fixed error for out of index in normalisation process

Author: Dragos0000

ted_sws/notice_metadata_processor/adapters/notice_metadata_extractor.py

@@ -363,8 +363,10 @@ def legal_basis_directive(self):
     @property
     def extracted_notice_subtype(self):
         return extract_text_from_element(
-            element=self.manifestation_root.find(self.xpath_registry.xpath_notice_subtype, namespaces=self.namespaces))
-
+            element=self.manifestation_root.find(self.xpath_registry.xpath_notice_subtype_first,
+                                                 namespaces=self.namespaces)) or extract_text_from_element(
+            element=self.manifestation_root.find(self.xpath_registry.xpath_notice_subtype_second,
+                                                 namespaces=self.namespaces))
     @property
     def extracted_eform_type(self):
         return extract_attribute_from_element(

ted_sws/notice_metadata_processor/adapters/notice_metadata_normaliser.py

@@ -284,7 +284,11 @@ def get_form_type_notice_type_and_legal_basis(cls, extracted_notice_subtype: str
          Get the values for form type, notice type and legal basis from the eForm mapping files
         """
         ef_map: pd.DataFrame = mapping_registry.ef_notice_df
-        filtered_df = ef_map.query(f"{E_FORMS_SUBTYPE_KEY}=='{extracted_notice_subtype}'").to_dict(orient='records')[0]
+        try:
+            filtered_df = ef_map.query(f"{E_FORMS_SUBTYPE_KEY}=='{extracted_notice_subtype}'").to_dict(orient='records')[0]
+        except:
+            raise Exception(
+                'No eform subtype was extracted from the notice content. Please check that the field exists in the XML content')
         try:
             form_type = filtered_df[FORM_TYPE_KEY]
             notice_type = filtered_df[E_FORM_NOTICE_TYPE_COLUMN]

ted_sws/notice_metadata_processor/adapters/xpath_registry.py

@@ -203,9 +203,13 @@ def xpath_legal_basis_directive(self):
         return ".//cbc:RegulatoryDomain"
 
     @property
-    def xpath_notice_subtype(self):
+    def xpath_notice_subtype_first(self):
         return ".//ext:UBLExtensions/ext:UBLExtension/ext:ExtensionContent/efext:EformsExtension/efac:NoticeSubType/cbc:SubTypeCode[@listName='notice-subtype']"
 
+    @property
+    def xpath_notice_subtype_second(self):
+        return ".//ext:UBLExtensions/ext:UBLExtension/ext:ExtensionContent/efext:EformsExtension/efac:NoticeSubType/cbc:SubTypeCode"
+
     @property
     def xpath_form_type(self):
         return ".//cbc:NoticeTypeCode"

tests/test_data/notice_normalisation/2023-OJS153-00486429.xml

@@ -16,6 +16,11 @@ def file_system_repository_path():
     return TEST_DATA_PATH / "notice_transformer" / "mapping_suite_processor_repository"
 
 
+@pytest.fixture
+def notice_normalisation_test_data_path():
+    return TEST_DATA_PATH / "notice_normalisation"
+
+
 @pytest.fixture
 def eforms_xml_notice_paths() -> List[pathlib.Path]:
     eforms_xml_notices_path = TEST_DATA_PATH / "eforms_samples"

tests/test_data/notice_normalisation/no_eform_subtype_notice.xml

@@ -1,5 +1,6 @@
 import pytest
 
+from ted_sws.core.model.manifestation import XMLManifestation
 from ted_sws.core.model.metadata import NormalisedMetadata
 from ted_sws.core.model.notice import NoticeStatus
 from ted_sws.notice_metadata_processor.adapters.notice_metadata_extractor import \
@@ -11,7 +12,8 @@
 from ted_sws.notice_metadata_processor.services.metadata_constraints import filter_df_by_variables
 from ted_sws.notice_metadata_processor.services.metadata_normalizer import normalise_notice, normalise_notice_by_id, \
     check_if_xml_manifestation_is_eform, find_metadata_extractor_based_on_xml_manifestation, \
-    find_metadata_normaliser_based_on_xml_manifestation, extract_notice_metadata, normalise_notice_metadata
+    find_metadata_normaliser_based_on_xml_manifestation, extract_notice_metadata, normalise_notice_metadata, \
+    extract_and_normalise_notice_metadata
 from ted_sws.resources.mapping_files_registry import MappingFilesRegistry
 
 
@@ -217,3 +219,19 @@ def test_get_form_type_notice_type_and_legal_basis():
     assert form_type == 'competition'
     assert notice_type == 'cn-social'
     assert legal_basis == '32014L0024'
+
+
+def test_normalising_notice_out_of_index(notice_normalisation_test_data_path):
+    notice_xml_path = notice_normalisation_test_data_path / "2023-OJS153-00486429.xml"
+    notice_content = notice_xml_path.read_text(encoding="utf-8")
+    normalised_notice_metadata = extract_and_normalise_notice_metadata(
+        xml_manifestation=XMLManifestation(object_data=notice_content))
+    assert normalised_notice_metadata.eforms_subtype == "16"
+    assert normalised_notice_metadata.notice_publication_number == "00486429-2023"
+
+    broken_notice_xml_path = notice_normalisation_test_data_path / "no_eform_subtype_notice.xml"
+    broke_notice_content = broken_notice_xml_path.read_text(encoding="utf-8")
+
+    with pytest.raises(Exception):
+        extract_and_normalise_notice_metadata(
+            xml_manifestation=XMLManifestation(object_data=broke_notice_content))