diff --git a/broker_sync/providers/parsers/schwab.py b/broker_sync/providers/parsers/schwab.py index 762a613..5a34f1b 100644 --- a/broker_sync/providers/parsers/schwab.py +++ b/broker_sync/providers/parsers/schwab.py @@ -7,9 +7,16 @@ cells holding date / direction / quantity / ticker / price. It does NOT email vest-release / Release Confirmation messages to the employee address for this account (verified against 4 years of inbox -history, 2022-2026 — see infra/docs in code-fqgr). Vest data must come -from the META payslip via payslip-ingest, not from email. The whole -vest-release parser that used to live here was dead code. +history, 2022-2026). The vest itself is invisible to IMAP. + +Same-day-sell synthesis: Meta RSUs vest and are sold the same day at +the same FMV (verified across 14 historical vests). When a SELL email +is parsed AND its trade date is on or after `VEST_INFER_FROM_DATE`, +we ALSO emit a paired BUY representing the underlying vest event — +same date, same quantity, same price. The date boundary stops this +back-filling historical vests that already have csv-sourced BUY rows +in Wealthfolio (which would duplicate at chart-level despite distinct +external_ids). On any parse failure we return an empty list — an unparseable email shouldn't crash the IMAP batch. @@ -17,6 +24,8 @@ shouldn't crash the IMAP batch. from __future__ import annotations import logging +import os +from datetime import date, datetime from decimal import Decimal, InvalidOperation from bs4 import BeautifulSoup @@ -29,9 +38,34 @@ log = logging.getLogger(__name__) _ACCOUNT_ID = "schwab-workplace" _DEFAULT_CURRENCY = "USD" +# Inferred-BUY synthesis boundary. SELL emails on or after this date +# emit a paired BUY for the underlying vest; earlier ones do not (they +# already have csv-sourced BUYs in Wealthfolio from the one-shot +# historical backfill, last vest 2026-02-18). Override at runtime with +# the env var if a different cutover is needed. ISO-8601 yyyy-mm-dd. +_DEFAULT_VEST_INFER_FROM = "2026-04-01" + + +def _vest_infer_from() -> date: + raw = os.environ.get("SCHWAB_VEST_INFER_FROM_DATE", _DEFAULT_VEST_INFER_FROM).strip() + try: + return datetime.strptime(raw, "%Y-%m-%d").date() + except ValueError: + log.warning( + "SCHWAB_VEST_INFER_FROM_DATE=%r is not yyyy-mm-dd; using default %s", + raw, _DEFAULT_VEST_INFER_FROM, + ) + return datetime.strptime(_DEFAULT_VEST_INFER_FROM, "%Y-%m-%d").date() + def parse_schwab_email(raw_html: str) -> list[Activity]: - """Return a one-element list of Activity on success, empty on failure.""" + """Return Activities for a Schwab trade-executed email. + + Returns: empty list on parse failure; one Activity for a BUY-direction + email (rare — the workplace account is essentially sell-only); for a + SELL email, returns [SELL] plus an inferred paired BUY (=vest event) + when the trade date is on or after the synthesis-boundary date. + """ try: soup = BeautifulSoup(raw_html, "html.parser") cells = [ @@ -52,22 +86,40 @@ def parse_schwab_email(raw_html: str) -> list[Activity]: for sign in ("$", "£", "€", "USD", "GBP", "EUR"): price_clean = price_clean.replace(sign, "") unit_price = Decimal(price_clean.replace(",", "").strip()) + ticker_clean = ticker.strip() - external_id = (f"schwab:{trade_date.date().isoformat()}:{ticker}:" + external_id = (f"schwab:{trade_date.date().isoformat()}:{ticker_clean}:" f"{direction.value}:{quantity}") - return [ - Activity( - external_id=external_id, - account_id=_ACCOUNT_ID, - account_type=AccountType.GIA, - date=trade_date, - activity_type=direction, - symbol=ticker.strip(), - quantity=quantity, - unit_price=unit_price, - currency=_DEFAULT_CURRENCY, - notes=f"schwab-email:{direction_txt}", - ) - ] + primary = Activity( + external_id=external_id, + account_id=_ACCOUNT_ID, + account_type=AccountType.GIA, + date=trade_date, + activity_type=direction, + symbol=ticker_clean, + quantity=quantity, + unit_price=unit_price, + currency=_DEFAULT_CURRENCY, + notes=f"schwab-email:{direction_txt}", + ) + + if direction is not ActivityType.SELL or trade_date.date() < _vest_infer_from(): + return [primary] + + inferred_buy = Activity( + external_id=(f"schwab:vest:{trade_date.date().isoformat()}:" + f"{ticker_clean}:BUY:{quantity}"), + account_id=_ACCOUNT_ID, + account_type=AccountType.GIA, + date=trade_date, + activity_type=ActivityType.BUY, + symbol=ticker_clean, + quantity=quantity, + unit_price=unit_price, + currency=_DEFAULT_CURRENCY, + notes=(f"schwab-vest-inferred-from-same-day-sell | " + f"paired_sell_external_id={external_id}"), + ) + return [inferred_buy, primary] except (ValueError, InvalidOperation, IndexError, AttributeError): return [] diff --git a/tests/providers/parsers/test_schwab.py b/tests/providers/parsers/test_schwab.py index 8e3c736..2cc0213 100644 --- a/tests/providers/parsers/test_schwab.py +++ b/tests/providers/parsers/test_schwab.py @@ -80,5 +80,66 @@ def test_external_id_is_stable_across_reruns() -> None: def test_price_with_commas_parses() -> None: html = _SELL.replace("$612.34", "$1,612.34") - a = parse_schwab_email(html)[0] - assert a.unit_price == Decimal("1612.34") + # The first activity is the inferred BUY (date 2025-01-23 ≥ 2026-04-01? no → + # only one activity for this old-dated email), so index 0 is the SELL. + acts = parse_schwab_email(html) + sell = next(a for a in acts if a.activity_type is ActivityType.SELL) + assert sell.unit_price == Decimal("1612.34") + + +# --- Inferred vest BUY --------------------------------------------------- + + +def _recent_sell(date_iso: str = "2026-05-19", qty: str = "55", price: str = "609.35") -> str: + return f""" + + + + + + +
{date_iso}
Sold
{qty}
META
${price}
+""" + + +def test_recent_sell_emits_paired_buy() -> None: + """SELL dated on/after the synthesis boundary triggers a paired BUY.""" + acts = parse_schwab_email(_recent_sell()) + assert len(acts) == 2 + + buy = next(a for a in acts if a.activity_type is ActivityType.BUY) + sell = next(a for a in acts if a.activity_type is ActivityType.SELL) + + assert buy.quantity == sell.quantity == Decimal("55") + assert buy.unit_price == sell.unit_price == Decimal("609.35") + assert buy.date == sell.date + assert buy.symbol == sell.symbol == "META" + assert "schwab-vest-inferred-from-same-day-sell" in (buy.notes or "") + assert buy.external_id == "schwab:vest:2026-05-19:META:BUY:55" + assert sell.external_id == "schwab:2026-05-19:META:SELL:55" + + +def test_old_sell_emits_only_sell() -> None: + """SELL dated before 2026-04-01 (default boundary) skips the paired BUY — + those vests already have csv-sourced BUY rows in Wealthfolio.""" + acts = parse_schwab_email(_recent_sell(date_iso="2025-08-19")) + assert len(acts) == 1 + assert acts[0].activity_type is ActivityType.SELL + + +def test_boundary_env_var_overrides(monkeypatch: object) -> None: + """The synthesis boundary is configurable via env var.""" + import os + os.environ["SCHWAB_VEST_INFER_FROM_DATE"] = "2025-01-01" + try: + acts = parse_schwab_email(_recent_sell(date_iso="2025-08-19")) + assert len(acts) == 2 # now in scope + finally: + del os.environ["SCHWAB_VEST_INFER_FROM_DATE"] + + +def test_buy_email_does_not_emit_inferred_buy() -> None: + """BUY-direction emails (rare for workplace account) don't get paired.""" + acts = parse_schwab_email(_BUY.replace("2024-11-15", "2026-05-15")) + assert len(acts) == 1 + assert acts[0].activity_type is ActivityType.BUY