From cb9edd2fa603b9ff8328dd877c75bce2e0dd174c Mon Sep 17 00:00:00 2001
From: Bertrand Zuchuat <bertrand.zuchuat@rero.ch>
Date: Tue, 9 Nov 2021 11:13:12 +0100
Subject: [PATCH] deposit: improve document type detection.

If field 502 is available in the record, it is
analyzed to determine the type of document.

Co-Authored-by: Bertrand Zuchuat <bertrand.zuchuat@rero.ch>
---
 sonar/modules/documents/dojson/sru/model.py   | 27 ++++++++++---------
 .../api/swisscovery/test_swisscovery_rest.py  | 14 ++++++++++
 2 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/sonar/modules/documents/dojson/sru/model.py b/sonar/modules/documents/dojson/sru/model.py
index 4f395f81..7dcc33c4 100644
--- a/sonar/modules/documents/dojson/sru/model.py
+++ b/sonar/modules/documents/dojson/sru/model.py
@@ -507,23 +507,22 @@ def determine_type(field):
         """Determine type of document."""
         # Bachelor thesis
         if 'bachelor' in field.get(
-                'a', '') or 'bachelor' in field.get('b', ''):
+                'a', '').lower() or 'bachelor' in field.get('b', '').lower():
             return 'coar:c_7a1f'
 
         # Master thesis
-        if 'master' in field.get('a', '') or 'master' in field.get(
-                'b', ''):
+        if 'master' in field.get('a', '').lower() or 'master' in field.get(
+                'b', '').lower():
             return 'coar:c_bdcc'
 
         # Doctoral thesis
         if 'dissertation' in field.get(
-                'a', '') or 'dissertation' in field.get(
-                    'b', '') or 'thèse' in field.get(
-                        'a', '') or 'thèse' in field.get('b', ''):
+                'a', '').lower() or 'dissertation' in field.get(
+                    'b', '').lower() or 'thèse' in field.get(
+                        'a', '').lower() or 'thèse' in field.get('b', '')\
+                            .lower():
             return 'coar:c_db06'
 
-        # Thesis
-        return 'coar:c_46ec'
 
     leader_06 = value[6]
 
@@ -570,11 +569,15 @@ def determine_type(field):
         if field_502:
             if type(field_502) == tuple:
                 for fld502 in field_502:
-                    # Bachelor thesis
-                    return determine_type(fld502)
+                    doctype = determine_type(fld502)
+                    if doctype:
+                        return doctype
             else:
-                # Bachelor thesis
-                return determine_type(field_502)
+                doctype = determine_type(field_502)
+                if doctype:
+                    return doctype
+            # Thesis
+            return 'coar:c_46ec'
         # Book
         if leader_07 == 'm':
             return 'coar:c_2f33'
diff --git a/tests/api/swisscovery/test_swisscovery_rest.py b/tests/api/swisscovery/test_swisscovery_rest.py
index 9a68308a..70d9adf2 100644
--- a/tests/api/swisscovery/test_swisscovery_rest.py
+++ b/tests/api/swisscovery/test_swisscovery_rest.py
@@ -22,6 +22,7 @@
 
 
 def test_get_record(client, user, submitter):
+    """Test get record on swisscovery."""
     url = url_for('swisscovery.get_record')
 
     # Not logged
@@ -184,3 +185,16 @@ def test_get_record(client, user, submitter):
             }
         }
     }
+
+
+def test_document_type(client, submitter):
+    """Test of the document type with the content of field 502."""
+    login_user_via_session(client, email=submitter['email'])
+    res = client.get(
+        url_for('swisscovery.get_record',
+                query='991050676859705501',
+                type='mms_id',
+                format='deposit'))
+    assert res.status_code == 200
+    data = res.json
+    assert data['metadata']['documentType'] == 'coar:c_db06'