import decimal
import pandas

csv_path = ''
parquet_path = ''
rates = ['activityRate', 'unitPriceUSD', 'withholdingTaxUSD']

headers = [
    'txnId',
    'activityRate',
    'countryId',
    'isrc',
    'labelId',
    'quantity',
    'saleCurrencyId',
    'storeId',
    'transactionDate',
    'transactionTypeId',
    'unitPriceUSD',
    'upc',
    'withholdingTaxUSD'
]

schema = {
    'txnId': 'int64',
    'activityRate': 'object',
    'countryId': 'int32',
    'isrc': 'object',
    'labelId': 'int32',
    'quantity': 'int32',
    'saleCurrencyId': 'int32',
    'storeId': 'int32',
    'transactionDate': 'object',
    'transactionTypeId': 'int32',
    'unitPriceUSD': 'object',
    'upc': 'int64',
    'withholdingTaxUSD': 'object'
}

df = pandas.read_csv(csv_path, header=None, names=headers, delimiter='\t')
df.dtypes
df = df.astype(schema)
df.dtypes

for rate in rates:
    df[rate] = df[rate].astype(str).map(decimal.Decimal)

df.dtypes

df.to_parquet(parquet_path)