Merge pull request #36 from kitconcept/fhnw-develop-keyword-extra-condition

Add extra conditions for string matching

Author: reekitconcept

src/kitconcept/solr/services/solr_utils_extra.py

@@ -74,6 +74,33 @@ def query_list(self):
                     result += f"{value}}}"
                 else:
                     result += "*]"
+            elif kind == "string":
+                keys = set(condition.keys())
+                if not keys.issubset({"in"}):
+                    raise RuntimeError(
+                        f"invalid keys in options for condition 'string', supported: 'in' [{keys}]"
+                    )
+                if "in" in condition:
+                    if type(condition["in"]) is not list:
+                        raise RuntimeError(
+                            f"invalid type for condition 'string' [{type(condition['in'])}]"
+                        )
+                    if len(condition["in"]) == 0:
+                        # Empty list, ignore
+                        continue
+                    result = f"{fieldname}:"
+                    result += (
+                        "("
+                        + " OR ".join(
+                            list(
+                                map(
+                                    lambda term: replace_reserved(term),
+                                    condition["in"],
+                                )
+                            )
+                        )
+                        + ")"
+                    )
             else:
                 raise RuntimeError(f"Wrong condition type [{kind}]")
             results.append(result)

tests/services/extra_conditions/conftest.py

@@ -14,13 +14,15 @@ def contents() -> List:
             "subjects": ["mymembersubject", "mymembersubjecttwo"],
             "_image": b"iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAAAXNSR0IArs4c6QAAAA1JREFUGFdjCDO+/R8ABKsCZD++CcMAAAAASUVORK5CYII=",  # noQA
             "language": "en",
+            "searchwords": ["term1", "term2"],
         },
         {
             "_container": "",
             "type": "Document",
             "id": "myfolder",
             "title": "My Folder to store everything about Noam Chomsky",
             "language": "en",
+            "searchwords": ["term1", "term2"],
         },
         {
             "_container": "myfolder",
@@ -37,6 +39,7 @@ def contents() -> List:
             "description": "Chomsky alpha",
             "language": "en",
             "start": "2021-01-01T00:00:00",
+            "searchwords": ["term1"],
         },
         {
             "_container": "",
@@ -46,6 +49,7 @@ def contents() -> List:
             "description": "Chomsky beta",
             "language": "en",
             "start": "2021-02-01T00:00:00",
+            "searchwords": ["term1"],
         },
         {
             "_container": "",
@@ -55,6 +59,7 @@ def contents() -> List:
             "description": "Chomsky gamma",
             "language": "en",
             "start": "2021-03-01T00:00:00",
+            "searchwords": ["term1"],
         },
         {
             "_container": "",
@@ -64,6 +69,7 @@ def contents() -> List:
             "description": "Chomsky alpha",
             "language": "en",
             "start": "2021-01-01T00:00:00",
+            "searchwords": ["term2"],
         },
         {
             "_container": "",
@@ -73,6 +79,7 @@ def contents() -> List:
             "description": "Chomsky beta",
             "language": "en",
             "start": "2021-02-01T00:00:00",
+            "searchwords": ["term2"],
         },
         {
             "_container": "",
@@ -82,5 +89,6 @@ def contents() -> List:
             "description": "Chomsky gamma",
             "language": "en",
             "start": "2021-03-01T00:00:00",
+            "searchwords": ["term2"],
         },
     ]

