chore(base): add boundary check for transactions

src/monopoly/config.py

@@ -37,10 +37,16 @@ class StatementConfig:
     - `transaction_date_order` represents the datetime format that a specific bank uses
     for transactions. For example, "DMY" will parse 01/02/2024 as 1 Feb 2024.
     Defaults to DMY.
+    - `statement_date_format` represents the datetime format that a specific bank uses
+    to represent a statement date.
     - `multiline_transactions` controls whether Monopoly tries to concatenate
     transactions that are split across two lines
     - `header_pattern` is a regex pattern that is used to find the 'header' line
     of a statement, and determine if it is a debit or credit card statement.
+    - `transaction_bound` will cause transactions that have an amount past a certain
+    number of spaces will be ignored. For example, if `transaction_bound` = 5:
+        "01 NOV  BALANCE B/F                       190.77" (will be ignored)
+        "01 NOV  YA KUN KAYA TOAST  12.00" (will be kept)
     """
 
     bank_name: BankNames | InternalBankNames
@@ -51,6 +57,7 @@ class StatementConfig:
     transaction_date_order: DateOrder = field(default_factory=lambda: DateOrder("DMY"))
     statement_date_order: DateOrder = field(default_factory=lambda: DateOrder("DMY"))
     multiline_transactions: bool = False
+    transaction_bound: Optional[int] = None
     has_withdraw_deposit_column: bool = False
     prev_balance_pattern: Optional[Pattern[str] | RegexEnum] = None
 

src/monopoly/statements/base.py

@@ -71,6 +71,9 @@ def get_transactions(self) -> list[Transaction] | None:
         for page_num, page in enumerate(self.pages):
             for line_num, line in enumerate(page.lines):
                 if match := self.pattern.search(line):
+                    if self._check_bound(match):
+                        continue
+
                     groupdict = TransactionGroupDict(**match.groupdict())
                     transaction_match = TransactionMatch(
                         groupdict, match, page_number=page_num
@@ -91,6 +94,13 @@ def get_transactions(self) -> list[Transaction] | None:
         post_processed_transactions = self.post_process_transactions(transactions)
         return post_processed_transactions
 
+    def _check_bound(self, match: re.Match):
+        if bound := self.config.transaction_bound:
+            if match.span(Columns.AMOUNT)[0] >= bound:
+                logger.debug("Transaction exists beyond boundary, ignoring")
+                return True
+        return False
+
     def pre_process_match(
         self, transaction_match: TransactionMatch
     ) -> TransactionMatch:

tests/unit/test_statement_process_line.py

@@ -35,6 +35,27 @@ def test_get_transactions(statement: BaseStatement):
     assert transactions == expected
 
 
+def test_check_bound(statement: BaseStatement):
+    pattern = Ocbc.credit_config.transaction_pattern
+    statement.config.transaction_pattern = pattern
+    statement.config.transaction_bound = 30
+
+    statement.pages = [
+        PdfPage("19/06 YA KUN KAYA TOAST 3.20\n20/06 FAIRPRICE FINEST             9.90")
+    ]
+    transactions = statement.get_transactions()
+    expected = [
+        Transaction(
+            transaction_date="19/06",
+            description="YA KUN KAYA TOAST",
+            amount=-3.2,
+            suffix=None,
+        ),
+    ]
+    statement.config.transaction_bound = None
+    assert transactions == expected
+
+
 def test_get_multiline_transactions(statement: BaseStatement):
     pattern = Hsbc.credit_config.transaction_pattern
     statement.config.multiline_transactions = True