dojson: fix work access point and title

* Fixes work access points $t.
* Fixes bf:KeyTitle tag 222

Co-Authored-by: Peter Weber <[email protected]>

rero_ils/modules/documents/dojson/contrib/marc21tojson/dnb/model.py

@@ -490,7 +490,7 @@ def marc21_to_part_of(self, key, value):
     do_part_of(self, marc21, key, value)
 
 
-@marc21.over('work_access_point', '(^130..|^700.2|^710.2|^730..)')
+@marc21.over('work_access_point', '(^130..|^730..)')
 @utils.for_each_value
 @utils.ignore_value
 def marc21_to_work_access_point(self, key, value):

rero_ils/modules/documents/dojson/contrib/marc21tojson/kul/model.py

@@ -502,7 +502,7 @@ def marc21_to_part_of(self, key, value):
     do_part_of(self, marc21, key, value)
 
 
-@marc21.over('work_access_point', '(^130..|^700.2|^710.2|^730..)')
+@marc21.over('work_access_point', '(^130..|^730..)')
 @utils.for_each_value
 @utils.ignore_value
 def marc21_to_work_access_point(self, key, value):

rero_ils/modules/documents/dojson/contrib/marc21tojson/rero/model.py

@@ -28,8 +28,9 @@
     get_field_items, not_repetitive, re_identified, \
     remove_trailing_punctuation
 
-from ..utils import _CONTRIBUTION_ROLE, do_acquisition_terms_from_field_037, \
-    do_copyright_date, do_credits, do_dissertation, do_edition_statement, \
+from ..utils import _CONTRIBUTION_ROLE, do_abbreviated_title, \
+    do_acquisition_terms_from_field_037, do_copyright_date, do_credits, \
+    do_dissertation, do_edition_statement, \
     do_electronic_locator_from_field_856, do_frequency_field_310_321, \
     do_identified_by_from_field_020, do_identified_by_from_field_022, \
     do_identified_by_from_field_024, do_identified_by_from_field_028, \
@@ -111,6 +112,13 @@ def marc21_to_title(self, key, value):
 @utils.ignore_value
 def marc21_to_contribution(self, key, value):
     """Get contribution."""
+    # exclude work access points
+    if key[:3] in ['700', '710'] and value.get('t'):
+        work_access_point = do_work_access_point(marc21, key, value)
+        if work_access_point:
+            self.setdefault('work_access_point', [])
+            self['work_access_point'].append(work_access_point)
+        return None
     agent = {}
     subfields_0 = utils.force_list(value.get('0'))
     if subfields_0:
@@ -267,6 +275,14 @@ def marc21_to_copyright_date(self, key, value):
     return copyright_dates or None
 
 
+@marc21.over('title', '(^210|^222)..')
+@utils.ignore_value
+def marc21_to_abbreviated_title(self, key, value):
+    """Get abbreviated title data."""
+    title_list = do_abbreviated_title(self, marc21, key, value)
+    return title_list or None
+
+
 @marc21.over('editionStatement', '^250..')
 @utils.for_each_value
 @utils.ignore_value
@@ -1050,7 +1066,7 @@ def marc21_to_masked(self, key, value):
     return value.get('a') == 'masked'
 
 
-@marc21.over('work_access_point', '(^130..|^700.2|^710.2|^730..)')
+@marc21.over('work_access_point', '(^130..|^730..)')
 @utils.for_each_value
 @utils.ignore_value
 def marc21_to_work_access_point(self, key, value):

rero_ils/modules/documents/dojson/contrib/marc21tojson/slsp/model.py

@@ -512,7 +512,7 @@ def marc21_to_part_of(self, key, value):
     do_part_of(self, marc21, key, value)
 
 
-@marc21.over('work_access_point', '(^130..|^700.2|^710.2|^730..)')
+@marc21.over('work_access_point', '(^130..|^730..)')
 @utils.for_each_value
 @utils.ignore_value
 def marc21_to_work_access_point(self, key, value):

rero_ils/modules/documents/dojson/contrib/marc21tojson/utils.py

@@ -363,7 +363,8 @@ def do_language(data, marc21):
 def do_abbreviated_title(data, marc21, key, value):
     """Get abbreviated title data.
 
-    * bf:Title = 210|222
+    * bf:AbbreviatedTitle = 210
+    * bf:KeyTitle = 222
     * mainTitle = $a
     * subtitle = $e
     * responsibilityStatement = $f|$g
@@ -376,8 +377,11 @@ def do_abbreviated_title(data, marc21, key, value):
     if value.get('a'):
         main_title = not_repetitive(
             marc21.bib_id, marc21.bib_id, key, value, 'a')
+        title_type = 'bf:AbbreviatedTitle'
+        if key[:3] == '222':
+            title_type = 'bf:KeyTitle'
         title = {
-            'type': 'bf:AbbreviatedTitle',
+            'type': title_type,
             'mainTitle': [{'value': main_title}]
         }
         if value.get('b'):
@@ -595,6 +599,10 @@ def build_agent():
 
