Merge pull request #163 from MJedr/texkeys

fix texkey query

.github/workflows/build-and-release.yml

@@ -8,7 +8,7 @@ on:
 
 jobs:
   Test:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-20.04
 
     steps:
       - name: Checkout

inspire_query_parser/config.py

@@ -184,7 +184,7 @@
     't': 'title',
 
     # texkey
-    'texkey': 'texkeys',
+    'texkey': 'texkeys.raw',
 
     # Topcite, i.e. citation count
     # Cited used to be for Invenio style syntax while topcite for SPIRES

inspire_query_parser/visitors/elastic_search_visitor.py

@@ -122,6 +122,7 @@ class ElasticSearchVisitor(Visitor):
         'refersto': 'references.record.$ref',
         'reportnumber': 'report_numbers.value.fuzzy',
         'subject': 'facet_inspire_categories',
+        'texkey': 'texkeys.raw',
         'title': 'titles.full_title',
         'type-code': 'document_type',
         'topcite': 'citation_count',
@@ -807,6 +808,9 @@ def visit_value(self, node, fieldnames=None):
 
                 return generate_match_query(fieldnames, re.sub('ar[xX]iv:', "", node.value), with_operator_and=True)
 
+            elif self.KEYWORD_TO_ES_FIELDNAME['texkey'] == fieldnames:
+                return generate_match_query('texkeys.raw', node.value, with_operator_and=False)
+
             elif fieldnames not in self.KEYWORD_TO_ES_FIELDNAME.values():
                 colon_value = ':'.join([fieldnames, node.value])
                 given_field_query = generate_match_query(fieldnames, node.value, with_operator_and=True)

tests/test_elastic_search_visitor.py

@@ -3235,3 +3235,50 @@ def test_elastic_search_visitor_complex_query():
     }
     generated_es_query = _parse_query(query_str)
     assert generated_es_query == expected_es_query
+
+
+def test_elastic_search_visitor_texkeys_regression():
+    query_str = "texkey Chen:2014cwa"
+    expected_es_query = {
+        "match": {
+            "texkeys.raw": "Chen:2014cwa"
+        }
+    }
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query
+
+
+def test_elastic_search_visitor_texkeys_regression_bool_query():
+    query_str = "texkey Chen:2014cwa and a Moskovic"
+    expected_es_query = {
+        "bool": {
+            "must": [
+                {
+                    "match": {
+                        "texkeys.raw": "Chen:2014cwa"
+                    }
+                },
+                {
+                    "nested": {
+                        "path": "authors",
+                        "query": {
+                            "bool": {
+                                "must": [
+                                    {
+                                        "match": {
+                                            "authors.last_name": {
+                                                "query": "Moskovic",
+                                                "operator": "AND"
+                                            }
+                                        }
+                                    }
+                                ]
+                            }
+                        }
+                    }
+                }
+            ]
+        }
+    }
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query

tests/test_restructuring_visitor.py

@@ -474,7 +474,7 @@
          ),
         (
             'find texkey Hirata:1992*',
-            KeywordOp(Keyword('texkeys'), Value('Hirata:1992*', contains_wildcard=True))
+            KeywordOp(Keyword('texkeys.raw'), Value('Hirata:1992*', contains_wildcard=True))
         ),
 
         # Queries for implicit "and" removal