Improve product review display workflow

This commit is contained in:
ben
2026-03-17 12:25:47 -04:00
parent e494386e64
commit 7b8141cd42
5 changed files with 372 additions and 46 deletions

View File

@@ -1,6 +1,5 @@
from collections import defaultdict
from datetime import date
from pathlib import Path
import click
@@ -99,17 +98,140 @@ def save_catalog_rows(path, rows):
write_csv_rows(path, rows, build_purchases.CATALOG_FIELDS)
def prompt_resolution(queue_row, catalog_rows):
INFO_COLOR = "cyan"
PROMPT_COLOR = "bright_yellow"
WARNING_COLOR = "magenta"
def sort_related_items(rows):
return sorted(
rows,
key=lambda row: (
row.get("purchase_date", ""),
row.get("order_id", ""),
int(row.get("line_no", "0") or "0"),
),
reverse=True,
)
def build_canonical_suggestions(related_rows, catalog_rows, limit=3):
normalized_names = {
row.get("normalized_item_name", "").strip().upper()
for row in related_rows
if row.get("normalized_item_name", "").strip()
}
upcs = {
row.get("upc", "").strip()
for row in related_rows
if row.get("upc", "").strip()
}
suggestions = []
seen_ids = set()
def add_matches(rows, reason):
for row in rows:
canonical_product_id = row.get("canonical_product_id", "")
if not canonical_product_id or canonical_product_id in seen_ids:
continue
seen_ids.add(canonical_product_id)
suggestions.append(
{
"canonical_product_id": canonical_product_id,
"canonical_name": row.get("canonical_name", ""),
"reason": reason,
}
)
if len(suggestions) >= limit:
return True
return False
exact_upc_rows = [
row
for row in catalog_rows
if row.get("upc", "").strip() and row.get("upc", "").strip() in upcs
]
if add_matches(exact_upc_rows, "exact upc"):
return suggestions
exact_name_rows = [
row
for row in catalog_rows
if row.get("canonical_name", "").strip().upper() in normalized_names
]
if add_matches(exact_name_rows, "exact normalized name"):
return suggestions
contains_rows = []
for row in catalog_rows:
canonical_name = row.get("canonical_name", "").strip().upper()
if not canonical_name:
continue
for normalized_name in normalized_names:
if normalized_name in canonical_name or canonical_name in normalized_name:
contains_rows.append(row)
break
add_matches(contains_rows, "canonical name contains match")
return suggestions
def build_display_lines(queue_row, related_rows):
lines = []
for row in sort_related_items(related_rows):
lines.append(
" - {purchase_date} | {line_total} | {raw_item_name} | {normalized_item_name} | "
"{upc} | {retailer}".format(
purchase_date=row.get("purchase_date", ""),
line_total=row.get("line_total", ""),
raw_item_name=row.get("raw_item_name", ""),
normalized_item_name=row.get("normalized_item_name", ""),
upc=row.get("upc", ""),
retailer=row.get("retailer", ""),
)
)
if row.get("image_url"):
lines.append(f" image: {row['image_url']}")
if not lines:
lines.append(" - no matched item rows found")
return lines
def prompt_resolution(queue_row, related_rows, catalog_rows, queue_index, queue_total):
suggestions = build_canonical_suggestions(related_rows, catalog_rows)
click.echo("")
click.secho(
f"Review observed product ({queue_index}/{queue_total})",
fg=INFO_COLOR,
)
click.echo(
"Resolve this observed product group to an existing canonical_name, "
"a new canonical_name, exclude it, or skip it."
)
click.echo(f"observed_product_id: {queue_row['observed_product_id']}")
click.echo(f"retailer: {queue_row['retailer']}")
click.echo(f"raw names: {queue_row['raw_item_names']}")
click.echo(f"normalized names: {queue_row['normalized_names']}")
click.echo(f"upcs: {queue_row['upc_values']}")
click.echo(f"example prices: {queue_row['example_prices']}")
click.echo(f"seen count: {queue_row['seen_count']}")
click.echo("actions: [l]ink existing [n]ew canonical [x]exclude [s]kip [q]uit")
action = click.prompt("action", type=click.Choice(["l", "n", "x", "s", "q"]))
click.echo(f"observed_name(s): {queue_row['normalized_names']}")
click.echo(f"upc(s): {queue_row['upc_values']}")
click.echo(f"seen_count: {queue_row['seen_count']}")
