read.markets/app/services/csv_import.py
Giorgio Gilestro bc55ab7d26 csv-parser: keep LLM-mapped tickers; don't pass them through T212 mapping
The route's resolve-slice loop is T212-specific — it looks tickers up
against the InstrumentMap, which only has T212's universe. For the LLM
path the ticker is already Yahoo-ready (e.g. VOD.L, ASML.AS), so
sending it through resolve_slice produced spurious "could not be
resolved" warnings and dropped the positions.

Fix: ParsedPie gains a ``tickers_resolved`` flag (default False for
T212 backward-compat); _apply_mapping in the LLM path sets it True
and also extracts currency from the LLM-mapped currency_col into a
new ``ParsedPosition.currency`` field. The route branches on the flag:
LLM-path positions are kept verbatim with a best-effort InstrumentMap
lookup for nicer name/currency overrides, never dropped.

Integration test tightened to assert all 5 IBKR fixture positions
round-trip with the right currencies (USD / GBP / EUR).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-27 12:48:27 +02:00

227 lines
8.3 KiB
Python

"""Defensive parser for Trading 212 pie-export CSVs.
The parser is pure: no DB, no HTTP, no I/O. Returns a ParsedPie that
`/api/portfolio/parse` ships to the browser; in Phase G the browser
keeps the pie in localStorage and the server keeps only the anonymous
ticker_universe.
"""
from __future__ import annotations
import csv
import io
from dataclasses import dataclass
class CSVImportError(ValueError):
"""Raised when the CSV is unparseable or missing required columns."""
# Header name -> normalised key used in the parsed dict. Lowercase, ignore
# leading/trailing whitespace, treat case-insensitively. Extra columns are
# silently ignored.
_HEADER_MAP = {
"slice": "slice",
"name": "name",
"invested value": "invested_value",
"value": "current_value",
"result": "result",
"owned quantity": "quantity",
"dividends gained": "dividends_gained",
"dividends cash": "dividends_cash",
"dividends reinvested": "dividends_reinvested",
}
# These must be present for the import to be meaningful at all.
_REQUIRED_FIELDS = ("slice", "quantity")
@dataclass(frozen=True)
class ParsedPosition:
slice: str # T212 shortcode (e.g. "SGLN") or a
# Yahoo-ready ticker (e.g. "VOD.L")
# when produced by the LLM path —
# see ParsedPie.tickers_resolved.
name: str
invested_value: float | None
current_value: float | None
result: float | None # P/L in pie currency
quantity: float
dividends_gained: float | None = None
dividends_cash: float | None = None
dividends_reinvested: float | None = None
currency: str | None = None # Populated by the LLM path from the
# mapped currency_col; the T212 path
# leaves it None and gets currency
# from the InstrumentMap row.
@property
def average_price(self) -> float | None:
if self.invested_value is None or not self.quantity:
return None
return self.invested_value / self.quantity
@property
def current_price(self) -> float | None:
if self.current_value is None or not self.quantity:
return None
return self.current_value / self.quantity
@dataclass(frozen=True)
class ParsedPie:
name: str | None # from the Total row's Name column
positions: tuple[ParsedPosition, ...]
invested: float | None # totals from the Total row
value: float | None
result: float | None
tickers_resolved: bool = False # True when ``slice`` on each position
# is already a Yahoo-ready ticker
# (LLM path). False (default) means
# tickers must still be resolved via
# the T212 InstrumentMap.
def _normalise_header(h: str) -> str:
return h.strip().lower()
def _parse_num(raw: str | None) -> float | None:
"""Empty / 'N/A' / '-' / '' → None. Otherwise float."""
if raw is None:
return None
s = raw.strip()
if not s or s in {"-", "", "N/A", "n/a", "NA"}:
return None
# T212 occasionally exports with thousand-comma. Strip safely.
s = s.replace(",", "")
try:
return float(s)
