fix name and description bm25 search (#160)

* fix name and description bm25 search

* bump version to 1.10.1

* filter out documents in passage tests

src/cpr_sdk/version.py

@@ -1,6 +1,6 @@
 _MAJOR = "1"
 _MINOR = "10"
-_PATCH = "0"
+_PATCH = "1"
 _SUFFIX = ""
 
 VERSION_SHORT = "{0}.{1}".format(_MAJOR, _MINOR)

src/cpr_sdk/yql_builder.py

@@ -60,8 +60,8 @@ def build_search_term(self) -> str:
             return """
                 (
                     {"targetHits": 1000} weakAnd(
-                        family_name contains(@query_string),
-                        family_description contains(@query_string),
+                        family_name_index contains(@query_string),
+                        family_description_index contains(@query_string),
                         text_block contains(@query_string)
                     )
                 )
@@ -71,8 +71,8 @@ def build_search_term(self) -> str:
                 (
                     (
                     {"targetHits": 1000} weakAnd(
-                        family_name contains(@query_string),
-                        family_description contains(@query_string),
+                        family_name_index contains(@query_string),
+                        family_description_index contains(@query_string),
                         text_block contains(@query_string)
                     )
                     ) or (

tests/test_search_adaptors.py

@@ -339,7 +339,11 @@ def test_vespa_search_adaptor__continuation_tokens__passages(test_vespa):
 
     # Collect family & hits for comparison later
     initial_family_id = initial_response.families[0].id
-    initial_passages = [h.text_block_id for h in initial_response.families[0].hits]
+    initial_passages = [
+        h.text_block_id
+        for h in initial_response.families[0].hits
+        if isinstance(h, Passage)
+    ]
 
     this_continuation = initial_response.this_continuation_token
     passage_continuation = initial_response.families[0].continuation_token
@@ -358,7 +362,9 @@ def test_vespa_search_adaptor__continuation_tokens__passages(test_vespa):
     assert response.families[0].id == initial_family_id
 
     # But Passages SHOULD have changed
-    new_passages = sorted([h.text_block_id for h in response.families[0].hits])
+    new_passages = sorted(
+        [h.text_block_id for h in response.families[0].hits if isinstance(h, Passage)]
+    )
     assert sorted(new_passages) != sorted(initial_passages)
 
     # Previous passage continuation gives initial results
@@ -370,7 +376,9 @@ def test_vespa_search_adaptor__continuation_tokens__passages(test_vespa):
     )
     response = vespa_search(test_vespa, request)
     assert response.families[0].id == initial_family_id
-    prev_passages = sorted([h.text_block_id for h in response.families[0].hits])
+    prev_passages = sorted(
+        [h.text_block_id for h in response.families[0].hits if isinstance(h, Passage)]
+    )
     assert sorted(prev_passages) != sorted(new_passages)
     assert sorted(prev_passages) == sorted(initial_passages)
 
@@ -426,10 +434,34 @@ def test_vespa_search_adaptor__continuation_tokens__families_and_passages(
 
     # All of these should have different passages from each other
     assert (
-        sorted([h.text_block_id for h in response_one.families[0].hits])
-        != sorted([h.text_block_id for h in response_two.families[0].hits])
-        != sorted([h.text_block_id for h in response_three.families[0].hits])
-        != sorted([h.text_block_id for h in response_four.families[0].hits])
+        sorted(
+            [
+                h.text_block_id
+                for h in response_one.families[0].hits
+                if isinstance(h, Passage)
+            ]
+        )
+        != sorted(
+            [
+                h.text_block_id
+                for h in response_two.families[0].hits
+                if isinstance(h, Passage)
+            ]
+        )
+        != sorted(
+            [
+                h.text_block_id
+                for h in response_three.families[0].hits
+                if isinstance(h, Passage)
+            ]
+        )
+        != sorted(
+            [
+                h.text_block_id
+                for h in response_four.families[0].hits
+                if isinstance(h, Passage)
+            ]
+        )
     )