# pricing_calculator

Python model generated by Power Migrate 0.1.0a1 from `pricing_calculator.xlsx`
(SHA-256 `d2a7b89598fa1cf4...`), in no-AI mode: every name and
every line comes from deterministic rules, so the code can be regenerated
and audited without a model call.

## Layout

| File | Purpose |
| --- | --- |
| `model.py` | `compute(book)` runs every logic block in dependency order; `load_inputs()` loads the workbook's own inputs |
| `sheets/*.py` | One module per worksheet; one function per logic block (a formula and all the cells it was copied to) |
| `xl.py` | Excel value semantics (blank = 0, text/number coercion, error values, VLOOKUP and friends). No dependencies |
| `inputs.json` | Every constant cell a formula reads, plus defined names. Edit it, or set cells on the Book, to run new scenarios |
| `expected.json` | What Excel last calculated for every formula cell: the reconciliation reference |
| `baseline.json` | Cells that reconciled at generation time; `test_model.py` checks they still do |
| `udf.py` | User-defined functions from the VBA project, machine-translated where simple, otherwise stubs with the VBA in the docstring (only when the workbook has macros) |

## Run it

```python
from pricing_calculator import compute, load_inputs

book = compute(load_inputs())
print(book["Sheet1"]["B7"].v)          # a single cell's value
book["Sheet1"]["B3"] = 42             # change an input...
compute(book)                          # ...and recompute
```

Dates are Excel serial numbers. Error cells hold `XlError` values (`#N/A`,
`#DIV/0!`...). Anything the generator could not convert is a call to
`xl.UNSUPPORTED(reason)` whose result propagates to every dependent cell and
is listed in the reconciliation report; 0 of 194 blocks contain one.

This package shows tested equivalence with the workbook's stored values for
the cells listed in `baseline.json`. It is not a guarantee for other inputs.