     # exclude work access points
     if key[:3] in ['700', '710'] and value.get('t'):
+        work_access_point = do_work_access_point(marc21, key, value)
+        if work_access_point:
+            data.setdefault('work_access_point', [])
+            data['work_access_point'].append(work_access_point)
         return None
 
     agent = {}
@@ -614,23 +622,22 @@ def build_agent():
         agent = build_agent()
 
     if value.get('4'):
-        roles = []
+        roles = set()
         for role in utils.force_list(value.get('4')):
-            if len(role) != 3 and 'http' not in role:
+            role = role.split('/')[-1].lower()
+            if len(role) != 3:
                 error_print('WARNING CONTRIBUTION ROLE LENGTH:',
                             marc21.bib_id, marc21.rero_id, role)
-                role = role[:3]
             if role == 'sce':
                 error_print('WARNING CONTRIBUTION ROLE SCE:',
                             marc21.bib_id, marc21.rero_id,
                             'sce --> aus')
                 role = 'aus'
-            role = role.lower()
-            if role not in _CONTRIBUTION_ROLE and 'http' not in role:
+            if role not in _CONTRIBUTION_ROLE:
                 error_print('WARNING CONTRIBUTION ROLE DEFINITION:',
                             marc21.bib_id, marc21.rero_id, role)
                 role = 'ctb'
-            roles.append(role)
+            roles.add(role)
     else:
         if key[:3] == '100':
             roles = ['cre']
@@ -641,7 +648,7 @@ def build_agent():
     if agent:
         return {
             'agent': agent,
-            'role': list(set(roles))
+            'role': list(roles)
         }
     return None
 
@@ -1647,20 +1654,17 @@ def do_work_access_point(marc21, key, value):
                 not_repetitive(marc21.bib_id, marc21.bib_id, key, value, 'b'))
         dates = not_repetitive(marc21.bib_id, marc21.bib_id, key, value, 'd')
         if dates:
-            dates = dates.rstrip(',')
-            dates = remove_trailing_punctuation(dates).split('-')
+            split_dates = dates.split('-')
+            date_of_birth = split_dates[0].strip()
+            if date_of_birth:
+                agent['date_of_birth'] = date_of_birth
             try:
-                date_of_birth = dates[0].strip()
-                if date_of_birth:
-                    agent['date_of_birth'] = date_of_birth
-            except Exception:
-                pass
-            try:
-                date_of_death = dates[1].strip()
+                date_of_death = split_dates[1].strip()
                 if date_of_death:
                     agent['date_of_death'] = date_of_death
             except Exception:
                 pass
+
         if value.get('c'):
             agent['qualifier'] = remove_trailing_punctuation(
                 not_repetitive(marc21.bib_id, marc21.bib_id, key, value, 'c'))
@@ -1672,8 +1676,7 @@ def do_work_access_point(marc21, key, value):
             agent['preferred_name'] = not_repetitive(
                 marc21.bib_id, marc21.bib_id, key, value, 'a')
         if value.get('b'):
-            agent['subordinate_unit'] = not_repetitive(
-                marc21.bib_id, marc21.bib_id, key, value, 'b')
+            agent['subordinate_unit'] = list(utils.force_list(value.get('b')))
     if agent:
         work_access_point['agent'] = agent
     if value.get(title_tag):

rero_ils/modules/documents/jsonschemas/documents/document_work_access_point-v0.0.1.json

@@ -108,7 +108,7 @@
                     "title": "Title",
                     "description": "Title of the part, section, or supplement",
                     "type": "string",
-                    "minLength": 3
+                    "minLength": 1
                   }
                 }
               }
@@ -130,13 +130,13 @@
               "items": {
                 "title": "Medium of performance (music)",
                 "type": "string",
-                "minLength": 3
+                "minLength": 1
               }
             },
             "arranged_statement_for_music": {
               "title": "Arranged statement (music)",
               "type": "string",
-              "minLength": 3
+              "minLength": 1
             },
             "key_for_music": {
               "title": "Key (music)",

tests/unit/test_documents_dojson.py

@@ -1349,6 +1349,11 @@ def test_marc21_to_contribution(mock_get):
         <subfield code="d">1921-2014</subfield>
         <subfield code="4">edt</subfield>
       </datafield>
+      <datafield tag="700" ind1="1" ind2=" ">
+        <subfield code="a">Santamaría, Germán</subfield>
+        <subfield code="t">No morirás</subfield>
+        <subfield code="l">français</subfield>
+      </datafield>
       <datafield tag="710" ind1=" " ind2=" ">
         <subfield code="a">RERO</subfield>
       </datafield>
@@ -5602,3 +5607,26 @@ def test_marc21_to_original_language():
     marc21json = create_record(marc21xml)
     data = marc21.do(marc21json)
     assert data.get('originalLanguage') == ['eng']
+
+
+def test_abbreviated_title(app, marc21_record):
+    """Test abbreviated title to MARC21 transformation."""
+    marc21xml = """
+    <record>
+      <datafield tag="210" ind1="0" ind2=" ">
+        <subfield code="a">Günter Gianni Piontek Skulpt.</subfield>
+      </datafield>
+      <datafield tag="222" ind1=" " ind2="0">
+        <subfield code="a">Günter Gianni Piontek, Skulpturen</subfield>
+      </datafield>
+    </record>
+    """
+    marc21json = create_record(marc21xml)
+    data = marc21.do(marc21json)
+    assert data.get('title') == [{
+        'type': 'bf:AbbreviatedTitle',
+        'mainTitle': [{'value': 'Günter Gianni Piontek Skulpt.'}]
+    }, {
+        'type': 'bf:KeyTitle',
+        'mainTitle': [{'value': 'Günter Gianni Piontek, Skulpturen'}]
+    }]