# Purchase Trade Import Prices Backlog

Backlog of potential issues found during a targeted review of the
`purchase_trade.import_prices` wizard.

## Findings

### Incomplete auto-created price indexes

- Area: `modules/purchase_trade/pricing.py`, `_price_index_values`
- Risk: medium
- Current behavior: when `create_missing_price_index` is enabled, the wizard
  creates a `price.price` even if default references cannot be found.
- Details: default references are searched by exact names: `Market price`,
  `USD`, `Argus EU`, and `Mt`.
- Possible impact: imported prices may be attached to indexes without unit,
  currency, type, or calendar, which can later break or distort price
  conversions.
- Suggested action: fail clearly when required defaults are missing, or expose
 /import a controlled mapping instead of relying on silent best-effort lookup.

### Duplicate price values for the same index and date

- Area: `modules/purchase_trade/pricing.py`, `_import_rows`; `modules/price/price_value.py`
- Risk: medium
- Current behavior: the wizard searches an existing `price.price_value` with
  `limit=1` for a given `price` and `price_date`.
- Details: no uniqueness constraint was found on `price.price_value` for
  `(price, price_date)`.
- Possible impact: if duplicates already exist, only one record is updated, and
  later price reads may use a non-deterministic record.
- Suggested action: add a duplicate detection step in the import wizard. Longer
  term, consider a unique constraint or a cleanup/migration if the database
  already contains duplicates.

### Strict forward month term formats

- Area: `modules/purchase_trade/pricing.py`, `_as_month_term`
- Risk: low to medium
- Current behavior: forward month columns accept Excel date values or text in
  `YYYY-MM` format.
- Details: common business labels such as `MAR26`, `Mar-26`, or `March 2026`
  are rejected.
- Possible impact: valid user files may fail with `Invalid month term`.
- Suggested action: confirm the real Excel template used by operators. If
  needed, support month abbreviations while keeping unambiguous validation.

### Fragile text number parsing

- Area: `modules/purchase_trade/pricing.py`, `_as_float`
- Risk: low to medium
- Current behavior: blank values become `None`; comma decimals such as `75,25`
  are accepted.
- Details: formatted text numbers such as `1 234,56`, `1,234.56`, or values
  with non-breaking spaces are not handled safely.
- Possible impact: imports may fail on files where prices are pasted as
  formatted text instead of stored as numeric Excel cells.
- Suggested action: normalize spaces and common thousand separators, or produce
  a clearer validation error with row/index context.

### Excel 1904 date system is not handled

- Area: `modules/purchase_trade/pricing.py`, `_as_date`
- Risk: low
- Current behavior: numeric Excel dates are interpreted with the 1899-12-30
  base.
- Details: Excel workbooks using the 1904 date system would be shifted by about
  four years.
- Possible impact: imported `price_date` or forward month terms could be wrong
  for uncommon workbook settings.
- Suggested action: inspect workbook properties for the date system before
  converting numeric dates.

## Recently Fixed

### Forward prices only populated `price_value`

- Area: `modules/purchase_trade/pricing.py`, `_read_forward_price_rows`
- Status: fixed locally
- Change: forward imports now copy each forward cell value into `price_value`,
  `open_price`, `low_price`, and `high_price`.
- Regression coverage: `test_import_forward_prices_copies_value_to_ohl_fields`
  in `modules/purchase_trade/tests/test_module.py`.