documents: search in full-text

sonar/modules/documents/query.py

@@ -17,6 +17,8 @@
 
 """Query for documents."""
 
+import re
+
 from elasticsearch_dsl.query import Q
 from flask import current_app, request
 
@@ -40,12 +42,21 @@ def documents_query_parser(qstr=None):
     if not qstr:
         return Q()
 
+    fields = FIELDS.copy()
+
+    # Special treatment for fulltext, we want to search in all fields and
+    # additionally in the fulltext field.
+    if 'fulltext:' in qstr:
+        result = re.match(r'^fulltext:(.*)$', qstr)
+        qstr = result.group(1)
+        fields.append('fulltext')
+
     operator, query_type = get_operator_and_query_type(qstr)
 
     return Q(query_type,
              query=qstr,
              default_operator=operator,
-             fields=FIELDS,
+             fields=fields,
              lenient=True)
     # lenient property is necessary to make it wildcards working, see
     # https://github.com/elastic/elasticsearch/issues/39577#issuecomment-468751713

tests/api/test_api_query.py

@@ -22,7 +22,8 @@
 from invenio_accounts.testutils import login_user_via_session
 
 
-def test_api_query(client, document, document_json, make_document, superuser):
+def test_api_query(client, document_with_file, document_json, make_document,
+                   superuser):
     """Test simple flow using REST API."""
     headers = [('Content-Type', 'application/json')]
     login_user_via_session(client, email=superuser['email'])
@@ -92,6 +93,15 @@ def test_api_query(client, document, document_json, make_document, superuser):
     assert response.status_code == 200
     assert response.json['hits']['total'] == 1
 
+    # Test search in fulltext
+    response = client.get(url_for('invenio_records_rest.doc_list',
+                                  q='fulltext:the',
+                                  debug=1),
+                          headers=headers)
+    assert response.status_code == 200
+    assert response.json['hits']['total'] == 1
+    assert len(response.json['hits']['hits'][0]['explanation']['details']) == 3
+
     # Not allowed operator
     with pytest.raises(Exception) as exception:
         response = client.get(url_for('invenio_records_rest.doc_list',