normaliser changes

ted_sws/notice_metadata_processor/services/metadata_normalizer.py

@@ -1,4 +1,5 @@
 import abc
+import re
 from datetime import datetime
 from typing import Dict, Tuple, List
 
@@ -137,12 +138,22 @@ def normalise_legal_basis_value(cls, value: str) -> str:
     @classmethod
     def normalise_form_number(cls, value: str) -> str:
         """
-        Normalise form number to be F{number} format
+        Normalise form number to be F{number} format.
+        ##Decided to keep normalisation of the input data
+            Rules:
+            * The form number should start with a letter ("F", "T")
+            * The form number isn't always a number (CEI,EEIG)
+            * If the number is between 1 - 9 then it must have 0 as prefix (F02 not F2)
         :param value:
         :return:
         """
-        if value and not value.startswith("F") and not value[0].isalpha():
-            return "F" + value
+        form_number_parts = re.split(r"(?=\d)", value, 1)
+        if len(form_number_parts) == 2:
+            text_part: str = form_number_parts[0] if form_number_parts[0] else "F"
+            number_part: str = form_number_parts[1]
+            if text_part.isalpha() and number_part.isdecimal():
+                number_part = "0" + number_part if number_part and len(number_part) < 2 else number_part
+                return text_part + number_part
         return value
 
     @classmethod
@@ -168,7 +179,8 @@ def get_filter_variables_values(cls, form_number: str, extracted_notice_type: st
                 filter_map.query(f"{FORM_NUMBER_KEY}=='{variables[FORM_NUMBER_KEY]}'").to_dict(orient='records')[0]
         except:
             raise Exception(
-                f"This notice doesn't have a form number or the extracted form number is not in the mapping. Form number found is {form_number}")
+                f"This notice doesn't have a form number or the extracted form number is not in the mapping. "
+                f"Form number found is {form_number}, document code is {document_type_code} and legal basis is {legal_basis}")
 
         for key, value in filter_variables.items():
             if value == 0:

ted_sws/resources/mapping_files/eforms_mapping.csv

@@ -39,8 +39,5 @@ result,can-desg,32014L0025,37
 cont-modif,can-modif,32014L0024,38
 cont-modif,can-modif,32014L0025,39
 cont-modif,can-modif,32014L0023,40
-change,corr,32014L0024,41
-change,corr,32014L0025,41
-change,corr,32014L0023,41
 planning,pin-tran,32007R1370,T01
 result,can-tran,32007R1370,T02

ted_sws/resources/mapping_files/sforms_mapping.csv

@@ -17,7 +17,6 @@ F12,,32014L0024,,23
 F12,,32014L0025,,24
 F13,,32014L0024,,36
 F13,,32014L0025,,37
-F14,,,,41
 F15,,32009L0081,,27
 F15,,32014L0023,,28
 F15,,32014L0024,,25
@@ -34,10 +33,12 @@ F20,,32014L0025,,39
 F20,,32018R1046,,38
 F21,AWARD_CONTRACT,,,33
 F21,CONTRACT,,,20
+F21,PRI_ONLY,,,4
 F21,PRI_CALL_COMPETITION,,,12
 F22,AWARD_CONTRACT,,,34
 F22,CONTRACT,,,21
 F22,PER_CALL_COMPETITION,,,13
+F22,PER_ONLY,,,5
 F22,QSU_ONLY,,,15
 F23,,,,35
 F24,,,,19

tests/unit/notice_metadata_processor/test_metadata_normaliser.py

@@ -6,7 +6,8 @@
 from ted_sws.notice_metadata_processor.services.metadata_normalizer import normalise_notice, normalise_notice_by_id, \
     MetadataNormaliser, ExtractedMetadataNormaliser, FORM_NUMBER_KEY, SF_NOTICE_TYPE_KEY, LEGAL_BASIS_KEY, \
     DOCUMENT_CODE_KEY
-from ted_sws.notice_metadata_processor.services.xml_manifestation_metadata_extractor import XMLManifestationMetadataExtractor
+from ted_sws.notice_metadata_processor.services.xml_manifestation_metadata_extractor import \
+    XMLManifestationMetadataExtractor
 
 
 def test_metadata_normaliser_by_notice(indexed_notice):
@@ -40,29 +41,37 @@ def test_metadata_normaliser(indexed_notice):
 
