From 8b4a40dcddfef1d4de76fba161039826a144c761 Mon Sep 17 00:00:00 2001
From: Sergey Volnov <volnov.serg@gmail.com>
Date: Mon, 26 Jun 2023 11:12:44 +0100
Subject: [PATCH] Update Morgan Stanley parser (#399)

- Handle GOOG stock split correctly
- Treat "Staged" releases similarly to "Completed"
---
 cgt_calc/parsers/mssb.py                      | 60 ++++++++++++++++---
 tests/test_data/mssb/Releases Report.csv      |  3 +-
 tests/test_data/mssb/Withdrawals Report.csv   |  5 +-
 .../test_run_with_example_files_output.txt    | 12 ++--
 4 files changed, 65 insertions(+), 15 deletions(-)

diff --git a/cgt_calc/parsers/mssb.py b/cgt_calc/parsers/mssb.py
index 173c6d5b..eb10a150 100644
--- a/cgt_calc/parsers/mssb.py
+++ b/cgt_calc/parsers/mssb.py
@@ -6,7 +6,8 @@
 from __future__ import annotations
 
 import csv
-from datetime import datetime
+from dataclasses import dataclass
+import datetime
 from decimal import Decimal
 from pathlib import Path
 from typing import Final
@@ -48,6 +49,20 @@
 }
 
 
+@dataclass
+class StockSplit:
+    """Info about stock split."""
+
+    symbol: str
+    date: datetime.date
+    factor: int
+
+
+STOCK_SPLIT_INFO = [
+    StockSplit(symbol="GOOG", date=datetime.datetime(2022, 6, 15).date(), factor=20),
+]
+
+
 def _hacky_parse_decimal(decimal: str) -> Decimal:
     return Decimal(decimal.replace(",", ""))
 
@@ -60,7 +75,7 @@ def _init_from_release_report(row_raw: list[str], filename: str) -> BrokerTransa
     if row["Type"] != "Release":
         raise ParsingError(filename, f"Unknown type: {row_raw[3]}")
 
-    if row["Status"] != "Complete":
+    if row["Status"] != "Complete" and row["Status"] != "Staged":
         raise ParsingError(filename, f"Unknown status: {row_raw[5]}")
 
     if row["Price"][0] != "$":
@@ -79,7 +94,7 @@ def _init_from_release_report(row_raw: list[str], filename: str) -> BrokerTransa
     symbol = TICKER_RENAMES.get(symbol, symbol)
 
     return BrokerTransaction(
-        date=datetime.strptime(row["Vest Date"], "%d-%b-%Y").date(),
+        date=datetime.datetime.strptime(row["Vest Date"], "%d-%b-%Y").date(),
         action=ActionType.STOCK_ACTIVITY,
         symbol=symbol,
         description=row["Plan"],
@@ -92,9 +107,38 @@ def _init_from_release_report(row_raw: list[str], filename: str) -> BrokerTransa
     )
 
 
+# Morgan Stanley decided to put a notice in the end of the withdrawal report that looks
+# like that:
+# "Please note that any Alphabet share sales, transfers, or deposits that occurred on
+# or prior to the July 15, 2022 stock split are reflected in pre-split. Any sales,
+# transfers, or deposits that occurred after July 15, 2022 are in post-split values.
+# For GSU vests, your activity is displayed in post-split values."
+# It makes sense, but it totally breaks the CSV parsing
+def _is_notice(row: list[str]) -> bool:
+    return row[0][:11] == "Please note"
+
+
+def _handle_stock_split(transaction: BrokerTransaction) -> BrokerTransaction:
+    for split in STOCK_SPLIT_INFO:
+        if (
+            transaction.symbol == split.symbol
+            and transaction.action == ActionType.SELL
+            and transaction.date < split.date
+        ):
+            if transaction.quantity:
+                transaction.quantity *= split.factor
+            if transaction.price:
+                transaction.price /= split.factor
+
+    return transaction
+
+
 def _init_from_withdrawal_report(
     row_raw: list[str], filename: str
-) -> BrokerTransaction:
+) -> BrokerTransaction | None:
+    if _is_notice(row_raw):
+        return None
+
     if len(COLUMNS_WITHDRAWAL) != len(row_raw):
         raise UnexpectedColumnCountError(row_raw, len(COLUMNS_WITHDRAWAL), filename)
     row = {col: row_raw[i] for i, col in enumerate(COLUMNS_WITHDRAWAL)}
@@ -122,8 +166,8 @@ def _init_from_withdrawal_report(
     else:
         action = ActionType.SELL
 
