def test_update_business_type_categories(recipients_data): le = LegalEntity.objects.filter(legal_entity_id=1111).first() le.business_types = 'P' le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert le.business_categories == ['individuals'] le.business_types = 'L' le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'authorities_and_commissions' in le.business_categories assert 'government' in le.business_categories le.business_types = 'M' le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'nonprofit' in le.business_categories
def test_update_business_type_categories(): business_types = "P" business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert business_categories == ["individuals"] business_types = "L" business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "authorities_and_commissions" in business_categories assert "government" in business_categories business_types = "M" business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "nonprofit" in business_categories
def extract_business_categories(recipient_name, recipient_duns): """ Extract the business categories via the recipient hash Args: recipient_name: name of the recipient recipient_duns: duns of the recipient Returns: list of business categories """ business_categories = set() if recipient_name in SPECIAL_CASES: return list(business_categories) # Go through DUNS first d_business_cat = DUNS.objects.filter(legal_business_name=recipient_name, awardee_or_recipient_uniqu=recipient_duns)\ .order_by('-update_date').values('business_types_codes', 'entity_structure').first() if d_business_cat: duns_types_mapping = get_duns_business_types_mapping() business_types_codes = d_business_cat['business_types_codes'] if d_business_cat['entity_structure']: business_types_codes.append(d_business_cat['entity_structure']) business_types = {duns_types_mapping[type]: 'true' for type in d_business_cat['business_types_codes'] if type in duns_types_mapping} business_categories |= set(get_business_categories(business_types, data_type='fpds')) # combine with latest transaction's business categories le_business_cat = LegalEntity.objects.filter(recipient_name=recipient_name, recipient_unique_id=recipient_duns) \ .order_by('-update_date').values('business_categories').first() if le_business_cat and le_business_cat['business_categories']: business_categories |= set(le_business_cat['business_categories']) return list(business_categories)
def test_update_business_type_categories(recipients_data): le = LegalEntity.objects.filter(legal_entity_id=1111).first() le.business_types = "P" le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert le.business_categories == ["individuals"] le.business_types = "L" le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "authorities_and_commissions" in le.business_categories assert "government" in le.business_categories le.business_types = "M" le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "nonprofit" in le.business_categories
def test_update_business_type_categories_faads_format(recipients_data): le = LegalEntity.objects.filter(legal_entity_id=1111).first() le.business_types = '01' # B equivalent le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'government' in le.business_categories assert 'local_government' in le.business_categories le.business_types = '12' # M equivalent le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'nonprofit' in le.business_categories le.business_types = '21' # P equivalent le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'individuals' in le.business_categories le.business_types = '23' # R equivalent le.business_categories = get_business_categories({'business_types': le.business_types}, 'fabs') assert 'small_business' in le.business_categories assert 'category_business' in le.business_categories
def test_update_business_type_categories_faads_format(): business_types = "01" # B equivalent business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "government" in business_categories assert "local_government" in business_categories business_types = "12" # M equivalent business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "nonprofit" in business_categories business_types = "21" # P equivalent business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "individuals" in business_categories business_types = "23" # R equivalent business_categories = get_business_categories( {"business_types": business_types}, "fabs") assert "small_business" in business_categories assert "category_business" in business_categories
def test_update_business_type_categories_faads_format(recipients_data): le = LegalEntity.objects.filter(legal_entity_id=1111).first() le.business_types = "01" # B equivalent le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "government" in le.business_categories assert "local_government" in le.business_categories le.business_types = "12" # M equivalent le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "nonprofit" in le.business_categories le.business_types = "21" # P equivalent le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "individuals" in le.business_categories le.business_types = "23" # R equivalent le.business_categories = get_business_categories( {"business_types": le.business_types}, "fabs") assert "small_business" in le.business_categories assert "category_business" in le.business_categories
