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').first()
if d_business_cat:
duns_types_mapping = get_duns_business_types_mapping()
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:
business_categories |= set(le_business_cat['business_categories'])
return list(business_categories)
def insert_new_fpds(self, to_insert, total_rows):
logger.info('Starting insertion of new FPDS data')
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"
}
start_time = datetime.now()
for index, row in enumerate(to_insert, 1):
if not (index % 1000):
logger.info(
'Inserting Stale FPDS: Inserting row {} of {} ({})'.format(
str(index), str(total_rows),
datetime.now() - start_time))
for key in row:
if isinstance(row[key], str):
row[key] = row[key].upper()
# 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)
try:
last_mod_date = datetime.strptime(str(
row['last_modified']), "%Y-%m-%d %H:%M:%S.%f").date()
except ValueError:
last_mod_date = datetime.strptime(str(row['last_modified']),
"%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,
"transaction_unique_id":
row['detached_award_proc_unique'],
"generated_unique_award_id":
generated_unique_id,
"is_fpds":
True
}
contract_field_map = {
"type": "contract_award_type",
"type_description": "contract_award_type_desc",
"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(), # thrown away
row,
as_dict=True)
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.utcnow()
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()
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)
try:
last_mod_date = datetime.strptime(str(
row["last_modified"]), "%Y-%m-%d %H:%M:%S.%f").date()
except ValueError:
last_mod_date = datetime.strptime(str(row["last_modified"]),
"%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,
"transaction_unique_id":
row["detached_award_proc_unique"],
"generated_unique_award_id":
generated_unique_id,
"is_fpds":
True,
}
contract_field_map = {
"type": "contract_award_type",
"type_description": "contract_award_type_desc",
"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(self, 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",
}
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
AWARD_UPDATE_ID_LIST.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
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 = TransactionNormalized(
**transaction_normalized_dict)
transaction.save()
# Create TransactionFABS
transaction_fabs = TransactionFABS(transaction=transaction,
**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()
def insert_new_fabs(self, to_insert, total_rows):
logger.info('Starting insertion of new FABS data')
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"
}
start_time = datetime.now()
for index, row in enumerate(to_insert, 1):
if not (index % 1000):
logger.info('Inserting Stale FABS: Inserting row {} of {} ({})'.format(str(index), str(total_rows),
datetime.now() - start_time))
for key in row:
if isinstance(row[key], str):
row[key] = row[key].upper()
# 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 ""
)
legal_entity_value_map = {
"location": legal_entity_location,
"business_categories": get_business_categories(row=row, data_type='fabs'),
"business_types_description": get_business_type_description(row['business_types'])
}
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 == 2:
uri = '-NONE-'
fain = row['fain'] if row['fain'] else '-NONE-'
else:
raise Exception('Invalid record type encountered for the following afa_generated_unique record: %s' %
row['afa_generated_unique'])
generated_unique_id = 'ASST_AW_' +\
(row['awarding_sub_tier_agency_c'] if row['awarding_sub_tier_agency_c'] else '-NONE-') + '_' + \
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
award_update_id_list.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": get_assistance_type_description(row['assistance_type']),
"transaction_unique_id": row['afa_generated_unique'],
"generated_unique_award_id": generated_unique_id
}
fad_field_map = {
"type": "assistance_type",
"description": "award_description",
}
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(), # thrown away
row,
as_dict=True)
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.utcnow()
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 = TransactionNormalized(**transaction_normalized_dict)
transaction.save()
# Create TransactionFABS
transaction_fabs = TransactionFABS(transaction=transaction, **financial_assistance_data)
transaction_fabs.save()