Merge pull request #136 from MJedr/date-queries

visitor: support range queries for date-added and date-updated
inspirehep · Jul 28, 2020 · 9f36c6d · 9f36c6d
2 parents 1795ade + 51f7f52
commit 9f36c6d
Show file tree

Hide file tree

Showing 2 changed files with 114 additions and 10 deletions.
diff --git a/inspire_query_parser/visitors/elastic_search_visitor.py b/inspire_query_parser/visitors/elastic_search_visitor.py
@@ -493,12 +493,15 @@ def _generate_range_queries(self, fieldnames, operator_value_pairs):
             Additionally, in the aforementioned case, if a malformed date has been given, then the the method will
             return an empty dictionary.
         """
-        if self.KEYWORD_TO_ES_FIELDNAME['date'] == fieldnames:
+        if self.KEYWORD_TO_ES_FIELDNAME['date'] == fieldnames or all(
+                field in [self.KEYWORD_TO_ES_FIELDNAME['date-added'],
+                          self.KEYWORD_TO_ES_FIELDNAME['date-updated']] for field in fieldnames
+        ):
             range_queries = []
             for fieldname in fieldnames:
-                updated_operator_value_pairs = \
-                    update_date_value_in_operator_value_pairs_for_fieldname(fieldname, operator_value_pairs)
-
+                updated_operator_value_pairs = update_date_value_in_operator_value_pairs_for_fieldname(
+                    fieldname, operator_value_pairs
+                )
                 if not updated_operator_value_pairs:
                     break  # Malformed date
                 else:
@@ -791,13 +794,12 @@ def visit_value(self, node, fieldnames=None):
 
         if node.contains_wildcard:
             return self.handle_value_wildcard(node, fieldnames=fieldnames)
-
+        if fieldnames in [self.KEYWORD_TO_ES_FIELDNAME['date'], self.KEYWORD_TO_ES_FIELDNAME['date-added'],
+                          self.KEYWORD_TO_ES_FIELDNAME['date-updated']]:
+            # Date queries with simple values are transformed into range queries, among the given and the exact
+            # next date, according to the granularity of the given date.
+            return self._generate_range_queries(force_list(fieldnames), {ES_RANGE_EQ_OPERATOR: node.value})
         if isinstance(fieldnames, list):
-            if self.KEYWORD_TO_ES_FIELDNAME['date'] == fieldnames:
-                # Date queries with simple values are transformed into range queries, among the given and the exact
-                # next date, according to the granularity of the given date.
-                return self._generate_range_queries(force_list(fieldnames), {ES_RANGE_EQ_OPERATOR: node.value})
-
             if self.KEYWORD_TO_ES_FIELDNAME['journal'] == fieldnames:
                 return self._generate_journal_nested_queries(node.value)
 

diff --git a/tests/test_elastic_search_visitor.py b/tests/test_elastic_search_visitor.py
@@ -3429,3 +3429,105 @@ def test_range_date_queries_are_nested():
 
     generated_es_query = _parse_query(query_str)
     assert generated_es_query == expected_es_query
+
+
+def test_date_updated_keyword_is_handled_with_range_query():
+    query_str = 'du 2019->2020'
+    expected_es_query = {
+       "range": {
+          "_updated": {
+             "gte": "2019||/y",
+             "lte": "2020||/y"
+          }
+       }
+    }
+
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query
+
+
+def test_date_added_keyword_is_handled_with_range_query():
+    query_str = 'da 1997'
+    expected_es_query = {
+       "range": {
+          "_created": {
+             "gte": "1997||/y",
+             "lt": "1998||/y"
+          }
+       }
+    }
+
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query
+
+
+def test_all_date_fields_are_handled_correctly_with_range_query():
+    query_str = 'du 2000 and date 1997'
+    expected_es_query = {
+       "bool": {
+          "must": [
+              {
+                  "range": {
+                      "_updated": {
+                          "gte": "2000||/y",
+                          "lt": "2001||/y"
+                      }
+                  }
+              },
+              {
+                "bool": {
+                   "should": [
+                      {
+                         "range": {
+                            "earliest_date": {
+                               "gte": "1997||/y",
+                               "lt": "1998||/y"
+                            }
+                         }
+                      },
+                      {
+                         "range": {
+                            "imprints.date": {
+                               "gte": "1997||/y",
+                               "lt": "1998||/y"
+                            }
+                         }
+                      },
+                      {
+                         "range": {
+                            "preprint_date": {
+                               "gte": "1997||/y",
+                               "lt": "1998||/y"
+                            }
+                         }
+                      },
+                      {
+                         "nested": {
+                            "path": "publication_info",
+                            "query": {
+                               "range": {
+                                  "publication_info.year": {
+                                     "gte": "1997||/y",
+                                     "lt": "1998||/y"
+                                  }
+                               }
+                            }
+                         }
+                      },
+                      {
+                         "range": {
+                            "thesis_info.date": {
+                               "gte": "1997||/y",
+                               "lt": "1998||/y"
+                            }
+                         }
+                      }
+                   ]
+                }
+              }
+          ]
+       }
+    }
+
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query