def extract_business_categories(recipient_name, recipient_duns, recipient_hash): """ Extract the business categories via the recipient hash Args: recipient_name: name of the recipient recipient_duns: duns of the recipient recipient_hash: hash of name and duns Returns: list of business categories """ business_categories = set() if recipient_name in SPECIAL_CASES: return list(business_categories) # Go through DUNS first d_business_cat = (DUNS.objects.filter( legal_business_name=recipient_name, awardee_or_recipient_uniqu=recipient_duns).order_by( "-update_date").values("business_types_codes", "entity_structure").first()) if d_business_cat: duns_types_mapping = get_duns_business_types_mapping() business_types_codes = d_business_cat["business_types_codes"] if d_business_cat["entity_structure"]: business_types_codes.append(d_business_cat["entity_structure"]) business_types = { duns_types_mapping[type]: "true" for type in d_business_cat["business_types_codes"] if type in duns_types_mapping } business_categories |= set( get_business_categories(business_types, data_type="fpds")) # combine with latest transaction's business categories # from usaspending_api.references.models import LegalEntity latest_transaction = (UniversalTransactionView.objects.filter( recipient_hash=recipient_hash, action_date__gte=settings.API_SEARCH_MIN_DATE).order_by( "-action_date", "-transaction_id").values("business_categories").first()) if latest_transaction and latest_transaction["business_categories"]: business_categories |= set(latest_transaction["business_categories"]) return sorted(business_categories)
def extract_business_categories(recipient_name, recipient_duns): """ Extract the business categories via the recipient hash Args: recipient_name: name of the recipient recipient_duns: duns of the recipient Returns: list of business categories """ business_categories = set() if recipient_name in SPECIAL_CASES: return list(business_categories) # Go through DUNS first d_business_cat = (DUNS.objects.filter( legal_business_name=recipient_name, awardee_or_recipient_uniqu=recipient_duns).order_by( "-update_date").values("business_types_codes", "entity_structure").first()) if d_business_cat: duns_types_mapping = get_duns_business_types_mapping() business_types_codes = d_business_cat["business_types_codes"] if d_business_cat["entity_structure"]: business_types_codes.append(d_business_cat["entity_structure"]) business_types = { duns_types_mapping[type]: "true" for type in d_business_cat["business_types_codes"] if type in duns_types_mapping } business_categories |= set( get_business_categories(business_types, data_type="fpds")) # combine with latest transaction's business categories le_business_cat = (LegalEntity.objects.filter( recipient_name=recipient_name, recipient_unique_id=recipient_duns).order_by("-update_date").values( "business_categories").first()) if le_business_cat and le_business_cat["business_categories"]: business_categories |= set(le_business_cat["business_categories"]) return list(business_categories)
def insert_new_fpds(self, to_insert, total_rows): place_of_performance_field_map = { "location_country_code": "place_of_perform_country_c", "country_name": "place_of_perf_country_desc", "state_code": "place_of_performance_state", "state_name": "place_of_perfor_state_desc", "city_name": "place_of_perform_city_name", "county_name": "place_of_perform_county_na", "county_code": "place_of_perform_county_co", "zip_4a": "place_of_performance_zip4a", "congressional_code": "place_of_performance_congr", "zip_last4": "place_of_perform_zip_last4", "zip5": "place_of_performance_zip5", } legal_entity_location_field_map = { "location_country_code": "legal_entity_country_code", "country_name": "legal_entity_country_name", "state_code": "legal_entity_state_code", "state_name": "legal_entity_state_descrip", "city_name": "legal_entity_city_name", "county_name": "legal_entity_county_name", "county_code": "legal_entity_county_code", "address_line1": "legal_entity_address_line1", "address_line2": "legal_entity_address_line2", "address_line3": "legal_entity_address_line3", "zip4": "legal_entity_zip4", "congressional_code": "legal_entity_congressional", "zip_last4": "legal_entity_zip_last4", "zip5": "legal_entity_zip5", } for index, row in enumerate(to_insert, 1): upper_case_dict_values(row) # Create new LegalEntityLocation and LegalEntity from the row data legal_entity_location = create_location( legal_entity_location_field_map, row, {"recipient_flag": True, "is_fpds": True} ) recipient_name = row["awardee_or_recipient_legal"] legal_entity = LegalEntity.objects.create( recipient_unique_id=row["awardee_or_recipient_uniqu"], recipient_name=recipient_name if recipient_name is not None else "", ) legal_entity_value_map = { "location": legal_entity_location, "business_categories": get_business_categories(row=row, data_type="fpds"), "is_fpds": True, } set_legal_entity_boolean_fields(row) legal_entity = load_data_into_model(legal_entity, row, value_map=legal_entity_value_map, save=True) # Create the place of performance location pop_location = create_location(place_of_performance_field_map, row, {"place_of_performance_flag": True}) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only(row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only(row["funding_sub_tier_agency_co"]) # Generate the unique Award ID # "CONT_AW_" + agency_id + referenced_idv_agency_iden + piid + parent_award_id generated_unique_id = ( "CONT_AW_" + (row["agency_id"] if row["agency_id"] else "-NONE-") + "_" + (row["referenced_idv_agency_iden"] if row["referenced_idv_agency_iden"] else "-NONE-") + "_" + (row["piid"] if row["piid"] else "-NONE-") + "_" + (row["parent_award_id"] if row["parent_award_id"] else "-NONE-") ) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=generated_unique_id, piid=row["piid"] ) award.parent_award_piid = row.get("parent_award_id") award.save() # Append row to list of Awards updated AWARD_UPDATE_ID_LIST.append(award.id) if row["last_modified"] and len(str(row["last_modified"])) == len("YYYY-MM-DD HH:MM:SS"): # 19 characters dt_fmt = "%Y-%m-%d %H:%M:%S" else: dt_fmt = "%Y-%m-%d %H:%M:%S.