False,
+        )
+        assert request_empty.field_values == {}
+
+
+class TestMatchClassAttributes:
+    """Tests for Match class attributes used in auto-labeling.
+
+    The autolabel service uses Match objects from FieldMatcher.
+    Verifies the correct attribute names are used.
+    """
+
+    def test_match_has_matched_text_attribute(self):
+        """Test that Match class has matched_text attribute (not matched_value)."""
+        from shared.matcher.models import Match
+
+        # Create a Match object
+        match = Match(
+            field="invoice_number",
+            value="12345",
+            bbox=(100, 100, 200, 150),
+            page_no=0,
+            score=0.95,
+            matched_text="INV-12345",
+            context_keywords=["faktura", "nummer"],
+        )
+
+        # Verify matched_text exists (this is what autolabel.py should use)
+        assert hasattr(match, "matched_text")
+        assert match.matched_text == "INV-12345"
+
+        # Verify matched_value does NOT exist
+        # This was the bug - autolabel.py was using matched_value instead of matched_text
+        assert not hasattr(match, "matched_value")
+
+    def test_match_attributes_for_annotation_creation(self):
+        """Test that Match has all attributes needed for annotation creation."""
+        from shared.matcher.models import Match
+
+        match = Match(
+            field="amount",
+            value="1000.00",
+            bbox=(50, 200, 150, 230),
+            page_no=0,
+            score=0.88,
+            matched_text="1 000,00",
+            context_keywords=["att betala", "summa"],
+        )
+
+        # These are all the attributes used in autolabel._create_annotations_from_matches
+        assert hasattr(match, "bbox")
+        assert hasattr(match, "matched_text")  # NOT matched_value
+        assert hasattr(match, "score")
+
+        # Verify bbox format
+        assert len(match.bbox) == 4  # (x0, y0, x1, y1)