From c654bb116f9e140d786ba8c5993b7071fe887dc7 Mon Sep 17 00:00:00 2001
From: Keenan Tims <ktims@gotroot.ca>
Date: Sun, 23 Nov 2025 19:06:27 -0800
Subject: [PATCH] Add ScotiaBank parser

---
 parsers.py | 100 ++++++++++++++++++++++++++++++++++++++---------------
 1 file changed, 72 insertions(+), 28 deletions(-)

diff --git a/parsers.py b/parsers.py
index de2c4b7..a77fcb3 100644
--- a/parsers.py
+++ b/parsers.py
@@ -3,8 +3,8 @@ from abc import ABC, abstractmethod
 from datetime import date, datetime
 from decimal import Decimal
 from email.message import EmailMessage, Message
-from logging import info
-from typing import Optional
+from logging import info, warning
+from typing import Any, Optional, Sequence
 from uuid import UUID
 
 from bs4 import BeautifulSoup
@@ -46,6 +46,18 @@ class TransactionParser(ABC):
         """
         pass
 
+    @staticmethod
+    def get_content(
+        msg: EmailMessage, preferencelist: Sequence[str] = ("html", "plain")
+    ) -> Any:
+        body = msg.get_body(preferencelist=preferencelist)
+        if body is None:
+            raise TransactionParsingFailed("No body of message found")
+        content = body.get_content()
+        if content is None:
+            raise TransactionParsingFailed("No content of message found")
+        return content
+
 
 class TransactionParsingFailed(Exception):
     pass
@@ -66,12 +78,7 @@ class RogersBankParser(TransactionParser):
         )
 
     def extract(self, msg: EmailMessage) -> Optional[Transaction]:
-        body = msg.get_body()
-        if body is None:
-            raise TransactionParsingFailed("No body of message found")
-        content = body.get_content()
-        if content is None:
-            raise TransactionParsingFailed("No content of message found")
+        content = self.get_content(msg)
         matches = self.EXTRACT_RE.search(content)
         if matches is None:
             raise TransactionParsingFailed("No matches for extraction RE")
@@ -110,12 +117,7 @@ class MBNAParser(TransactionParser):
         )
 
     def extract(self, msg: EmailMessage) -> Optional[Transaction]:
-        body = msg.get_body(preferencelist=("html", "plain"))
-        if body is None:
-            raise TransactionParsingFailed("No body of message found")
-        content = body.get_content()
-        if content is None:
-            raise TransactionParsingFailed("No content of message found")
+        content = self.get_content(msg)
         matches = self.EXTRACT_RE.search(content)
         if matches is None:
             raise TransactionParsingFailed("No matches for extraction RE")
@@ -145,12 +147,7 @@ class BMOParser(TransactionParser):
         return msg["From"] == "bmoalerts@bmo.com"
 
     def extract(self, msg: EmailMessage) -> Optional[Transaction]:
-        body = msg.get_body(preferencelist=("html", "plain"))
-        if body is None:
-            raise TransactionParsingFailed("No body of message found")
-        content = body.get_content()
-        if content is None:
-            raise TransactionParsingFailed("No content of message found")
+        content = self.get_content(msg)
         soup = BeautifulSoup(content, "html.parser")
         matches = self.EXTRACT_RE.search(soup.get_text())
         if matches is None:
@@ -162,7 +159,7 @@ class BMOParser(TransactionParser):
         date = parse_email_time(msg["Date"]).date()
         account_ref = int(matches[3])
         if account_ref not in self._account_map:
-            info("Account %s not in account map", account_ref)
+            warning("Account %s not in account map, skipping transaction", account_ref)
             return None
         account_id = self._account_map[account_ref]
         return Transaction(
@@ -188,12 +185,7 @@ class CIBCParser(TransactionParser):
         return msg["From"] == "CIBC Banking <mailbox.noreply@cibc.com>"
 
     def extract_payment(self, msg: EmailMessage):
-        body = msg.get_body(preferencelist=("html", "plain"))
-        if body is None:
-            raise TransactionParsingFailed("No body of message found")
-        content = body.get_content()
-        if content is None:
-            raise TransactionParsingFailed("No content of message found")
+        content = self.get_content(msg)
 
         matches = self.PAYMENT_EXTRACT_RE.search(content)
         if matches is None:
@@ -204,7 +196,7 @@ class CIBCParser(TransactionParser):
         date = parse_email_time(msg["Date"]).date()
 
         if account_ref not in self._account_map:
-            info("Account %s not in account map", account_ref)
+            warning("Account %s not in account map, skipping transaction", account_ref)
             return None
         account_id = self._account_map[account_ref]
 
@@ -222,3 +214,55 @@ class CIBCParser(TransactionParser):
             case "New payment to your credit card":
                 return self.extract_payment(msg)
         return None
+
+
+class ScotiaBankParser(TransactionParser):
+    PAYMENT_EXTRACT_RE = re.compile(
+        r"There was an authorization (?P<card>without the credit card present )?for \$(?P<amount>[0-9]+\.[0-9]{2}) at (?P<payee>.+) on account (?P<account>[0-9*]+) at\s+(?P<time>[0-9]{1,2}:[0-9]{2} (am|pm))",  # noqa: E501
+        re.MULTILINE,
+    )
+
+    def __init__(self, account_map: dict[str, UUID]):
+        self._account_map = account_map
+
+    def match(self, msg: Message) -> bool:
+        return msg["From"] == "Scotia InfoAlerts <infoalerts@scotiabank.com>"
+
+    def extract_transaction(self, msg: EmailMessage) -> Optional[Transaction]:
+        content = self.get_content(msg)
+
+        matches = self.PAYMENT_EXTRACT_RE.search(content)
+        if matches is None:
+            raise TransactionParsingFailed("no matches for extraction RE")
+
+        amount = Decimal(matches["amount"].replace(",", ""))
+        date = parse_email_time(msg["Date"]).date()
+
+        if matches["account"] not in self._account_map:
+            warning(
+                "Account %s not in account map, skipping transaction",
+                matches["account"],
+            )
+            return None
+        account_id = self._account_map[matches["account"]]
+
+        return Transaction(
+            account=account_id,
+            date=date,
+            amount=amount,
+            payee=matches["payee"],
+            notes="without card (via email)"
+            if matches.group("card") is not None
+            else "with card (via email)",
+            imported_id=msg["Message-ID"],
+        )
+
+    def extract(self, msg: EmailMessage) -> Optional[Transaction]:
+        match msg["Subject"]:
+            case (
+                "Authorization on your credit account"
+                | "Authorization without credit card present"
+            ):
+                return self.extract_transaction(msg)
+        info("Subject `%s` didn't match any extractors", msg["Subject"])
+        return None