Prompt file imported from nevedomski/SAS_to_PY (.github/prompts/07-normalization-utils.prompt.md). Copyright stays with the author.

Scaffold normalization helpers to ensure deterministic DF comparisons

Dependencies:

• Requires schema contracts from 02-schema-contracts.prompt.md.
• Feeds into 08-test-chunk-compare.prompt.md and 10-test-e2e.prompt.md.

Instructions:
Implement normalization utilities to ensure deterministic test comparisons. Include:

1. Data type normalization (e.g., date formats, numeric precision).
2. Column ordering enforcement.
3. Missing value handling rules.

Create tests/test_utils/normalize.py with:

from __future__ import annotations
from typing import Iterable, Mapping
import pandas as pd
import numpy as np


def normalize_df(
    df: pd.DataFrame,
    *,
    sort_keys: Iterable[str] | None = None,
    column_order: Iterable[str] | None = None,
    datetime_cols: Iterable[str] = (),
    strip_string_cols: Iterable[str] | None = None,
) -> pd.DataFrame:
    """
    Apply canonical ordering, dtype normalization, NA normalization, and optional string trimming.
    Returns a new DataFrame.
    """
    ...

• Ensure immutability (don't mutate the argument in place).
• Avoid expensive copies where they are not needed.
• Ensure the helper is unit-test friendly and well-documented.