import pandas as pd import pytest from pdf_parser import DBSCreditCardStatementParser def test_dbs_credit_card_statement_parser(): # The path to a sample PDF file to use for testing sample_pdf_path = "tests/sample_dbs_statement.pdf" # Initialize the parser parser = DBSCreditCardStatementParser(sample_pdf_path) # Parse the PDF file df = parser.parse() # Check the DataFrame's columns assert list(df.columns) == [ "Card Last 4 Digits", "TRANS DATE", "POST DATE", "DESCRIPTION", "AMOUNT HKD", "RUNNING BALANCE", ] # Check the DataFrame's number of rows # (Replace 21 with the actual number of transactions in the sample PDF file) assert len(df) == 21