use elastic cross fields search by default (superdesk#527)

* use elastic cross fields search by default

CPCN-316

newsroom/search/service.py

@@ -1,5 +1,5 @@
 import logging
-from typing import List, Optional, Union, Dict, Any, TypedDict
+from typing import List, Literal, Optional, Union, Dict, Any, TypedDict
 from copy import deepcopy
 
 from flask import current_app as app, json, abort
@@ -41,15 +41,22 @@ def strtobool(val):
     return _strtobool(val)
 
 
-def query_string(query, default_operator="AND", fields=["*"]):
+def query_string(
+    query: str,
+    default_operator: Literal["AND", "OR"] = "AND",
+    fields: List[str] = ["*"],
+    multimatch_type: Literal["cross_fields", "best_fields"] = "cross_fields",
+    analyze_wildcard=False,
+):
     query_string_settings = app.config["ELASTICSEARCH_SETTINGS"]["settings"]["query_string"]
     return {
         "query_string": {
             "query": query,
             "default_operator": default_operator,
-            "analyze_wildcard": query_string_settings["analyze_wildcard"],
+            "analyze_wildcard": query_string_settings["analyze_wildcard"] or analyze_wildcard,
             "lenient": True,
             "fields": fields,
+            "type": multimatch_type,
         }
     }
 
@@ -772,28 +779,30 @@ def apply_request_advanced_search(self, search: SearchQuery):
         if not fields:
             return
 
-        def gen_advanced_query(keywords: str, operator: str, multi_match_type: str):
-            return {
-                "query_string": {
-                    "query": keywords,
-                    "fields": fields,
-                    "default_operator": operator,
-                    "type": multi_match_type,
-                    "lenient": True,
-                    "analyze_wildcard": True,
-                },
-            }
-
         if search.advanced.get("all"):
             search.query["bool"].setdefault("must", []).append(
-                gen_advanced_query(search.advanced["all"], "AND", "cross_fields")
+                query_string(
+                    search.advanced["all"], "AND", fields=fields, multimatch_type="cross_fields", analyze_wildcard=True
+                )
             )
+
         if search.advanced.get("any"):
             search.query["bool"].setdefault("must", []).append(
-                gen_advanced_query(search.advanced["any"], "OR", "best_fields")
+                query_string(
+                    search.advanced["any"], "OR", fields=fields, multimatch_type="best_fields", analyze_wildcard=True
+                )
             )
+
         if search.advanced.get("exclude"):
-            search.query["bool"]["must_not"].append(gen_advanced_query(search.advanced["exclude"], "OR", "best_fields"))
+            search.query["bool"]["must_not"].append(
+                query_string(
+                    search.advanced["exclude"],
+                    "OR",
+                    fields=fields,
+                    multimatch_type="best_fields",
+                    analyze_wildcard=True,
+                )
+            )
 
     def apply_embargoed_filters(self, search):
         """Generate filters for embargoed params"""

tests/search/test_search_fields.py

@@ -1,3 +1,5 @@
+from flask import json
+from urllib.parse import quote
 from tests.utils import get_json
 
 
@@ -16,6 +18,24 @@ def test_wire_search_fields(client, app):
     assert 0 == len(data["_items"])
 
 
+def test_wire_search_cross_fields(client, app):
+    app.data.insert(
+        "items",
+        [
+            {"headline": "foo", "body_html": "bar", "type": "text"},
+        ],
+    )
+
+    data = get_json(client, "/wire/search?q=foo+bar")
+    assert 1 == len(data["_items"])
+
+    data = get_json(client, f'/wire/search?advanced={quote(json.dumps({"all": "foo bar"}))}')
+    assert 1 == len(data["_items"])
+
+    data = get_json(client, f'/wire/search?advanced={quote(json.dumps({"any": "foo bar"}))}')
+    assert 1 == len(data["_items"])
+
+
 def test_agenda_search_fields(client, app):
     app.data.insert(
         "agenda",

tests/search/test_search_filters.py

@@ -54,6 +54,7 @@ def test_apply_section_filter(client, app):
                 "analyze_wildcard": query_string_settings["analyze_wildcard"],
                 "lenient": True,
                 "fields": ["*"],
+                "type": "cross_fields",
             }
         } in search.query["bool"]["filter"]
 
@@ -68,6 +69,7 @@ def test_apply_section_filter(client, app):
                 "analyze_wildcard": query_string_settings["analyze_wildcard"],
                 "lenient": True,
                 "fields": ["*"],
+                "type": "cross_fields",
             }
         } in search.query["bool"]["filter"]
 
@@ -147,6 +149,7 @@ def assert_products_query(user_id, args=None, products=None):
                         "analyze_wildcard": query_string_settings["analyze_wildcard"],
                         "lenient": True,
                         "fields": app.config["WIRE_SEARCH_FIELDS"],
+                        "type": "cross_fields",
                     }
                 } in search.query["bool"]["should"]
 
@@ -182,6 +185,7 @@ def test_apply_request_filter__query_string(client, app):
                 "analyze_wildcard": query_string_settings["analyze_wildcard"],
                 "lenient": True,
                 "fields": app.config["WIRE_SEARCH_FIELDS"],
+                "type": "cross_fields",
             }
         } in search.query["bool"]["must"]
 
@@ -194,6 +198,7 @@ def test_apply_request_filter__query_string(client, app):
                 "analyze_wildcard": query_string_settings["analyze_wildcard"],
                 "lenient": True,
                 "fields": app.config["WIRE_SEARCH_FIELDS"],
+                "type": "cross_fields",
             }
         } in search.query["bool"]["must"]