click.secho("matched items:", fg=INFO_COLOR)
for line in build_display_lines(queue_row, related_rows):
click.echo(line)
if suggestions:
click.secho("suggested canonical_name values:", fg=INFO_COLOR)
for suggestion in suggestions:
click.echo(
f" - {suggestion['canonical_product_id']} | {suggestion['canonical_name']} "
f"({suggestion['reason']})"
)
else:
click.secho("no deterministic canonical suggestions found", fg=WARNING_COLOR)
click.secho(
"actions: [l]ink existing [n]ew canonical e[x]clude [s]kip [q]uit",
fg=PROMPT_COLOR,
)
action = click.prompt(
click.style("action", fg=PROMPT_COLOR),
type=click.Choice(["l", "n", "x", "s", "q"]),
)
if action == "q":
return None, None
if action == "s":
@@ -122,7 +244,11 @@ def prompt_resolution(queue_row, catalog_rows):
"reviewed_at": str(date.today()),
}, None
if action == "x":
notes = click.prompt("exclude notes", default="", show_default=False)
notes = click.prompt(
click.style("exclude notes", fg=PROMPT_COLOR),
default="",
show_default=False,
)
return {
"observed_product_id": queue_row["observed_product_id"],
"canonical_product_id": "",
@@ -132,11 +258,28 @@ def prompt_resolution(queue_row, catalog_rows):
"reviewed_at": str(date.today()),
}, None
if action == "l":
click.echo("existing canonicals:")
for row in catalog_rows[:10]:
click.echo(f" {row['canonical_product_id']} {row['canonical_name']}")
canonical_product_id = click.prompt("canonical product id", type=str)
notes = click.prompt("link notes", default="", show_default=False)
display_rows = suggestions or [
{
"canonical_product_id": row["canonical_product_id"],
"canonical_name": row["canonical_name"],
"reason": "catalog sample",
}
for row in catalog_rows[:10]
]
click.secho("existing canonical_name values:", fg=INFO_COLOR)
for row in display_rows:
click.echo(
f" - {row['canonical_product_id']} | {row['canonical_name']} ({row['reason']})"
)
canonical_product_id = click.prompt(
click.style("canonical product id", fg=PROMPT_COLOR),
type=str,
)
notes = click.prompt(
click.style("link notes", fg=PROMPT_COLOR),
default="",
show_default=False,
)
return {
"observed_product_id": queue_row["observed_product_id"],
"canonical_product_id": canonical_product_id,
@@ -146,10 +289,22 @@ def prompt_resolution(queue_row, catalog_rows):
"reviewed_at": str(date.today()),
}, None
canonical_name = click.prompt("canonical name", type=str)
category = click.prompt("category", default="", show_default=False)
product_type = click.prompt("product type", default="", show_default=False)
notes = click.prompt("notes", default="", show_default=False)
canonical_name = click.prompt(click.style("canonical name", fg=PROMPT_COLOR), type=str)
category = click.prompt(
click.style("category", fg=PROMPT_COLOR),
default="",
show_default=False,
)
product_type = click.prompt(
click.style("product type", fg=PROMPT_COLOR),
default="",
show_default=False,
)
notes = click.prompt(
click.style("notes", fg=PROMPT_COLOR),
default="",
show_default=False,
)
canonical_product_id = stable_id("gcan", f"manual|{canonical_name}|{category}|{product_type}")
canonical_row = {
"canonical_product_id": canonical_product_id,
@@ -197,11 +352,17 @@ def main(purchases_csv, queue_csv, resolutions_csv, catalog_csv, limit, refresh_
resolution_lookup = build_purchases.load_resolution_lookup(resolution_rows)
catalog_by_id = {row["canonical_product_id"]: row for row in catalog_rows if row.get("canonical_product_id")}
rows_by_observed = defaultdict(list)
for row in purchase_rows:
observed_product_id = row.get("observed_product_id", "")
if observed_product_id:
rows_by_observed[observed_product_id].append(row)
reviewed = 0
for queue_row in queue_rows:
for index, queue_row in enumerate(queue_rows, start=1):
if limit and reviewed >= limit:
break
result = prompt_resolution(queue_row, catalog_rows)
related_rows = rows_by_observed.get(queue_row["observed_product_id"], [])
result = prompt_resolution(queue_row, related_rows, catalog_rows, index, len(queue_rows))
if result == (None, None):
break
resolution_row, canonical_row = result