%f" # try using this even if last_modified isn't a valid string try: last_mod_date = datetime.strptime(str(row["last_modified"]), dt_fmt).date() except ValueError: # handle odd-string formats and NULLs from the upstream FPDS-NG system info_message = "Invalid value '{}' does not match: '{}'".format(row["last_modified"], dt_fmt) logger.info(info_message) last_mod_date = None award_type, award_type_desc = award_types(row) parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "recipient": legal_entity, "place_of_performance": pop_location, "period_of_performance_start_date": format_date(row["period_of_performance_star"]), "period_of_performance_current_end_date": format_date(row["period_of_performance_curr"]), "action_date": format_date(row["action_date"]), "last_modified_date": last_mod_date, "transaction_unique_id": row["detached_award_proc_unique"], "generated_unique_award_id": generated_unique_id, "is_fpds": True, "type": award_type, "type_description": award_type_desc, } contract_field_map = {"description": "award_description"} transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=contract_field_map, value_map=parent_txn_value_map, as_dict=True, ) contract_instance = load_data_into_model(TransactionFPDS(), row, as_dict=True) # thrown away detached_award_proc_unique = contract_instance["detached_award_proc_unique"] unique_fpds = TransactionFPDS.objects.filter(detached_award_proc_unique=detached_award_proc_unique) if unique_fpds.first(): transaction_normalized_dict["update_date"] = datetime.now(timezone.utc) transaction_normalized_dict["fiscal_year"] = fy(transaction_normalized_dict["action_date"]) # update TransactionNormalized TransactionNormalized.objects.filter(id=unique_fpds.first().transaction.id).update( **transaction_normalized_dict ) # update TransactionFPDS unique_fpds.update(**contract_instance) else: # create TransactionNormalized transaction = TransactionNormalized(**transaction_normalized_dict) transaction.save() # create TransactionFPDS transaction_fpds = TransactionFPDS(transaction=transaction, **contract_instance) transaction_fpds.save() # Update legal entity to map back to transaction legal_entity.transaction_unique_id = detached_award_proc_unique legal_entity.save()
def business_categories(broker_input): return get_business_categories(broker_input, "fpds")
def insert_new_fabs(to_insert): fabs_normalized_field_map = { "type": "assistance_type", "description": "award_description", "funding_amount": "total_funding_amount", } fabs_field_map = { "officer_1_name": "high_comp_officer1_full_na", "officer_1_amount": "high_comp_officer1_amount", "officer_2_name": "high_comp_officer2_full_na", "officer_2_amount": "high_comp_officer2_amount", "officer_3_name": "high_comp_officer3_full_na", "officer_3_amount": "high_comp_officer3_amount", "officer_4_name": "high_comp_officer4_full_na", "officer_4_amount": "high_comp_officer4_amount", "officer_5_name": "high_comp_officer5_full_na", "officer_5_amount": "high_comp_officer5_amount", } update_award_ids = [] for row in to_insert: upper_case_dict_values(row) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only( row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only( row["funding_sub_tier_agency_co"]) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=row["unique_award_key"], fain=row["fain"], uri=row["uri"], record_type=row["record_type"], ) award.save() # Append row to list of Awards updated update_award_ids.append(award.id) try: last_mod_date = datetime.strptime(str(row["modified_at"]), "%Y-%m-%d %H:%M:%S.%f").date() except ValueError: last_mod_date = datetime.strptime(str(row["modified_at"]), "%Y-%m-%d %H:%M:%S").date() parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "period_of_performance_start_date": format_date(row["period_of_performance_star"]), "period_of_performance_current_end_date": format_date(row["period_of_performance_curr"]), "action_date": format_date(row["action_date"]), "last_modified_date": last_mod_date, "type_description": row["assistance_type_desc"], "transaction_unique_id": row["afa_generated_unique"], "business_categories": get_business_categories(row=row, data_type="fabs"), } transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=fabs_normalized_field_map, value_map=parent_txn_value_map, as_dict=True, ) financial_assistance_data = load_data_into_model( TransactionFABS(), row, field_map=fabs_field_map, as_dict=True # thrown away ) # Hack to cut back on the number of warnings dumped to the log. financial_assistance_data["updated_at"] = cast_datetime_to_utc( financial_assistance_data["updated_at"]) financial_assistance_data["created_at"] = cast_datetime_to_utc( financial_assistance_data["created_at"]) financial_assistance_data["modified_at"] = cast_datetime_to_utc( financial_assistance_data["modified_at"]) afa_generated_unique = financial_assistance_data[ "afa_generated_unique"] unique_fabs = TransactionFABS.objects.filter( afa_generated_unique=afa_generated_unique) if unique_fabs.first(): transaction_normalized_dict["update_date"] = datetime.now( timezone.utc) transaction_normalized_dict["fiscal_year"] = fy( transaction_normalized_dict["action_date"]) # Update TransactionNormalized TransactionNormalized.objects.filter( id=unique_fabs.first().transaction.id).update( **transaction_normalized_dict) # Update TransactionFABS unique_fabs.update(**financial_assistance_data) else: # Create TransactionNormalized transaction_normalized = TransactionNormalized( **transaction_normalized_dict) transaction_normalized.save() # Create TransactionFABS transaction_fabs = TransactionFABS( transaction=transaction_normalized, **financial_assistance_data) transaction_fabs.save() return update_award_ids
