feat(banks): add bank of america combined statement

src/monopoly/banks/__init__.py

@@ -1,6 +1,7 @@
 import logging
 from typing import Type
 
+from .bank_of_america import BankOfAmerica
 from .base import BankBase
 from .chase import Chase
 from .citibank import Citibank
@@ -15,6 +16,7 @@
 from .zkb import ZurcherKantonalBank
 
 banks: list[Type["BankBase"]] = [
+    BankOfAmerica,
     Chase,
     Citibank,
     Dbs,

src/monopoly/banks/bank_of_america/__init__.py

@@ -0,0 +1,3 @@
+from .boa import BankOfAmerica
+
+__all__ = ["BankOfAmerica"]

src/monopoly/banks/bank_of_america/boa.py

@@ -0,0 +1,39 @@
+import logging
+from re import compile as regex
+
+from monopoly.config import DateOrder, StatementConfig
+from monopoly.constants import BankNames, CreditTransactionPatterns, EntryType
+from monopoly.constants.date import ISO8601
+from monopoly.identifiers import MetadataIdentifier
+
+from ..base import BankBase
+
+logger = logging.getLogger(__name__)
+
+
+class BankOfAmerica(BankBase):
+    name = BankNames.BANK_OF_AMERICA
+
+    credit = StatementConfig(
+        statement_type=EntryType.CREDIT,
+        statement_date_pattern=regex(rf"for .* to {ISO8601.MMMM_DD_YYYY}"),
+        statement_date_order=DateOrder("MDY"),
+        transaction_date_order=DateOrder("MDY"),
+        header_pattern=regex(r"(Date.*Description.*Amount)"),
+        transaction_pattern=CreditTransactionPatterns.BANK_OF_AMERICA,
+        multiline_transactions=True,
+        safety_check=False,
+        auto_polarity=False,
+    )
+
+    identifiers = [
+        [
+            MetadataIdentifier(
+                format="PDF 1.5",
+                creator="Bank of America",
+                producer="TargetStream StreamEDS",
+            )
+        ]
+    ]
+
+    statement_configs = [credit]

src/monopoly/config.py

@@ -47,6 +47,9 @@ class StatementConfig:
     number of spaces will be ignored. For example, if `transaction_bound` = 32:
         "01 NOV  BALANCE B/F              190.77" (will be ignored)
         "01 NOV  YA KUN KAYA TOAST  12.00       " (will be kept)
+    - `auto_polarity` controls whether transaction amounts are set as negative.
+    or positive if they have 'CR' or '+' as a suffix. Enabled by default.
+    If enabled, only 'CR' or '+' will make a transaction positive. Disabled by default.
     - `safety_check` controls whether the safety check for banks. Use
     for banks that don't provide total amount (or total debit/credit)
     in the statement. Enabled by default.
@@ -62,6 +65,7 @@ class StatementConfig:
     transaction_bound: Optional[int] = None
     prev_balance_pattern: Optional[Pattern[str] | RegexEnum] = None
     safety_check: bool = True
+    auto_polarity: bool = True
 
 
 @dataclass

src/monopoly/constants/date.py

@@ -37,6 +37,7 @@ class ISO8601(RegexEnum):
         rf"\b({DateFormats.DD}[-\s]{DateFormats.MMM}[,\s]{{1,2}}{DateFormats.YYYY})"
     )
     MM_DD = rf"\b({DateFormats.MM}[\/\-\s]{DateFormats.DD})"
+    MM_DD_YY = rf"\b({DateFormats.MM}[\/\-\s]{DateFormats.DD}[\/\-\s]{DateFormats.YY})"
     MMMM_DD_YYYY = (
         rf"\b({DateFormats.MMMM}\s{DateFormats.DD}[,\s]{{1,2}}{DateFormats.YYYY})"
     )

src/monopoly/constants/statement.py

@@ -15,6 +15,7 @@ class EntryType(AutoEnum):
 
 
 class BankNames(AutoEnum):
+    BANK_OF_AMERICA = auto()
     CHASE = auto()
     CITIBANK = auto()
     DBS = auto()
@@ -42,7 +43,7 @@ class Columns(AutoEnum):
 class SharedPatterns(StrEnum):
     """
     AMOUNT matches the following patterns:
-    1,123.12 | 123.12 | (123.12) | ( 123.12) | 123.12 CR
+    1,123.12 | 123.12 | (123.12) | ( 123.12) | 123.12 CR | -1,123.12
 
     AMOUNT_EXTENDED is generally used for credit statements and to
     find statement balances, while AMOUNT_EXTENDED_WITHOUT_EOL is used
@@ -52,9 +53,10 @@ class SharedPatterns(StrEnum):
 
     COMMA_FORMAT = r"\d{1,3}(,\d{3})*\.\d*"
     ENCLOSED_COMMA_FORMAT = rf"\({COMMA_FORMAT}\s{{0,1}}\))"
+    OPTIONAL_NEGATIVE_SYMBOL = r"(?:-)?"
     DEBIT_CREDIT_SUFFIX = r"(?P<suffix>CR\b|DR\b|\+|\-)?\s*"
 
-    AMOUNT = rf"(?P<amount>{COMMA_FORMAT}|{ENCLOSED_COMMA_FORMAT}\s*"
+    AMOUNT = rf"(?P<amount>{OPTIONAL_NEGATIVE_SYMBOL}{COMMA_FORMAT}|{ENCLOSED_COMMA_FORMAT}\s*"
     AMOUNT_EXTENDED_WITHOUT_EOL = AMOUNT + DEBIT_CREDIT_SUFFIX
     AMOUNT_EXTENDED = AMOUNT_EXTENDED_WITHOUT_EOL + r"$"
 
@@ -99,6 +101,11 @@ class StatementBalancePatterns(RegexEnum):
 
 
 class CreditTransactionPatterns(RegexEnum):
+    BANK_OF_AMERICA = (
+        rf"(?P<transaction_date>{ISO8601.MM_DD_YY})\s+"
+        + SharedPatterns.DESCRIPTION
+        + SharedPatterns.AMOUNT_EXTENDED
+    )
     DBS = (
         rf"(?P<transaction_date>{ISO8601.DD_MMM})\s+"
         + SharedPatterns.DESCRIPTION

src/monopoly/statements/base.py

@@ -88,7 +88,10 @@ def get_transactions(self) -> list[Transaction] | None:
                         lines=page.lines,
                         idx=line_num,
                     )
-                    transaction = Transaction(**processed_match.groupdict)
+                    transaction = Transaction(
+                        **processed_match.groupdict,
+                        auto_polarity=self.config.auto_polarity,
+                    )
                     transactions.append(transaction)
 
         if not transactions:

src/monopoly/statements/transaction.py

@@ -85,6 +85,9 @@ class Transaction:
     amount: float
     date: str = Field(alias="transaction_date")
     suffix: Optional[str] = None
+    # avoid storing config logic, since the Transaction object is used to create
+    # a single unique hash which should not change
+    auto_polarity: bool = Field(default=True, init=True, repr=False)
 
     def as_raw_dict(self, show_suffix=False):
         """Returns stringified dictionary version of the transaction"""
@@ -138,6 +141,9 @@ def convert_credit_amount_to_negative(self: "Transaction") -> "Transaction":
         if self.amount == 0:
             return self
 
+        if not self.auto_polarity:
+            return
+
         if self.suffix in ("CR", "+"):
             self.amount = abs(self.amount)