 
 def test_normalise_form_number(indexed_notice):
-    extracted_metadata = XMLManifestationMetadataExtractor(xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
+    extracted_metadata = XMLManifestationMetadataExtractor(
+        xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
     extracted_metadata_normaliser = ExtractedMetadataNormaliser(extracted_metadata=extracted_metadata)
-    assert "18" == extracted_metadata.extracted_form_number
-    assert "F18" == extracted_metadata_normaliser.normalise_form_number(value=extracted_metadata.extracted_form_number)
-    assert "T01" == extracted_metadata_normaliser.normalise_form_number(value="T01")
-    assert None == extracted_metadata_normaliser.normalise_form_number(value=None)
+    assert extracted_metadata_normaliser.normalise_form_number("FFFSA") == "FFFSA"
+    assert extracted_metadata_normaliser.normalise_form_number("F18") == "F18"
+    assert extracted_metadata_normaliser.normalise_form_number("F01") == "F01"
+    assert extracted_metadata_normaliser.normalise_form_number("F2") == "F02"
+    assert extracted_metadata_normaliser.normalise_form_number("22") == "F22"
+    assert extracted_metadata_normaliser.normalise_form_number("2") == "F02"
+    assert extracted_metadata_normaliser.normalise_form_number("F") == "F"
+    assert extracted_metadata_normaliser.normalise_form_number("TX01") == "TX01"
+    assert extracted_metadata_normaliser.normalise_form_number("TX1") == "TX01"
+    assert extracted_metadata_normaliser.normalise_form_number("FX03FG") == "FX03FG"
+    assert extracted_metadata_normaliser.normalise_form_number("1F03FG") == "1F03FG"
 
 
 def test_normalise_legal_basis(indexed_notice):
-    extracted_metadata = XMLManifestationMetadataExtractor(xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
+    extracted_metadata = XMLManifestationMetadataExtractor(
+        xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
     extracted_metadata_normaliser = ExtractedMetadataNormaliser(extracted_metadata=extracted_metadata)
     assert "2009/81/EC" == extracted_metadata.legal_basis_directive
     assert "32009L0081" == extracted_metadata_normaliser.normalise_legal_basis_value(
         value=extracted_metadata.legal_basis_directive)
 
-
 def test_get_map_value(indexed_notice):
-    extracted_metadata = XMLManifestationMetadataExtractor(xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
+    extracted_metadata = XMLManifestationMetadataExtractor(
+        xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
     extracted_metadata_normaliser = ExtractedMetadataNormaliser(extracted_metadata=extracted_metadata)
     value = extracted_metadata_normaliser.get_map_value(mapping=MappingFilesRegistry().countries, value="DE")
     assert value == "http://publications.europa.eu/resource/authority/country/DEU"
 
-
 def test_filter_df_by_variables():
     df = MappingFilesRegistry().ef_notice_df
     filtered_df = filter_df_by_variables(df=df, form_type="planning",
@@ -71,9 +80,9 @@ def test_filter_df_by_variables():
     assert len(filtered_df.index) == 3
     assert "32014L0024" in filtered_df["eform_legal_basis"].values
 
-
 def test_get_form_type_and_notice_type(indexed_notice):
-    extracted_metadata = XMLManifestationMetadataExtractor(xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
+    extracted_metadata = XMLManifestationMetadataExtractor(
+        xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
     extracted_metadata_normaliser = ExtractedMetadataNormaliser(extracted_metadata=extracted_metadata)
     form_type, notice_type, legal_basis, eforms_subtype = extracted_metadata_normaliser.get_form_type_and_notice_type(
         ef_map=MappingFilesRegistry().ef_notice_df,
@@ -86,9 +95,9 @@ def test_get_form_type_and_notice_type(indexed_notice):
     assert "32014L0024" == legal_basis
     assert "16" == eforms_subtype
 
-
 def test_get_filter_values(indexed_notice):
-    extracted_metadata = XMLManifestationMetadataExtractor(xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
+    extracted_metadata = XMLManifestationMetadataExtractor(
+        xml_manifestation=indexed_notice.xml_manifestation).to_metadata()
     extracted_metadata_normaliser = ExtractedMetadataNormaliser(extracted_metadata=extracted_metadata)
     filter_map = MappingFilesRegistry().filter_map_df
     filter_variables_dict = extracted_metadata_normaliser.get_filter_variables_values(form_number="F07",
@@ -109,7 +118,6 @@ def test_get_filter_values(indexed_notice):
                                                                   document_type_code="7",
                                                                   legal_basis="legal")
 
-
 def test_normalising_process_on_failed_notice_in_dag(notice_2021):
     extracted_metadata = XMLManifestationMetadataExtractor(
         xml_manifestation=notice_2021.xml_manifestation).to_metadata()