def insert_new_fabs(to_insert): place_of_performance_field_map = { "location_country_code": "place_of_perform_country_c", "country_name": "place_of_perform_country_n", "state_code": "place_of_perfor_state_code", "state_name": "place_of_perform_state_nam", "city_name": "place_of_performance_city", "county_name": "place_of_perform_county_na", "county_code": "place_of_perform_county_co", "foreign_location_description": "place_of_performance_forei", "zip_4a": "place_of_performance_zip4a", "congressional_code": "place_of_performance_congr", "performance_code": "place_of_performance_code", "zip_last4": "place_of_perform_zip_last4", "zip5": "place_of_performance_zip5", } legal_entity_location_field_map = { "location_country_code": "legal_entity_country_code", "country_name": "legal_entity_country_name", "state_code": "legal_entity_state_code", "state_name": "legal_entity_state_name", "city_name": "legal_entity_city_name", "city_code": "legal_entity_city_code", "county_name": "legal_entity_county_name", "county_code": "legal_entity_county_code", "address_line1": "legal_entity_address_line1", "address_line2": "legal_entity_address_line2", "address_line3": "legal_entity_address_line3", "foreign_location_description": "legal_entity_foreign_descr", "congressional_code": "legal_entity_congressional", "zip_last4": "legal_entity_zip_last4", "zip5": "legal_entity_zip5", "foreign_postal_code": "legal_entity_foreign_posta", "foreign_province": "legal_entity_foreign_provi", "foreign_city_name": "legal_entity_foreign_city", } fabs_normalized_field_map = { "type": "assistance_type", "description": "award_description", "funding_amount": "total_funding_amount", } fabs_field_map = { "officer_1_name": "high_comp_officer1_full_na", "officer_1_amount": "high_comp_officer1_amount", "officer_2_name": "high_comp_officer2_full_na", "officer_2_amount": "high_comp_officer2_amount", "officer_3_name": "high_comp_officer3_full_na", "officer_3_amount": "high_comp_officer3_amount", "officer_4_name": "high_comp_officer4_full_na", "officer_4_amount": "high_comp_officer4_amount", "officer_5_name": "high_comp_officer5_full_na", "officer_5_amount": "high_comp_officer5_amount", } update_award_ids = [] for row in to_insert: upper_case_dict_values(row) # Create new LegalEntityLocation and LegalEntity from the row data legal_entity_location = create_location( legal_entity_location_field_map, row, {"recipient_flag": True}) recipient_name = row['awardee_or_recipient_legal'] legal_entity = LegalEntity.objects.create( recipient_unique_id=row['awardee_or_recipient_uniqu'], recipient_name=recipient_name if recipient_name is not None else "", parent_recipient_unique_id=row['ultimate_parent_unique_ide'], ) legal_entity_value_map = { "location": legal_entity_location, "business_categories": get_business_categories(row=row, data_type='fabs'), "business_types_description": row['business_types_desc'], } legal_entity = load_data_into_model(legal_entity, row, value_map=legal_entity_value_map, save=True) # Create the place of performance location pop_location = create_location(place_of_performance_field_map, row, {"place_of_performance_flag": True}) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only( row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only( row["funding_sub_tier_agency_co"]) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=row['unique_award_key'], fain=row['fain'], uri=row['uri'], record_type=row['record_type'], ) award.save() # Append row to list of Awards updated update_award_ids.append(award.id) try: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S.%f").date() except ValueError: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S").date() parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "recipient": legal_entity, "place_of_performance": pop_location, "period_of_performance_start_date": format_date(row['period_of_performance_star']), "period_of_performance_current_end_date": format_date(row['period_of_performance_curr']), "action_date": format_date(row['action_date']), "last_modified_date": last_mod_date, "type_description": row['assistance_type_desc'], "transaction_unique_id": row['afa_generated_unique'], } transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=fabs_normalized_field_map, value_map=parent_txn_value_map, as_dict=True, ) financial_assistance_data = load_data_into_model( TransactionFABS(), # thrown away row, field_map=fabs_field_map, as_dict=True) # Hack to cut back on the number of warnings dumped to the log. financial_assistance_data['updated_at'] = cast_datetime_to_utc( financial_assistance_data['updated_at']) financial_assistance_data['created_at'] = cast_datetime_to_utc( financial_assistance_data['created_at']) financial_assistance_data['modified_at'] = cast_datetime_to_utc( financial_assistance_data['modified_at']) afa_generated_unique = financial_assistance_data[ 'afa_generated_unique'] unique_fabs = TransactionFABS.objects.filter( afa_generated_unique=afa_generated_unique) if unique_fabs.first(): transaction_normalized_dict["update_date"] = datetime.now( timezone.utc) transaction_normalized_dict["fiscal_year"] = fy( transaction_normalized_dict["action_date"]) # Update TransactionNormalized TransactionNormalized.objects.filter( id=unique_fabs.first().transaction.id).update( **transaction_normalized_dict) # Update TransactionFABS unique_fabs.update(**financial_assistance_data) else: # Create TransactionNormalized transaction_normalized = TransactionNormalized( **transaction_normalized_dict) transaction_normalized.save() # Create TransactionFABS transaction_fabs = TransactionFABS( transaction=transaction_normalized, **financial_assistance_data) transaction_fabs.save() # Update legal entity to map back to transaction legal_entity.transaction_unique_id = afa_generated_unique legal_entity.save() return update_award_ids