tests/services/extra_conditions/test_extra_conditions.py

@@ -285,3 +285,119 @@ def test_facet_fields(self):
     def test_paths(self, all_path_string, path: str, expected: bool):
         path_strings = all_path_string(self.data)
         assert (path in path_strings) is expected
+
+
+class TestExtraConditionsStringInSingleTerm(TestEndpointCustom):
+    extra_conditions = [
+        [
+            "searchwords",
+            "string",
+            {"in": ["term1"]},
+        ],
+    ]
+    url = f"/@solr?q=chomsky&group_select=1&extra_conditions={encoded(extra_conditions)}"
+
+    @pytest.mark.parametrize(
+        "path,expected",
+        [
+            ("/plone/myimage", False),
+            ("/plone/myfolder", True),
+            ("/plone/myfolder/mynews", False),
+            ("/plone/foo_alpha", True),
+            ("/plone/foo_beta", True),
+            ("/plone/foo_gamma", True),
+            ("/plone/bar_alpha", False),
+            ("/plone/bar_beta", False),
+            ("/plone/bar_gamma", False),
+        ],
+    )
+    def test_paths(self, all_path_string, path: str, expected: bool):
+        path_strings = all_path_string(self.data)
+        assert (path in path_strings) is expected
+
+
+class TestExtraConditionsStringInSingleTerm2(TestEndpointCustom):
+    extra_conditions = [
+        [
+            "searchwords",
+            "string",
+            {"in": ["term2"]},
+        ],
+    ]
+    url = f"/@solr?q=chomsky&group_select=1&extra_conditions={encoded(extra_conditions)}"
+
+    @pytest.mark.parametrize(
+        "path,expected",
+        [
+            ("/plone/myimage", False),
+            ("/plone/myfolder", True),
+            ("/plone/myfolder/mynews", False),
+            ("/plone/foo_alpha", False),
+            ("/plone/foo_beta", False),
+            ("/plone/foo_gamma", False),
+            ("/plone/bar_alpha", True),
+            ("/plone/bar_beta", True),
+            ("/plone/bar_gamma", True),
+        ],
+    )
+    def test_paths(self, all_path_string, path: str, expected: bool):
+        path_strings = all_path_string(self.data)
+        assert (path in path_strings) is expected
+
+
+class TestExtraConditionsStringInMultipleTerms(TestEndpointCustom):
+    extra_conditions = [
+        [
+            "searchwords",
+            "string",
+            {"in": ["term1", "term2"]},
+        ],
+    ]
+    url = f"/@solr?q=chomsky&group_select=1&extra_conditions={encoded(extra_conditions)}"
+
+    @pytest.mark.parametrize(
+        "path,expected",
+        [
+            ("/plone/myimage", False),
+            ("/plone/myfolder", True),
+            ("/plone/myfolder/mynews", False),
+            ("/plone/foo_alpha", True),
+            ("/plone/foo_beta", True),
+            ("/plone/foo_gamma", True),
+            ("/plone/bar_alpha", True),
+            ("/plone/bar_beta", True),
+            ("/plone/bar_gamma", True),
+        ],
+    )
+    def test_paths(self, all_path_string, path: str, expected: bool):
+        path_strings = all_path_string(self.data)
+        assert (path in path_strings) is expected
+
+
+class TestExtraConditionsStringInNoTerms(TestEndpointCustom):
+    extra_conditions = [
+        [
+            "searchwords",
+            "string",
+            {"in": []},
+        ],
+    ]
+    url = f"/@solr?q=chomsky&group_select=1&extra_conditions={encoded(extra_conditions)}"
+
+    @pytest.mark.parametrize(
+        "path,expected",
+        [
+            ("/plone/myimage", False),
+            ("/plone/myfolder", True),
+            ("/plone/myfolder/mynews", False),
+            ("/plone/foo_alpha", True),
+            ("/plone/foo_beta", True),
+            ("/plone/foo_gamma", True),
+            ("/plone/bar_alpha", True),
+            ("/plone/bar_beta", True),
+            ("/plone/bar_gamma", True),
+        ],
+    )
+    def test_paths(self, all_path_string, path: str, expected: bool):
+        path_strings = all_path_string(self.data)
+        assert (path in path_strings) is expected

tests/utils/test_utils_extra_conditions.py

@@ -122,3 +122,29 @@ def test_query_list_invalid_combination_2(self):
         obj = SolrExtraConditions(config)
         with pytest.raises(RuntimeError):
             obj.query_list()
+
+
+class TestUtilsExtraConditionsString:
+    def test_query_list_string_in_single_term(self):
+        config = [["keywords", "string", {"in": ["term1"]}]]
+        obj = SolrExtraConditions(config)
+        result = obj.query_list()
+        assert result == ["keywords:(term1)"]
+
+    def test_query_list_string_in_multiple_terms(self):
+        config = [["keywords", "string", {"in": ["term1", "term2"]}]]
+        obj = SolrExtraConditions(config)
+        result = obj.query_list()
+        assert result == ["keywords:(term1 OR term2)"]
+
+    def test_query_list_string_in_empty_term(self):
+        config = [["keywords", "string", {"in": []}]]
+        obj = SolrExtraConditions(config)
+        result = obj.query_list()
+        assert result == []
+
+    def test_query_list_string_in_invalid_term_type(self):
+        config = [["keywords", "string", {"in": "NotAList"}]]
+        obj = SolrExtraConditions(config)
+        with pytest.raises(RuntimeError):
+            obj.query_list()