feat(generic): support inconsistent header spacing across pages

src/monopoly/generic/generic.py

@@ -155,9 +155,10 @@ def get_debit_statement_header_line(self, lines_before_first_transaction) -> str
             if line:
                 # assume that the header always starts with `date` or `DATE`
                 result = header_pattern.search(line)
-
                 if result:
-                    escaped_result = f"({re.escape(result.string)})"
+                    generalized_result = re.sub(r"\s{3,}", r"\\s+", result.string)
+                    escaped_result = re.escape(generalized_result)
+                    escaped_result = escaped_result.replace("\\\\s\\+", r"\s+")
                     logger.debug("Found header statement: %s", escaped_result)
                     return escaped_result
 

src/monopoly/statements/debit_statement.py

@@ -110,9 +110,10 @@ def get_header_pos(self, column_name: str, page_number: int) -> int:
                     continue
                 return header_start_pos + len(column_name)
 
-        raise ValueError(
+        logger.debug(
             f"Debit header {column_name} cannot be found on page {page_number}"
         )
+        return -1
 
     @lru_cache
     def perform_safety_check(self) -> bool:

tests/integration/banks/test_banks_generic_debit.py

@@ -6,6 +6,7 @@
 from test_utils.transactions import get_transactions_as_dict, read_transactions_from_csv
 
 from monopoly.banks import BankBase, Dbs, Maybank, Ocbc
+from monopoly.generic import GenericBank
 from monopoly.pdf import PdfDocument, PdfParser
 from monopoly.pipeline import Pipeline
 from monopoly.statements import DebitStatement
@@ -17,17 +18,11 @@
 ]
 
 
-@pytest.fixture
-def no_banks(monkeypatch):
-    monkeypatch.setattr("monopoly.banks.banks", [])
-
-
 @skip_if_encrypted
 @pytest.mark.parametrize(
     "bank, expected_debit_sum, expected_credit_sum, statement_date",
     test_cases,
 )
-@pytest.mark.usefixtures("no_banks")
 def test_bank_debit_statements(
     bank: BankBase,
     expected_debit_sum: float,
@@ -38,6 +33,7 @@ def test_bank_debit_statements(
 
     document = PdfDocument(test_directory / "input.pdf")
     parser = PdfParser(bank, document)
+    parser.bank = GenericBank
     pipeline = Pipeline(parser)
     statement: DebitStatement = pipeline.extract()
 

tests/unit/generic/test_date_pattern_analyzer.py

@@ -349,14 +349,15 @@ def __init__(self, lines):
     pages = [MockPdfPage(lines_before_first_transaction)]
     date_pattern_analyzer.pages = pages
     date_pattern_analyzer.matches = [MockDateMatch(line_number=22, page_number=0)]
-    expected_header = re.escape(
-        " DATE    DETAILS OF TRANSACTIONS     WITHDRAWAL($)    DEPOSIT($)     BALANCE($)"
+    expected_header = (
+        "\\ DATE\\s+DETAILS\\ OF\\ TRANSACTIONS\\s+WITHDRAWAL\\(\\$\\)"
+        + "\\s+DEPOSIT\\(\\$\\)\\s+BALANCE\\(\\$\\)"
     )
 
     result = date_pattern_analyzer.get_debit_statement_header_line(
-        frozenset(lines_before_first_transaction)
+        lines_before_first_transaction
     )
-    assert result == f"({expected_header})"
+    assert result == expected_header
 
 
 def test_check_if_multiline(date_pattern_analyzer: DatePatternAnalyzer):