def insert_new_fpds(self, to_insert, total_rows): place_of_performance_field_map = { "location_country_code": "place_of_perform_country_c", "country_name": "place_of_perf_country_desc", "state_code": "place_of_performance_state", "state_name": "place_of_perfor_state_desc", "city_name": "place_of_perform_city_name", "county_name": "place_of_perform_county_na", "county_code": "place_of_perform_county_co", "zip_4a": "place_of_performance_zip4a", "congressional_code": "place_of_performance_congr", "zip_last4": "place_of_perform_zip_last4", "zip5": "place_of_performance_zip5", } legal_entity_location_field_map = { "location_country_code": "legal_entity_country_code", "country_name": "legal_entity_country_name", "state_code": "legal_entity_state_code", "state_name": "legal_entity_state_descrip", "city_name": "legal_entity_city_name", "county_name": "legal_entity_county_name", "county_code": "legal_entity_county_code", "address_line1": "legal_entity_address_line1", "address_line2": "legal_entity_address_line2", "address_line3": "legal_entity_address_line3", "zip4": "legal_entity_zip4", "congressional_code": "legal_entity_congressional", "zip_last4": "legal_entity_zip_last4", "zip5": "legal_entity_zip5", } for index, row in enumerate(to_insert, 1): upper_case_dict_values(row) # Create new LegalEntityLocation and LegalEntity from the row data legal_entity_location = create_location( legal_entity_location_field_map, row, { "recipient_flag": True, "is_fpds": True }) recipient_name = row["awardee_or_recipient_legal"] legal_entity = LegalEntity.objects.create( recipient_unique_id=row["awardee_or_recipient_uniqu"], recipient_name=recipient_name if recipient_name is not None else "", ) legal_entity_value_map = { "location": legal_entity_location, "business_categories": get_business_categories(row=row, data_type="fpds"), "is_fpds": True, } set_legal_entity_boolean_fields(row) legal_entity = load_data_into_model( legal_entity, row, value_map=legal_entity_value_map, save=True) # Create the place of performance location pop_location = create_location(place_of_performance_field_map, row, {"place_of_performance_flag": True}) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only( row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only( row["funding_sub_tier_agency_co"]) # Generate the unique Award ID # "CONT_AW_" + agency_id + referenced_idv_agency_iden + piid + parent_award_id generated_unique_id = ( "CONT_AW_" + (row["agency_id"] if row["agency_id"] else "-NONE-") + "_" + (row["referenced_idv_agency_iden"] if row["referenced_idv_agency_iden"] else "-NONE-") + "_" + (row["piid"] if row["piid"] else "-NONE-") + "_" + (row["parent_award_id"] if row["parent_award_id"] else "-NONE-")) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=generated_unique_id, piid=row["piid"]) award.parent_award_piid = row.get("parent_award_id") award.save() # Append row to list of Awards updated AWARD_UPDATE_ID_LIST.append(award.id) if row["last_modified"] and len(str(row["last_modified"])) == len( "YYYY-MM-DD HH:MM:SS"): # 19 characters dt_fmt = "%Y-%m-%d %H:%M:%S" else: dt_fmt = "%Y-%m-%d %H:%M:%S.%f" # try using this even if last_modified isn't a valid string try: last_mod_date = datetime.strptime(str(row["last_modified"]), dt_fmt).date() except ValueError: # handle odd-string formats and NULLs from the upstream FPDS-NG system info_message = "Invalid value '{}' does not match: '{}'".format( row["last_modified"], dt_fmt) logger.info(info_message) last_mod_date = None award_type, award_type_desc = award_types(row) parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "recipient": legal_entity, "place_of_performance": pop_location, "period_of_performance_start_date": format_date(row["period_of_performance_star"]), "period_of_performance_current_end_date": format_date(row["period_of_performance_curr"]), "action_date": format_date(row["action_date"]), "last_modified_date": last_mod_date, "transaction_unique_id": row["detached_award_proc_unique"], "generated_unique_award_id": generated_unique_id, "is_fpds": True, "type": award_type, "type_description": award_type_desc, } contract_field_map = {"description": "award_description"} transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=contract_field_map, value_map=parent_txn_value_map, as_dict=True, ) contract_instance = load_data_into_model( TransactionFPDS(), row, as_dict=True) # thrown away detached_award_proc_unique = contract_instance[ "detached_award_proc_unique"] unique_fpds = TransactionFPDS.objects.filter( detached_award_proc_unique=detached_award_proc_unique) if unique_fpds.first(): transaction_normalized_dict["update_date"] = datetime.now( timezone.utc) transaction_normalized_dict["fiscal_year"] = fy( transaction_normalized_dict["action_date"]) # update TransactionNormalized TransactionNormalized.objects.filter( id=unique_fpds.first().transaction.id).update( **transaction_normalized_dict) # update TransactionFPDS unique_fpds.update(**contract_instance) else: # create TransactionNormalized transaction = TransactionNormalized( **transaction_normalized_dict) transaction.save() # create TransactionFPDS transaction_fpds = TransactionFPDS(transaction=transaction, **contract_instance) transaction_fpds.save() # Update legal entity to map back to transaction legal_entity.transaction_unique_id = detached_award_proc_unique legal_entity.save()
