Merge pull request #575 from nationalarchives/1751-faceted-search

1751 - Faceted Search

CHANGELOG.md

@@ -5,15 +5,17 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog 1.0.0].
 
 ## Unreleased
+
 - Remove fcl_ex_id_ prefix from UUID of reparse execution ID
+- Implement handling of facets received from MarkLogic search results
 
 ## [Release 22.1.0]
 
 - Add an `enriched_recently` property
 - Add a `validates_against_schema` property
 - Add a `can_enrich` property
 - Only enrich if not recently enriched and valid against current schema
-- Allow fetching linked docuements for `Judgement`s and `PressSummary`s
+- Allow fetching linked documents for `Judgement`s and `PressSummary`s
 - Add function to check if the docx exists for a judgment
 
 ## [Release 22.0.2]

src/caselawclient/responses/search_response.py

@@ -45,3 +45,20 @@ def results(self) -> List[SearchResult]:
             "//search:response/search:result", namespaces=self.NAMESPACES
         )
         return [SearchResult(result, self.client) for result in results]
+
+    @property
+    def facets(self) -> dict[str, str]:
+        """
+        Returns search facets from the SearchResponse as a dictionary
+
+        :return: A flattened dictionary of search facet values
+        """
+        # TODO: preserve the name of the facet (e.g. "court", "year")
+        results = self.node.xpath(
+            "//search:response/search:facet/search:facet-value",
+            namespaces={"search": "http://marklogic.com/appservices/search"},
+        )
+        facets_dictionary = {
+            result.attrib["name"]: result.attrib["count"] for result in results
+        }
+        return facets_dictionary

tests/conftest.py

@@ -34,6 +34,18 @@ def valid_search_result_xml_fixture() -> str:
     )
 
 
+@pytest.fixture(name="valid_facets_fixture_xml")
+def valid_facets_xml_fixture() -> str:
+    return (
+        '<search:facet name="court" type="xs:string">'
+        '<search:facet-value name="" count="14"/>'
+        '<search:facet-value name=" UKUT-AAC" count="1"> UKUT-AAC</search:facet-value>'
+        '<search:facet-value name="EAT" count="649">EAT</search:facet-value>'
+        '<search:facet-value name="EWCA-Civil" count="5768">EWCA-Civil</search:facet-value>'
+        "</search:facet>"
+    )
+
+
 @pytest.fixture(name="generate_search_response_xml")
 def generate_search_response_xml_fixture() -> Callable:
     """
@@ -43,7 +55,7 @@ def generate_search_response_xml_fixture() -> Callable:
         Callable: Function that generates search response XML.
     """
 
-    def _generate_search_response_xml(response_content: str) -> str:
+    def _generate_search_response_xml(response_content: str, facets="") -> str:
         """
         Generate a search response XML string.
 
@@ -56,6 +68,7 @@ def _generate_search_response_xml(response_content: str) -> str:
         return (
             '<search:response xmlns:search="http://marklogic.com/appservices/search" total="2">'  # noqa: E501
             f"{response_content}"
+            f"{facets}"
             "</search:response>"
         )
 

tests/responses/test_search_response.py

@@ -79,3 +79,32 @@ def test_when_search_namespace_prefix_not_defined_on_response_xml_syntax_error_r
             match="Namespace prefix search on response is not defined",
         ):
             SearchResponse(etree.fromstring(xml_without_namespace), self.client)
+
+    def test_facets(
+        self,
+        valid_search_result_xml,
+        valid_facets_fixture_xml,
+        generate_search_response_xml,
+    ):
+        """
+        Given a SearchResponse instance with n facets
+        When calling 'facets' on it
+        Then it should return a list of n elements
+        And each element's node attribute should be as expected
+        """
+        search_response = SearchResponse(
+            etree.fromstring(
+                generate_search_response_xml(
+                    valid_search_result_xml, valid_facets_fixture_xml
+                )
+            ),
+            self.client,
+        )
+
+        results = search_response.facets
+
+        assert len(results) == 4
+        assert results[""] == "14"
+        assert results[" UKUT-AAC"] == "1"
+        assert results["EAT"] == "649"
+        assert results["EWCA-Civil"] == "5768"