def validate_ids_unique(data):
seen = set()
for row in data:
row_id = row["id"]
if row_id in seen:
print(row)
critical_exit(f"Duplicate id found {row_id}")
else:
seen.add(row_id)
def add_missing_ids(data):
for row in data:
if "id" not in row or len(row["id"]) == 0:
row["id"] = gen_id(row)
print("Added id: " + row["id"])
if "name" not in row:
print(row)
data_builder.critical_exit(
"this row is broken with no name? (missing ###):")
return data
def add_missing_ids():
data = data_builder.read_all_data()
for row in data:
if "id" not in row or len(row["id"]) == 0:
row["id"] = gen_id(row)
print("Added id: " + row["id"])
if "name" not in row:
print(row)
critical_exit("this row is broken with no name? (missing ###):")
validate_ids_unique(data)
rewrite_data(data)
def gen_id(row):
state = row['state']
state_abbrev = us_state_to_abbrev[state].lower()
city = row['city']
city_abbrev = city.replace(' ', '').replace('.', '').lower()
if len(city_abbrev) == 0:
if state_abbrev == unknown_location_acronym:
city_abbrev = unknown_location_acronym
elif state_abbrev == 'dc':
city_abbrev = 'dc'
else:
critical_exit("invalid city abbreviation, exiting")
# id_line = f'id: {state_abbrev}-{city_abbrev}-{city_index}'
return f"{state_abbrev}-{city_abbrev}-{random_chars(4)}"
def gen_id(row):
state = row["state"]
state_abbrev = us_state_to_abbrev[state].lower()
city = row["city"]
city_abbrev = city.replace(" ", "").replace(".", "").lower()
if len(city_abbrev) == 0:
if state_abbrev == unknown_location_acronym:
city_abbrev = unknown_location_acronym
elif state_abbrev == "dc":
city_abbrev = "dc"
else:
data_builder.critical_exit("invalid city abbreviation, exiting")
# id_line = f'id: {state_abbrev}-{city_abbrev}-{city_index}'
return f"{state_abbrev}-{city_abbrev}-{random_chars(4)}"