def insert_new_fabs(to_insert): place_of_performance_field_map = { "location_country_code": "place_of_perform_country_c", "country_name": "place_of_perform_country_n", "state_code": "place_of_perfor_state_code", "state_name": "place_of_perform_state_nam", "city_name": "place_of_performance_city", "county_name": "place_of_perform_county_na", "county_code": "place_of_perform_county_co", "foreign_location_description": "place_of_performance_forei", "zip_4a": "place_of_performance_zip4a", "congressional_code": "place_of_performance_congr", "performance_code": "place_of_performance_code", "zip_last4": "place_of_perform_zip_last4", "zip5": "place_of_performance_zip5", } legal_entity_location_field_map = { "location_country_code": "legal_entity_country_code", "country_name": "legal_entity_country_name", "state_code": "legal_entity_state_code", "state_name": "legal_entity_state_name", "city_name": "legal_entity_city_name", "city_code": "legal_entity_city_code", "county_name": "legal_entity_county_name", "county_code": "legal_entity_county_code", "address_line1": "legal_entity_address_line1", "address_line2": "legal_entity_address_line2", "address_line3": "legal_entity_address_line3", "foreign_location_description": "legal_entity_foreign_descr", "congressional_code": "legal_entity_congressional", "zip_last4": "legal_entity_zip_last4", "zip5": "legal_entity_zip5", "foreign_postal_code": "legal_entity_foreign_posta", "foreign_province": "legal_entity_foreign_provi", "foreign_city_name": "legal_entity_foreign_city", } update_award_ids = [] for row in to_insert: upper_case_dict_values(row) # Create new LegalEntityLocation and LegalEntity from the row data legal_entity_location = create_location(legal_entity_location_field_map, row, {"recipient_flag": True}) recipient_name = row['awardee_or_recipient_legal'] legal_entity = LegalEntity.objects.create( recipient_unique_id=row['awardee_or_recipient_uniqu'], recipient_name=recipient_name if recipient_name is not None else "", parent_recipient_unique_id=row['ultimate_parent_unique_ide'], ) legal_entity_value_map = { "location": legal_entity_location, "business_categories": get_business_categories(row=row, data_type='fabs'), "business_types_description": row['business_types_desc'], } legal_entity = load_data_into_model(legal_entity, row, value_map=legal_entity_value_map, save=True) # Create the place of performance location pop_location = create_location(place_of_performance_field_map, row, {"place_of_performance_flag": True}) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only(row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only(row["funding_sub_tier_agency_co"]) # Generate the unique Award ID # "ASST_AW_" + awarding_sub_tier_agency_c + fain + uri # this will raise an exception if the cast to an int fails, that's ok since we don't want to process # non-numeric record type values record_type_int = int(row['record_type']) if record_type_int == 1: uri = row['uri'] if row['uri'] else '-NONE-' fain = '-NONE-' elif record_type_int in (2, 3): uri = '-NONE-' fain = row['fain'] if row['fain'] else '-NONE-' else: msg = "Invalid record type encountered for the following afa_generated_unique record: {}" raise Exception(msg.format(row['afa_generated_unique'])) astac = row["awarding_sub_tier_agency_c"] if row["awarding_sub_tier_agency_c"] else "-NONE-" generated_unique_id = "ASST_AW_{}_{}_{}".format(astac, fain, uri) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=generated_unique_id, fain=row['fain'], uri=row['uri'], record_type=row['record_type'], ) award.save() # Append row to list of Awards updated update_award_ids.append(award.id) try: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S.%f").date() except ValueError: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S").date() parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "recipient": legal_entity, "place_of_performance": pop_location, "period_of_performance_start_date": format_date(row['period_of_performance_star']), "period_of_performance_current_end_date": format_date(row['period_of_performance_curr']), "action_date": format_date(row['action_date']), "last_modified_date": last_mod_date, "type_description": row['assistance_type_desc'], "transaction_unique_id": row['afa_generated_unique'], "generated_unique_award_id": generated_unique_id, } fad_field_map = { "type": "assistance_type", "description": "award_description", "funding_amount": "total_funding_amount", } transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=fad_field_map, value_map=parent_txn_value_map, as_dict=True, ) financial_assistance_data = load_data_into_model(TransactionFABS(), row, as_dict=True) # thrown away # Hack to cut back on the number of warnings dumped to the log. financial_assistance_data['updated_at'] = cast_datetime_to_utc(financial_assistance_data['updated_at']) financial_assistance_data['created_at'] = cast_datetime_to_utc(financial_assistance_data['created_at']) financial_assistance_data['modified_at'] = cast_datetime_to_utc(financial_assistance_data['modified_at']) afa_generated_unique = financial_assistance_data['afa_generated_unique'] unique_fabs = TransactionFABS.objects.filter(afa_generated_unique=afa_generated_unique) if unique_fabs.first(): transaction_normalized_dict["update_date"] = datetime.now(timezone.utc) transaction_normalized_dict["fiscal_year"] = fy(transaction_normalized_dict["action_date"]) # Update TransactionNormalized TransactionNormalized.objects.filter(id=unique_fabs.first().transaction.id).update( **transaction_normalized_dict ) # Update TransactionFABS unique_fabs.update(**financial_assistance_data) else: # Create TransactionNormalized transaction_normalized = TransactionNormalized(**transaction_normalized_dict) transaction_normalized.save() # Create TransactionFABS transaction_fabs = TransactionFABS(transaction=transaction_normalized, **financial_assistance_data) transaction_fabs.save() # Update legal entity to map back to transaction legal_entity.transaction_unique_id = afa_generated_unique legal_entity.save() return update_award_ids
