test: add unit tests for URI replacements and resource clearing

tests/test_format_fidelity_checker.py

@@ -29,10 +29,13 @@ class TestDcatCatalogCheck(unittest.TestCase):
             "XML": ["application/xml"],
         }
 
-        self.assertTrue(self.dcc.is_mime_type_compatible("JSON", "application/json"))
-        self.assertFalse(self.dcc.is_mime_type_compatible("JSON", "application/xml"))
+        self.assertTrue(self.dcc.is_mime_type_compatible(
+            "JSON", "application/json"))
+        self.assertFalse(self.dcc.is_mime_type_compatible(
+            "JSON", "application/xml"))
         self.assertFalse(
-            self.dcc.is_mime_type_compatible("UnknownFormat", "application/json")
+            self.dcc.is_mime_type_compatible(
+                "UnknownFormat", "application/json")
         )
 
     def test_read_allowed_file_formats(self):
@@ -44,7 +47,8 @@ class TestDcatCatalogCheck(unittest.TestCase):
         ):
             formats = self.dcc.read_allowed_file_formats()
             self.assertEqual(
-                formats, {"JSON": ["application/json"], "XML": ["application/xml"]}
+                formats, {"JSON": ["application/json"],
+                          "XML": ["application/xml"]}
             )
 
     def test_load_uri_replacements(self):
@@ -55,7 +59,8 @@ class TestDcatCatalogCheck(unittest.TestCase):
             ),
         ):
             replacements = self.dcc.load_uri_replacements()
-            self.assertEqual(replacements, [{"regex": "old", "replaced_by": "new"}])
+            self.assertEqual(
+                replacements, [{"regex": "old", "replaced_by": "new"}])
 
     # Simulate that the file does not exist
 
@@ -375,6 +380,82 @@ class TestDcatCatalogCheck(unittest.TestCase):
             'Line 1 is missing \'url\': {"status": "valid", "format": "JSON"}'
         )
 
+    def test_apply_uri_replacements(self):
+        """Test the apply_uri_replacements method."""
+        # Setup URI replacements
+        self.dcc.uri_replacements = [
+            {"regex": r"example\.com", "replaced_by": "test.com"},
+            {"regex": r"http://", "replaced_by": "https://"},
+        ]
+
+        # URL matching both replacements
+        url = "http://example.com/path"
+        result = self.dcc.apply_uri_replacements(url)
+        self.assertEqual(result, "https://test.com/path")
+
+        # URL matching only one replacement
+        url = "http://other.com/path"
+        result = self.dcc.apply_uri_replacements(url)
+        self.assertEqual(result, "https://other.com/path")
+
+        # URL with no matches
+        url = "https://unchanged.com/path"
+        result = self.dcc.apply_uri_replacements(url)
+        self.assertEqual(result, "https://unchanged.com/path")
+
+        # Empty URL
+        url = ""
+        result = self.dcc.apply_uri_replacements(url)
+        self.assertEqual(result, "")
+
+        # No URI replacements defined
+        self.dcc.uri_replacements = []
+        url = "http://example.com/path"
+        result = self.dcc.apply_uri_replacements(url)
+        self.assertEqual(result, "http://example.com/path")
+
+    def test_clear_result(self):
+        """Test the _clear_result method."""
+        # Define a resource dictionary with keys to clear and some additional keys
+        resource = {
+            "accessible": True,
+            "checksum_ok": True,
+            "duration": 1.23,
+            "error": "Some error",
+            "etag": "some-etag",
+            "http_status": 200,
+            "last_check": "2024-12-27T12:34:56Z",
+            "mimetype": "application/json",
+            "mimetype_mismatch": False,
+            "valid": True,
+            "url": "http://example.com/data",  # This key should remain untouched
+            "format": "JSON",  # This key should remain untouched
+        }
+
+        # Call the _clear_result method
+        self.dcc._clear_result(resource)
+
+        # Check that all keys to clear have been removed
+        for key in [
+            "accessible",
+            "checksum_ok",
+            "duration",
+            "error",
+            "etag",
+            "http_status",
+            "last_check",
+            "mimetype",
+            "mimetype_mismatch",
+            "valid",
+        ]:
+            self.assertNotIn(key, resource)
+
+        # Check that unrelated keys remain
+        self.assertIn("url", resource)
+        self.assertIn("format", resource)
+        self.assertEqual(resource["url"], "http://example.com/data")
+        self.assertEqual(resource["format"], "JSON")
+
 
 if __name__ == "__main__":
     unittest.main()