-    return BrokerTransaction(
-        date=datetime.strptime(row["Date"], "%d-%b-%Y").date(),
+    transaction = BrokerTransaction(
+        date=datetime.datetime.strptime(row["Date"], "%d-%b-%Y").date(),
         action=action,
         symbol=KNOWN_SYMBOL_DICT[row["Plan"]],
         description=row["Plan"],
@@ -135,6 +179,8 @@ def _init_from_withdrawal_report(
         broker="Morgan Stanley",
     )
 
+    return _handle_stock_split(transaction)
+
 
 def _validate_header(
     header: list[str], golden_header: list[str], filename: str
@@ -172,4 +218,4 @@ def read_mssb_transactions(transactions_folder: str) -> list[BrokerTransaction]:
                     _init_from_release_report(row, str(file)) for row in lines
                 ]
 
-    return transactions
+    return [transaction for transaction in transactions if transaction]
diff --git a/tests/test_data/mssb/Releases Report.csv b/tests/test_data/mssb/Releases Report.csv
index 50f74e9b..232b9bd6 100644
--- a/tests/test_data/mssb/Releases Report.csv	
+++ b/tests/test_data/mssb/Releases Report.csv	
@@ -1,2 +1,3 @@
 Vest Date,Order Number,Plan,Type,Status,Price,Quantity,Net Cash Proceeds,Net Share Proceeds,Tax Payment Method
-25-Mar-2021,ORDER_NUMBER_HIDDEN,GSU Class C,Release,Complete,"$2,045.06",20.000,$0.00,10.60,Fractional Shares
+25-Mar-2021,ORDER_NUMBER_HIDDEN,GSU Class C,Release,Complete,$102.25,400.000,$0.00,212.0,Fractional Shares
+25-Mar-2023,ORDER_NUMBER_HIDDEN,GSU Class C,Release,Complete,$121.64,40.000,$0.00,21.201,Fractional Shares
diff --git a/tests/test_data/mssb/Withdrawals Report.csv b/tests/test_data/mssb/Withdrawals Report.csv
index 1b806d5a..c6a3c3f5 100644
--- a/tests/test_data/mssb/Withdrawals Report.csv	
+++ b/tests/test_data/mssb/Withdrawals Report.csv	
@@ -1,3 +1,6 @@
 Date,Order Number,Plan,Type,Order Status,Price,Quantity,Net Amount,Net Share Proceeds,Tax Payment Method
 01-Apr-2021,ORDER_NUMBER_HIDDEN,GSU Class C,Sale,Complete,"$2,110.00",-2,"$4,219.95",0,N/A
-02-Apr-2023,ORDER_NUMBER_HIDDEN,Cash,Sale,Complete,$1.00,"-4,218.95","$4,218.95",0,N/A
+02-Apr-2021,ORDER_NUMBER_HIDDEN,Cash,Sale,Complete,$1.00,"-4,218.95","$4,218.95",0,N/A
+09-Feb-2023,ORDER_NUMBER_HIDDEN,Cash,Sale,Complete,$1.00,"-3,170.93","$3,170.93",0,N/A
+07-Feb-2023,ORDER_NUMBER_HIDDEN,GSU Class C,Sale,Complete,$105.70,-30.00,"$3,170.93",0,N/A
+Please note that any Alphabet share sales, transfers, or deposits that occurred on or prior to the July 15, 2022 stock split are reflected in pre-split. Any sales, transfers, or deposits that occurred after July 15, 2022 are in post-split values. For GSU vests, your activity is displayed in post-split values.
diff --git a/tests/test_data/test_run_with_example_files_output.txt b/tests/test_data/test_run_with_example_files_output.txt
index 5e2ad988..dab7befd 100644
--- a/tests/test_data/test_run_with_example_files_output.txt
+++ b/tests/test_data/test_run_with_example_files_output.txt
@@ -6,7 +6,7 @@ First pass completed
 Final portfolio:
   GE: 1.00
   NVDA: 1.00
-  GOOG: 8.60
+  GOOG: 163.20
   FOO: 30.50
 Final balance:
   Charles Schwab: 9477.90 (USD)
@@ -24,15 +24,15 @@ Portfolio at the end of 2020/2021 tax year:
   GE: 1.00, £8.07
   NVDA: 1.00, £401.01
   FOO: 30.50, £543.97
-  GOOG: 8.60, £12516.92
+  GOOG: 172.00, £12516.55
 For tax year 2020/2021:
 Number of disposals: 4
 Disposal proceeds: £42769.56
-Allowable costs: £17656.17
-Capital gain: £25113.39
+Allowable costs: £17656.08
+Capital gain: £25113.48
 Capital loss: £0.00
-Total capital gain: £25113.39
-Taxable capital gain: £12813.39
+Total capital gain: £25113.48
+Taxable capital gain: £12813.48
 
 Generate calculations report
 All done!