def test_dev_2318_business_category_changes(): """ The following changes were made to business categories as part of DEV-2318. Let's make sure they took. - alaskan_native_owned_business => alaskan_native_corporation_owned_firm - foreign_owned_and_located_business => foreign_owned - native_hawaiian_owned_business => native_hawaiian_organization_owned_firm - tribally_owned_business => tribally_owned_firm """ expected_business_categories = [ "alaskan_native_corporation_owned_firm", "foreign_owned", "minority_owned_business", "native_hawaiian_organization_owned_firm", "special_designations", "tribally_owned_firm", ] business_categories = get_business_categories( { "alaskan_native_owned_corpo": "true", "native_hawaiian_owned_busi": "true", "tribally_owned_business": "true", "foreign_owned_and_located": "true", }, "fpds", ) assert business_categories == expected_business_categories # Same test using the database function. with connection.cursor() as cursor: cursor.execute(""" select compile_fpds_business_categories( '', -- contracting_officers_deter false, -- corporate_entity_tax_exemp false, -- corporate_entity_not_tax_e false, -- partnership_or_limited_lia false, -- sole_proprietorship false, -- manufacturer_of_goods false, -- subchapter_s_corporation false, -- limited_liability_corporat false, -- for_profit_organization TRUE, -- alaskan_native_owned_corpo false, -- american_indian_owned_busi false, -- asian_pacific_american_own false, -- black_american_owned_busin false, -- hispanic_american_owned_bu false, -- native_american_owned_busi TRUE, -- native_hawaiian_owned_busi false, -- subcontinent_asian_asian_i TRUE, -- tribally_owned_business false, -- other_minority_owned_busin false, -- minority_owned_business false, -- women_owned_small_business false, -- economically_disadvantaged false, -- joint_venture_women_owned false, -- joint_venture_economically false, -- woman_owned_business false, -- service_disabled_veteran_o false, -- veteran_owned_business false, -- c8a_program_participant false, -- the_ability_one_program false, -- dot_certified_disadvantage false, -- emerging_small_business false, -- federally_funded_research false, -- historically_underutilized false, -- labor_surplus_area_firm false, -- sba_certified_8_a_joint_ve false, -- self_certified_small_disad false, -- small_agricultural_coopera false, -- small_disadvantaged_busine false, -- community_developed_corpor '', -- domestic_or_foreign_entity TRUE, -- foreign_owned_and_located false, -- foreign_government false, -- international_organization false, -- domestic_shelter false, -- hospital_flag false, -- veterinary_hospital false, -- foundation false, -- community_development_corp false, -- nonprofit_organization false, -- educational_institution false, -- other_not_for_profit_organ false, -- state_controlled_instituti false, -- c1862_land_grant_college false, -- c1890_land_grant_college false, -- c1994_land_grant_college false, -- private_university_or_coll false, -- minority_institution false, -- historically_black_college false, -- tribal_college false, -- alaskan_native_servicing_i false, -- native_hawaiian_servicing false, -- hispanic_servicing_institu false, -- school_of_forestry false, -- veterinary_college false, -- us_federal_government false, -- federal_agency false, -- us_government_entity false, -- interstate_entity false, -- us_state_government false, -- council_of_governments false, -- city_local_government false, -- county_local_government false, -- inter_municipal_local_gove false, -- municipality_local_governm false, -- township_local_government false, -- us_local_government false, -- local_government_owned false, -- school_district_local_gove false, -- us_tribal_government false, -- indian_tribe_federally_rec false, -- housing_authorities_public false, -- airport_authority false, -- port_authority false, -- transit_authority false -- planning_commission ) """) assert cursor.fetchall()[0][0] == expected_business_categories display_names = get_business_category_display_names( business_categories) assert display_names == [ "Alaskan Native Corporation Owned Firm", "Foreign Owned", "Minority Owned Business", "Native Hawaiian Organization Owned Firm", "Special Designations", "Tribally Owned Firm", ]
