Merge pull request #5429 from hypothesis/normalize-uris-in-websocket-…

…filtering

Normalize URIs for comparison when filtering Web Socket messages

h/streamer/filter.py

@@ -5,6 +5,7 @@
 import unicodedata
 
 from jsonpointer import resolve_pointer
+from h.util.uri import normalize as normalize_uri
 from h._compat import text_type
 
 SCHEMA = {
@@ -50,15 +51,25 @@ def evaluate_clause(self, clause, target):
 
         filter_term = clause['value']
 
+        def normalize(term):
+            # Apply generic normalization.
+            normalized = uni_fold(term)
+
+            # Apply field-specific normalization.
+            if clause['field'] == '/uri':
+                normalized = normalize_uri(term)
+
+            return normalized
+
         if isinstance(filter_term, list):
-            filter_term = [t for t in uni_fold(filter_term)]
+            filter_term = [normalize(t) for t in filter_term]
         else:
-            filter_term = uni_fold(filter_term)
+            filter_term = normalize(filter_term)
 
         if isinstance(field_value, list):
-            field_value = [v for v in uni_fold(field_value)]
+            field_value = [normalize(v) for v in field_value]
         else:
-            field_value = uni_fold(field_value)
+            field_value = normalize(field_value)
 
         if clause['operator'] == 'one_of':
             # The `one_of` operator behaves differently depending on whether

tests/h/streamer/filter_test.py

@@ -1,28 +1,38 @@
 from __future__ import unicode_literals
 
+import pytest
+
 from h.streamer.filter import FilterHandler
 
 
 class TestFilterHandler(object):
-    def test_it_matches_uri(self):
+    @pytest.mark.parametrize('query_uris,ann_uri,should_match', [
+        # Test cases that require only exact comparisons.
+        (["https://example.com", "https://example.org"], 'https://example.com', True),
+        (["https://example.com", "https://example.org"], 'https://example.net', False),
+
+        # Test cases that require comparison of normalized URIs.
+        (["https://example.com"], "http://example.com", True),
+        (["http://example.com"], "https://example.com", True),
+        (["http://example.com/?"], "https://example.com", True),
+        (["http://example.com"], "https://example.com/?", True),
+    ])
+    def test_it_matches_uri(self, query_uris, ann_uri, should_match):
         query = {
             "match_policy": "include_any",
             "actions": {},
             "clauses": [
                 {
                     "field": "/uri",
                     "operator": "one_of",
-                    "value": ["https://example.com", "https://example.org"],
+                    "value": query_uris,
                 }
             ],
         }
         handler = FilterHandler(query)
 
-        ann = {"id": "123", "uri": "https://example.com"}
-        assert handler.match(ann) is True
-
-        ann = {"id": "123", "uri": "https://example.net"}
-        assert handler.match(ann) is False
+        ann = {"id": "123", "uri": ann_uri}
+        assert handler.match(ann) is should_match
 
     def test_it_matches_id(self):
         query = {