Refactor retailer normalization outputs

enrich_costco.py

@@ -8,7 +8,10 @@ import click
 
 from enrich_giant import (
     OUTPUT_FIELDS,
+    derive_normalized_quantity,
+    derive_price_fields,
     format_decimal,
+    normalization_identity,
     normalize_number,
     normalize_unit,
     normalize_whitespace,
@@ -177,12 +180,42 @@ def parse_costco_item(order_id, order_date, raw_path, line_no, item):
     price_per_each, price_per_lb, price_per_oz = derive_costco_prices(
         item, measure_type, size_value, size_unit, pack_qty
     )
+    normalized_row_id = f"{RETAILER}:{order_id}:{line_no}"
+    normalized_quantity, normalized_quantity_unit = derive_normalized_quantity(
+        size_value,
+        size_unit,
+        pack_qty,
+        measure_type,
+    )
+    identity_key, normalization_basis = normalization_identity(
+        {
+            "retailer": RETAILER,
+            "normalized_row_id": normalized_row_id,
+            "upc": "",
+            "retailer_item_id": str(item.get("itemNumber", "")),
+            "item_name_norm": item_name_norm,
+            "size_value": size_value,
+            "size_unit": size_unit,
+            "pack_qty": pack_qty,
+        }
+    )
+    price_fields = derive_price_fields(
+        price_per_each,
+        price_per_lb,
+        price_per_oz,
+        str(item.get("amount", "")),
+        str(item.get("unit", "")),
+        pack_qty,
+    )
 
     return {
         "retailer": RETAILER,
         "order_id": str(order_id),
         "line_no": str(line_no),
-        "observed_item_key": f"{RETAILER}:{order_id}:{line_no}",
+        "normalized_row_id": normalized_row_id,
+        "normalized_item_id": f"cnorm:{identity_key}",
+        "normalization_basis": normalization_basis,
+        "observed_item_key": normalized_row_id,
         "order_date": normalize_whitespace(order_date),
         "retailer_item_id": str(item.get("itemNumber", "")),
         "pod_id": "",
@@ -210,13 +243,14 @@ def parse_costco_item(order_id, order_date, raw_path, line_no, item):
         "size_unit": size_unit,
         "pack_qty": pack_qty,
         "measure_type": measure_type,
+        "normalized_quantity": normalized_quantity,
+        "normalized_quantity_unit": normalized_quantity_unit,
         "is_store_brand": "true" if brand_guess else "false",
+        "is_item": "false" if is_discount_line else "true",
         "is_fee": "false",
         "is_discount_line": "true" if is_discount_line else "false",
         "is_coupon_line": is_coupon_line,
-        "price_per_each": price_per_each,
-        "price_per_lb": price_per_lb,
-        "price_per_oz": price_per_oz,
+        **price_fields,
         "parse_version": PARSER_VERSION,
         "parse_notes": "",
     }
@@ -321,6 +355,7 @@ def write_csv(path, rows):
     help="CSV path for enriched Costco item rows.",
 )
 def main(input_dir, output_csv):
+    click.echo("legacy entrypoint: prefer normalize_costco_web.py for data-model outputs")
     rows = build_items_enriched(Path(input_dir))
     write_csv(Path(output_csv), rows)
     click.echo(f"wrote {len(rows)} rows to {output_csv}")