except ValueError:
return None
def parse_t212_csv(content: str | bytes) -> ParsedPie:
"""Parse a T212 pie-export CSV.
Args:
content: bytes or str containing the CSV (raw export file contents).
Returns:
ParsedPie with positions list and aggregate totals.
Raises:
CSVImportError: if the file is empty, missing required headers,
or contains no usable rows.
"""
if isinstance(content, bytes):
try:
content = content.decode("utf-8-sig") # handle Excel BOM
except UnicodeDecodeError:
content = content.decode("latin-1")
reader = csv.reader(io.StringIO(content))
try:
header_row = next(reader)
except StopIteration:
raise CSVImportError("Empty CSV file")
# Map column index -> normalised field name. Unknown headers are ignored.
field_by_index: dict[int, str] = {}
for i, h in enumerate(header_row):
key = _HEADER_MAP.get(_normalise_header(h))
if key:
field_by_index[i] = key
missing = [f for f in _REQUIRED_FIELDS if f not in field_by_index.values()]
if missing:
raise CSVImportError(
f"CSV missing required column(s): {', '.join(missing)}. "
f"Found headers: {header_row}"
)
positions: list[ParsedPosition] = []
total: ParsedPosition | None = None
pie_name: str | None = None
zero_qty_slices = 0 # real slice rows skipped for missing/zero quantity
for row_num, row in enumerate(reader, start=2):
if not row or not any(cell.strip() for cell in row):
continue # skip blank lines
record: dict[str, object] = {}
for idx, field in field_by_index.items():
raw = row[idx] if idx < len(row) else ""
if field in {"slice", "name"}:
record[field] = raw.strip()
else:
record[field] = _parse_num(raw)
slice_code = record.get("slice") or ""
if not slice_code:
continue # malformed; skip silently rather than abort
# The 'Total' row uses slice='Total' and quantity='-' — capture it
# for aggregate totals but don't list it as a position.
if slice_code.lower() == "total":
pie_name = (record.get("name") or "").strip() or None
total = ParsedPosition(
slice=slice_code,
name=pie_name or "Total",
invested_value=record.get("invested_value"),
current_value=record.get("current_value"),
result=record.get("result"),
quantity=0.0,
dividends_gained=record.get("dividends_gained"),
dividends_cash=record.get("dividends_cash"),
dividends_reinvested=record.get("dividends_reinvested"),
)
continue
qty = record.get("quantity")
if qty is None or qty == 0:
# Position row with no usable quantity — skip rather than fail.
# Counted so an all-zero (unfunded) pie yields a precise error.
zero_qty_slices += 1
continue
positions.append(ParsedPosition(
slice=slice_code,
name=(record.get("name") or "").strip(),
invested_value=record.get("invested_value"),
current_value=record.get("current_value"),
result=record.get("result"),
quantity=qty,
dividends_gained=record.get("dividends_gained"),
dividends_cash=record.get("dividends_cash"),
dividends_reinvested=record.get("dividends_reinvested"),
))
if not positions:
# Distinguish an unfunded pie (slices present, all 0 quantity)
# from a genuinely unreadable file — the two need very different
# user action, and the generic message misleads people into
# debugging the file format.
if zero_qty_slices:
raise CSVImportError(
f"This pie holds no shares — all {zero_qty_slices} "
f"slice(s) have an Owned quantity of 0. Export the pie from "
f"Trading 212 after it has been funded."
)
raise CSVImportError(
"CSV contained no parseable position rows. "
"Expected at least one row with a Slice code and quantity."
)
return ParsedPie(
name=pie_name,
positions=tuple(positions),
invested=total.invested_value if total else None,
value=total.current_value if total else None,
result=total.result if total else None,
)
# persist_pie removed in Phase G — the parsed pie is returned to the
# browser by /api/portfolio/parse and lives in localStorage. The server
# now keeps only the anonymous ticker_universe (see
# app/services/ticker_universe.py).