Add Costco acquisition and enrich flow

2026-03-16 09:17:46 -04:00
parent 9497565978
commit da00288f10
4 changed files with 1090 additions and 0 deletions
--- a/validate_cross_retailer_flow.py
+++ b/validate_cross_retailer_flow.py
@@ -0,0 +1,154 @@
+import json
+from pathlib import Path
+
+import click
+
+import build_canonical_layer
+import build_observed_products
+from layer_helpers import stable_id, write_csv_rows
+
+
+PROOF_FIELDS = [
+    "proof_name",
+    "canonical_product_id",
+    "giant_observed_product_id",
+    "costco_observed_product_id",
+    "giant_example_item",
+    "costco_example_item",
+    "notes",
+]
+
+
+def read_rows(path):
+    import csv
+
+    with Path(path).open(newline="", encoding="utf-8") as handle:
+        return list(csv.DictReader(handle))
+
+
+def find_proof_pair(observed_rows):
+    giant = None
+    costco = None
+    for row in observed_rows:
+        if row["retailer"] == "giant" and row["representative_name_norm"] == "BANANA":
+            giant = row
+        if row["retailer"] == "costco" and row["representative_name_norm"] == "BANANA":
+            costco = row
+    return giant, costco
+
+
+def merge_proof_pair(canonical_rows, link_rows, giant_row, costco_row):
+    if not giant_row or not costco_row:
+        return canonical_rows, link_rows, []
+
+    proof_canonical_id = stable_id("gcan", "proof|banana")
+    link_rows = [
+        row
+        for row in link_rows
+        if row["observed_product_id"]
+        not in {giant_row["observed_product_id"], costco_row["observed_product_id"]}
+    ]
+    canonical_rows = [
+        row
+        for row in canonical_rows
+        if row["canonical_product_id"] != proof_canonical_id
+    ]
+    canonical_rows.append(
+        {
+            "canonical_product_id": proof_canonical_id,
+            "canonical_name": "BANANA",
+            "product_type": "banana",
+            "brand": "",
+            "variant": "",
+            "size_value": "",
+            "size_unit": "",
+            "pack_qty": "",
+            "measure_type": "weight",
+            "normalized_quantity": "",
+            "normalized_quantity_unit": "",
+            "notes": "manual proof merge for cross-retailer validation",
+            "created_at": "",
+            "updated_at": "",
+        }
+    )
+    for observed_row in [giant_row, costco_row]:
+        link_rows.append(
+            {
+                "observed_product_id": observed_row["observed_product_id"],
+                "canonical_product_id": proof_canonical_id,
+                "link_method": "manual_proof_merge",
+                "link_confidence": "medium",
+                "review_status": "",
+                "reviewed_by": "",
+                "reviewed_at": "",
+                "link_notes": "cross-retailer validation proof",
+            }
+        )
+
+    proof_rows = [
+        {
+            "proof_name": "banana",
+            "canonical_product_id": proof_canonical_id,
+            "giant_observed_product_id": giant_row["observed_product_id"],
+            "costco_observed_product_id": costco_row["observed_product_id"],
+            "giant_example_item": giant_row["example_item_name"],
+            "costco_example_item": costco_row["example_item_name"],
+            "notes": "BANANA proof pair built from Giant and Costco enriched rows",
+        }
+    ]
+    return canonical_rows, link_rows, proof_rows
+
+
+@click.command()
+@click.option(
+    "--giant-items-enriched-csv",
+    default="giant_output/items_enriched.csv",
+    show_default=True,
+)
+@click.option(
+    "--costco-items-enriched-csv",
+    default="costco_output/items_enriched.csv",
+    show_default=True,
+)
+@click.option(
+    "--outdir",
+    default="combined_output",
+    show_default=True,
+)
+def main(giant_items_enriched_csv, costco_items_enriched_csv, outdir):
+    outdir = Path(outdir)
+    rows = read_rows(giant_items_enriched_csv) + read_rows(costco_items_enriched_csv)
+    observed_rows = build_observed_products.build_observed_products(rows)
+    canonical_rows, link_rows = build_canonical_layer.build_canonical_layer(observed_rows)
+    giant_row, costco_row = find_proof_pair(observed_rows)
+    if not giant_row or not costco_row:
+        raise click.ClickException(
+            "could not find BANANA proof pair across Giant and Costco observed products"
+        )
+    canonical_rows, link_rows, proof_rows = merge_proof_pair(
+        canonical_rows, link_rows, giant_row, costco_row
+    )
+
+    write_csv_rows(
+        outdir / "products_observed.csv",
+        observed_rows,
+        build_observed_products.OUTPUT_FIELDS,
+    )
+    write_csv_rows(
+        outdir / "products_canonical.csv",
+        canonical_rows,
+        build_canonical_layer.CANONICAL_FIELDS,
+    )
+    write_csv_rows(
+        outdir / "product_links.csv",
+        link_rows,
+        build_canonical_layer.LINK_FIELDS,
+    )
+    write_csv_rows(outdir / "proof_examples.csv", proof_rows, PROOF_FIELDS)
+    click.echo(
+        f"wrote combined outputs to {outdir} using {len(observed_rows)} observed rows"
+    )
+
+
+if __name__ == "__main__":
+    main()