def insert_new_fabs(to_insert): place_of_performance_field_map = { "location_country_code": "place_of_perform_country_c", "country_name": "place_of_perform_country_n", "state_code": "place_of_perfor_state_code", "state_name": "place_of_perform_state_nam", "city_name": "place_of_performance_city", "county_name": "place_of_perform_county_na", "county_code": "place_of_perform_county_co", "foreign_location_description": "place_of_performance_forei", "zip_4a": "place_of_performance_zip4a", "congressional_code": "place_of_performance_congr", "performance_code": "place_of_performance_code", "zip_last4": "place_of_perform_zip_last4", "zip5": "place_of_performance_zip5", } legal_entity_location_field_map = { "location_country_code": "legal_entity_country_code", "country_name": "legal_entity_country_name", "state_code": "legal_entity_state_code", "state_name": "legal_entity_state_name", "city_name": "legal_entity_city_name", "city_code": "legal_entity_city_code", "county_name": "legal_entity_county_name", "county_code": "legal_entity_county_code", "address_line1": "legal_entity_address_line1", "address_line2": "legal_entity_address_line2", "address_line3": "legal_entity_address_line3", "foreign_location_description": "legal_entity_foreign_descr", "congressional_code": "legal_entity_congressional", "zip_last4": "legal_entity_zip_last4", "zip5": "legal_entity_zip5", "foreign_postal_code": "legal_entity_foreign_posta", "foreign_province": "legal_entity_foreign_provi", "foreign_city_name": "legal_entity_foreign_city", } update_award_ids = [] for row in to_insert: upper_case_dict_values(row) # Create new LegalEntityLocation and LegalEntity from the row data legal_entity_location = create_location( legal_entity_location_field_map, row, {"recipient_flag": True}) recipient_name = row['awardee_or_recipient_legal'] legal_entity = LegalEntity.objects.create( recipient_unique_id=row['awardee_or_recipient_uniqu'], recipient_name=recipient_name if recipient_name is not None else "", parent_recipient_unique_id=row['ultimate_parent_unique_ide'], ) legal_entity_value_map = { "location": legal_entity_location, "business_categories": get_business_categories(row=row, data_type='fabs'), "business_types_description": row['business_types_desc'], } legal_entity = load_data_into_model(legal_entity, row, value_map=legal_entity_value_map, save=True) # Create the place of performance location pop_location = create_location(place_of_performance_field_map, row, {"place_of_performance_flag": True}) # Find the toptier awards from the subtier awards awarding_agency = Agency.get_by_subtier_only( row["awarding_sub_tier_agency_c"]) funding_agency = Agency.get_by_subtier_only( row["funding_sub_tier_agency_co"]) # Generate the unique Award ID # "ASST_AW_" + awarding_sub_tier_agency_c + fain + uri # this will raise an exception if the cast to an int fails, that's ok since we don't want to process # non-numeric record type values record_type_int = int(row['record_type']) if record_type_int == 1: uri = row['uri'] if row['uri'] else '-NONE-' fain = '-NONE-' elif record_type_int in (2, 3): uri = '-NONE-' fain = row['fain'] if row['fain'] else '-NONE-' else: msg = "Invalid record type encountered for the following afa_generated_unique record: {}" raise Exception(msg.format(row['afa_generated_unique'])) astac = row["awarding_sub_tier_agency_c"] if row[ "awarding_sub_tier_agency_c"] else "-NONE-" generated_unique_id = "ASST_AW_{}_{}_{}".format(astac, fain, uri) # Create the summary Award (created, award) = Award.get_or_create_summary_award( generated_unique_award_id=generated_unique_id, fain=row['fain'], uri=row['uri'], record_type=row['record_type'], ) award.save() # Append row to list of Awards updated update_award_ids.append(award.id) try: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S.%f").date() except ValueError: last_mod_date = datetime.strptime(str(row['modified_at']), "%Y-%m-%d %H:%M:%S").date() parent_txn_value_map = { "award": award, "awarding_agency": awarding_agency, "funding_agency": funding_agency, "recipient": legal_entity, "place_of_performance": pop_location, "period_of_performance_start_date": format_date(row['period_of_performance_star']), "period_of_performance_current_end_date": format_date(row['period_of_performance_curr']), "action_date": format_date(row['action_date']), "last_modified_date": last_mod_date, "type_description": row['assistance_type_desc'], "transaction_unique_id": row['afa_generated_unique'], "generated_unique_award_id": generated_unique_id, } fad_field_map = { "type": "assistance_type", "description": "award_description", "funding_amount": "total_funding_amount", } transaction_normalized_dict = load_data_into_model( TransactionNormalized(), # thrown away row, field_map=fad_field_map, value_map=parent_txn_value_map, as_dict=True, ) financial_assistance_data = load_data_into_model( TransactionFABS(), row, as_dict=True) # thrown away # Hack to cut back on the number of warnings dumped to the log. financial_assistance_data['updated_at'] = cast_datetime_to_utc( financial_assistance_data['updated_at']) financial_assistance_data['created_at'] = cast_datetime_to_utc( financial_assistance_data['created_at']) financial_assistance_data['modified_at'] = cast_datetime_to_utc( financial_assistance_data['modified_at']) afa_generated_unique = financial_assistance_data[ 'afa_generated_unique'] unique_fabs = TransactionFABS.objects.filter( afa_generated_unique=afa_generated_unique) if unique_fabs.first(): transaction_normalized_dict["update_date"] = datetime.now( timezone.utc) transaction_normalized_dict["fiscal_year"] = fy( transaction_normalized_dict["action_date"]) # Update TransactionNormalized TransactionNormalized.objects.filter( id=unique_fabs.first().transaction.id).update( **transaction_normalized_dict) # Update TransactionFABS unique_fabs.update(**financial_assistance_data) else: # Create TransactionNormalized transaction_normalized = TransactionNormalized( **transaction_normalized_dict) transaction_normalized.save() # Create TransactionFABS transaction_fabs = TransactionFABS( transaction=transaction_normalized, **financial_assistance_data) transaction_fabs.save() # Update legal entity to map back to transaction legal_entity.transaction_unique_id = afa_generated_unique legal_entity.save() return update_award_ids