Pre-Release Data Preparation Tool 3.5.0 (GND4C-Testing).

modules/connectors/gnd4c_nds/gnd4c_nds.py

@@ -6,6 +6,9 @@
 
 
 def parse_xml_content(session_data, xml_findbuch_in, input_type, input_file, error_status, propagate_logging, administrative_data, provider_rights, serializer):
+    # Aggregiertes Logging vorbereiten
+    logfile_in_multiple_date_elements = open("log_multiple_date_elements.txt", "a", encoding="utf-8")
+
     namespaces = {"gnd4c": "http://gnd4c.digicult-verbund.de"}
     xml_result = None
 
@@ -207,6 +210,11 @@ def parse_xml_content(session_data, xml_findbuch_in, input_type, input_file, err
                 if get_compare_value(geographic_area_code_uri_element) != "":
                     object_metadata["geographic_area_code"].append(get_compare_value(geographic_area_code_uri_element))
 
+            geographic_area_code_label_element = geographic_area_code_element.find("gnd4c:label", namespaces)
+            if geographic_area_code_label_element is not None:
+                if get_compare_value(geographic_area_code_label_element) != "":
+                    object_metadata["geographic_area_code"].append(get_compare_value(geographic_area_code_label_element))
+
         # gnd4c:person/gnd4c:placeOfBirth|placeOfDeath|associatedPlace
         object_metadata["place_of_birth"] = []
         object_metadata["place_of_death"] = []
@@ -273,15 +281,20 @@ def parse_xml_content(session_data, xml_findbuch_in, input_type, input_file, err
                     object_metadata["period_of_activity"].append(date_single)
 
         if len(object_metadata["date_of_birth"]) > 1:
-            logger.warning("Für Datensatz-ID {} sind {} dateOfBirth-Elemente vorhanden. Bitte prüfen, ob Geburts- und Sterbedatum korrekt zusammengesetzt wurden.".format(object_metadata["record_id"], len(object_metadata["date_of_birth"])))
-        if len(object_metadata["date_of_death"]) > 1:
-            logger.warning("Für Datensatz-ID {} sind {} dateOfDeath-Elemente vorhanden. Bitte prüfen, ob Geburts- und Sterbedatum korrekt zusammengesetzt wurden.".format(object_metadata["record_id"], len(object_metadata["date_of_death"])))
+            log_message = "Für Datensatz-ID {} sind {} dateOfBirth-Elemente vorhanden. Bitte prüfen, ob Geburts- und Sterbedatum korrekt zusammengesetzt wurden. (Datei: {})".format(object_metadata["record_id"], len(object_metadata["date_of_birth"]), input_file)
+            logger.warning(log_message)
+            logfile_in_multiple_date_elements.write("{}\n".format(log_message))
 
 
+        if len(object_metadata["date_of_death"]) > 1:
+            log_message = "Für Datensatz-ID {} sind {} dateOfDeath-Elemente vorhanden. Bitte prüfen, ob Geburts- und Sterbedatum korrekt zusammengesetzt wurden. (Datei: {})".format(object_metadata["record_id"], len(object_metadata["date_of_death"]), input_file)
+            logger.warning(log_message)
+            logfile_in_multiple_date_elements.write("{}\n".format(log_message))
 
-        if serializer == "marcxml":
-            xml_result = map2marcxml.serialize_metadata(session_data, object_id, object_level, object_type, object_parent_id, object_metadata, object_rights, object_binaries, administrative_data, xml_base=xml_result)
 
+        if serializer == "marcxml":
+            xml_result = map2marcxml.serialize_metadata(session_data, object_id, object_level, object_type, object_parent_id, object_metadata, object_rights, object_binaries, administrative_data, input_file, xml_base=xml_result)
 
+    logfile_in_multiple_date_elements.close()
     if xml_result is not None:
         return xml_result

modules/provider_specific/.provider_script_sets/0c59f30406904f9c9a563a66f53b3d5f.xml

@@ -1,2 +1,2 @@
 <?xml version='1.0' encoding='UTF-8'?>
-<provider_script_set><provider id="GND4C-NDS" range="global"><set id="0c59f30406904f9c9a563a66f53b3d5f"><name>GND4C-NDS-Format in MarcXML umwandeln</name><description></description><modules><module><module_provider>common</module_provider><module_name>maintenance_function.py</module_name><module_config>{'maintenance_type': 'mapping_definition'}</module_config></module><module><module_provider>GND4C</module_provider><module_name>split_marcxml_records_from_gnd4c_connector.py</module_name><module_config/></module></modules></set></provider></provider_script_set>
+<provider_script_set><provider id="GND4C-NDS" range="global"><set id="0c59f30406904f9c9a563a66f53b3d5f"><name>GND4C-NDS-Format in MarcXML umwandeln</name><description></description><modules><module><module_provider>common</module_provider><module_name>maintenance_function.py</module_name><module_config>{'maintenance_type': 'mapping_definition'}</module_config></module><module><module_provider>GND4C</module_provider><module_name>set_authentication_code_gnd3.py</module_name><module_config/></module><module><module_provider>GND4C</module_provider><module_name>set_general_note_labw.py</module_name><module_config/></module><module><module_provider>GND4C</module_provider><module_name>set_other_classification_number_from_config.py</module_name><module_config/></module><module><module_provider>GND4C</module_provider><module_name>set_source_data_found_from_config.py</module_name><module_config/></module><module><module_provider>GND4C</module_provider><module_name>set_teilbestandskennzeichen_d.py</module_name><module_config/></module><module><module_provider>GND4C</module_provider><module_name>sort_marcxml_fields.py</module_name><module_config/></module></modules></set></provider></provider_script_set>

modules/provider_specific/GND4C/sort_marcxml_fields.py

@@ -0,0 +1,34 @@
+from lxml import etree
+from loguru import logger
+
+
+def get_sort_key_from_tag(element):
+    if "tag" in element.attrib:
+        return element.attrib["tag"]
+    else:
+        return "00000000"
+
+def get_sort_key_from_code(element):
+    if "code" in element.attrib:
+        code_value = element.attrib["code"]
+        if code_value.isnumeric():
+            # Subfields deren Code einen Buchstaben enthält, voranstellen.
+            code_value = "zzzzzzzz{}".format(code_value)
+        return code_value
+    else:
+        return "00000000"
+
+
+def parse_xml_content(xml_findbuch_in, input_type, input_file):
+    """control-, data- und subfields sortieren"""
+    record_elements = xml_findbuch_in.findall("//{http://www.loc.gov/MARC21/slim}record")
+    for record_element in record_elements:
+        # controlfields und datafields sortieren
+        record_element[:] = sorted(record_element, key=get_sort_key_from_tag)
+
+        datafield_elements = record_element.findall("{http://www.loc.gov/MARC21/slim}datafield")
+        for datafield_element in datafield_elements:
+            # subfields sortieren
+            datafield_element[:] = sorted(datafield_element, key=get_sort_key_from_code)
+
+    return xml_findbuch_in