Handle filters conrner case

deepset-ai · Nov 9, 2023 · f4c2963 · f4c2963
1 parent 4346f7c
commit f4c2963
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 3 deletions.
diff --git a/document_stores/elasticsearch/src/elasticsearch_haystack/filters.py b/document_stores/elasticsearch/src/elasticsearch_haystack/filters.py
@@ -81,7 +81,18 @@ def _parse_comparison(field: str, comparison: Union[Dict, List, str, float]) ->
                     raise FilterError(msg)
                 result.append({"range": {field: {comparator[1:]: val}}})
             elif comparator in ["$not", "$or"]:
-                result.append(_normalize_filters(val, comparator))
+                if isinstance(val, list):
+                    # This handles corner cases like this:
+                    # `{"name": {"$or": [{"$eq": "name_0"}, {"$eq": "name_1"}]}}`
+                    # If we don't handle it like this we'd lose the "name" field and the
+                    # generated query would be wrong and return unexpected results.
+                    comparisons = [_parse_comparison(field, v)[0] for v in val]
+                    if comparator == "$not":
+                        result.append({"bool": {"must_not": comparisons}})
+                    elif comparator == "$or":
+                        result.append({"bool": {"should": comparisons}})
+                else:
+                    result.append(_normalize_filters(val, comparator))
             elif comparator == "$and" and isinstance(val, list):
                 # We're assuming there are no duplicate items in the list
                 flat_filters = {k: v for d in val for k, v in d.items()}

diff --git a/document_stores/elasticsearch/tests/test_filters.py b/document_stores/elasticsearch/tests/test_filters.py
@@ -126,8 +126,8 @@
                             {
                                 "bool": {
                                     "should": [
-                                        {"match": {"$eq": {"query": "name_0", "minimum_should_match": "100%"}}},
-                                        {"match": {"$eq": {"query": "name_1", "minimum_should_match": "100%"}}},
+                                        {"term": {"name": "name_0"}},
+                                        {"term": {"name": "name_1"}},
                                     ]
                                 }
                             },
@@ -168,6 +168,21 @@
         {"number": {"$and": [{"$lte": 2}, {"$gte": 0}]}},
         {"bool": {"must": [{"range": {"number": {"lte": 2, "gte": 0}}}]}},
     ),
+    (
+        {"name": {"$or": [{"$eq": "name_0"}, {"$eq": "name_1"}]}},
+        {
+            "bool": {
+                "must": {
+                    "bool": {
+                        "should": [
+                            {"term": {"name": "name_0"}},
+                            {"term": {"name": "name_1"}},
+                        ]
+                    }
+                }
+            }
+        },
+    ),
 ]