es_visitor: support for legacy jhep/jcap queries

* INSPIR-3400
inspirehep · Apr 14, 2020 · b81afce · b81afce
1 parent bd907de
commit b81afce
Show file tree

Hide file tree

Showing 2 changed files with 110 additions and 0 deletions.
diff --git a/inspire_query_parser/visitors/elastic_search_visitor.py b/inspire_query_parser/visitors/elastic_search_visitor.py
@@ -76,11 +76,13 @@ class ElasticSearchVisitor(Visitor):
     JOURNAL_VOLUME = 'journal_volume'
     JOURNAL_PAGE_START = 'page_start'
     JOURNAL_ART_ID = 'artid'
+    JOURNAL_YEAR = 'year'
     JOURNAL_FIELDS_MAPPING = {
         JOURNAL_TITLE: '.'.join((JOURNAL_FIELDS_PREFIX, JOURNAL_TITLE)),
         JOURNAL_VOLUME: '.'.join((JOURNAL_FIELDS_PREFIX, JOURNAL_VOLUME)),
         JOURNAL_PAGE_START: '.'.join((JOURNAL_FIELDS_PREFIX, JOURNAL_PAGE_START)),
         JOURNAL_ART_ID: '.'.join((JOURNAL_FIELDS_PREFIX, JOURNAL_ART_ID)),
+        JOURNAL_YEAR: '.'.join((JOURNAL_FIELDS_PREFIX, JOURNAL_YEAR)),
     }
     # ########################################
 
@@ -518,6 +520,7 @@ def _preprocess_journal_query_value(self, third_journal_field, old_publication_i
             self.JOURNAL_TITLE,
             self.JOURNAL_VOLUME,
             third_journal_field,
+            self.JOURNAL_YEAR
         ]
         values_list = [
             value.strip()
@@ -574,6 +577,15 @@ def _generate_journal_nested_queries(self, value):
                 )
             )
 
+        if self.JOURNAL_YEAR in new_publication_info:
+            queries_for_each_field.append(
+                generate_match_query(
+                    self.JOURNAL_FIELDS_MAPPING[self.JOURNAL_YEAR],
+                    new_publication_info[self.JOURNAL_YEAR],
+                    with_operator_and=False
+                )
+            )
+
         if third_journal_field in new_publication_info:
             artid_or_page_start = new_publication_info[third_journal_field]
             match_queries = [

diff --git a/tests/test_elastic_search_visitor.py b/tests/test_elastic_search_visitor.py
@@ -2600,3 +2600,101 @@ def test_nested_query_exact_last_name():
 
     generated_es_query = _parse_query(query_string)
     assert generated_es_query == expected_es_query
+
+
+def test_elastic_search_visitor_type_code_legacy_compatible_jhep_search():
+    query_str = "j jhep,0903,112,2009"
+    expected_es_query = {
+       "nested": {
+          "path": "publication_info",
+          "query": {
+             "bool": {
+                "must": [
+                   {
+                      "match": {
+                         "publication_info.journal_title": "jhep"
+                      }
+                   },
+                   {
+                      "match": {
+                         "publication_info.journal_volume": "0903"
+                      }
+                   },
+                   {
+                      "match": {
+                         "publication_info.year": "2009"
+                      }
+                   },
+                   {
+                      "bool": {
+                         "should": [
+                            {
+                               "match": {
+                                  "publication_info.page_start": "112"
+                               }
+                            },
+                            {
+                               "match": {
+                                  "publication_info.artid": "112"
+                               }
+                            }
+                         ]
+                      }
+                   }
+                ]
+             }
+          }
+       }
+    }
+
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query
+
+
+def test_elastic_search_visitor_type_code_legacy_compatible_jcap_search():
+    query_str = "j jcap,0903,112,2009"
+    expected_es_query = {
+       "nested": {
+          "path": "publication_info",
+          "query": {
+             "bool": {
+                "must": [
+                   {
+                      "match": {
+                         "publication_info.journal_title": "jcap"
+                      }
+                   },
+                   {
+                      "match": {
+                         "publication_info.journal_volume": "0903"
+                      }
+                   },
+                   {
+                      "match": {
+                         "publication_info.year": "2009"
+                      }
+                   },
+                   {
+                      "bool": {
+                         "should": [
+                            {
+                               "match": {
+                                  "publication_info.page_start": "112"
+                               }
+                            },
+                            {
+                               "match": {
+                                  "publication_info.artid": "112"
+                               }
+                            }
+                         ]
+                      }
+                   }
+                ]
+             }
+          }
+       }
+    }
+
+    generated_es_query = _parse_query(query_str)
+    assert generated_es_query == expected_es_query