Add catalog search to review flow

review_products.py

@@ -1,5 +1,6 @@
 from collections import defaultdict
 from datetime import date
+import re
 
 import click
 
@@ -29,6 +30,7 @@ QUEUE_FIELDS = [
 INFO_COLOR = "cyan"
 PROMPT_COLOR = "bright_yellow"
 WARNING_COLOR = "magenta"
+TOKEN_RE = re.compile(r"[A-Z0-9]+")
 
 
 def print_intro_text():
@@ -134,6 +136,13 @@ def sort_related_items(rows):
     )
 
 
+def tokenize_match_text(*values):
+    tokens = set()
+    for value in values:
+        tokens.update(TOKEN_RE.findall((value or "").upper()))
+    return tokens
+
+
 def build_catalog_suggestions(related_rows, purchase_rows, catalog_rows, limit=3):
     normalized_names = {
         row.get("normalized_item_name", "").strip().upper()
@@ -190,6 +199,55 @@ def build_catalog_suggestions(related_rows, purchase_rows, catalog_rows, limit=3
     return suggestions
 
 
+def search_catalog_rows(query, catalog_rows, purchase_rows, current_normalized_item_id, limit=10):
+    query_tokens = tokenize_match_text(query)
+    if not query_tokens:
+        return []
+
+    linked_purchase_counts = defaultdict(int)
+    linked_normalized_ids = defaultdict(set)
+    current_catalog_id = ""
+    for row in purchase_rows:
+        catalog_id = row.get("catalog_id", "")
+        normalized_item_id = row.get("normalized_item_id", "")
+        if catalog_id and normalized_item_id:
+            linked_purchase_counts[catalog_id] += 1
+            linked_normalized_ids[catalog_id].add(normalized_item_id)
+        if normalized_item_id == current_normalized_item_id and catalog_id:
+            current_catalog_id = catalog_id
+
+    ranked_rows = []
+    for row in catalog_rows:
+        catalog_id = row.get("catalog_id", "")
+        if not catalog_id or catalog_id == current_catalog_id:
+            continue
+        catalog_tokens = tokenize_match_text(
+            row.get("catalog_name", ""),
+            row.get("product_type", ""),
+            row.get("variant", ""),
+        )
+        overlap = query_tokens & catalog_tokens
+        if not overlap:
+            continue
+        ranked_rows.append(
+            {
+                "catalog_id": catalog_id,
+                "catalog_name": row.get("catalog_name", ""),
+                "product_type": row.get("product_type", ""),
+                "category": row.get("category", ""),
+                "variant": row.get("variant", ""),
+                "linked_normalized_items": len(linked_normalized_ids.get(catalog_id, set())),
+                "linked_purchase_rows": linked_purchase_counts.get(catalog_id, 0),
+                "score": len(overlap),
+            }
+        )
+
+    ranked_rows.sort(
+        key=lambda row: (-row["score"], row["catalog_name"], row["catalog_id"])
+    )
+    return ranked_rows[:limit]
+
+
 def suggestion_display_rows(suggestions, purchase_rows, catalog_rows):
     linked_purchase_counts = defaultdict(int)
     linked_normalized_ids = defaultdict(set)
@@ -235,6 +293,15 @@ def suggestion_display_rows(suggestions, purchase_rows, catalog_rows):
     return display_rows
 
 
+def print_catalog_rows(rows):
+    for index, row in enumerate(rows, start=1):
+        click.echo(
+            f" [{index}] {row['catalog_name']}, {row.get('product_type', '')}, "
+            f"{row.get('category', '')} ({row['linked_normalized_items']} items, "
+            f"{row['linked_purchase_rows']} rows)"
+        )
+
+
 def build_display_lines(related_rows):
     lines = []
     for index, row in enumerate(sort_related_items(related_rows), start=1):
@@ -267,12 +334,7 @@ def choose_existing_catalog(display_rows, normalized_name, matched_count):
         f"Select the catalog_name to associate {matched_count} items with:",
         fg=INFO_COLOR,
     )
-    for index, row in enumerate(display_rows, start=1):
-        click.echo(
-            f"  [{index}] {row['catalog_name']}, {row.get('product_type', '')}, "
-            f"{row.get('category', '')} ({row['linked_normalized_items']} items, "
-            f"{row['linked_purchase_rows']} rows)"
-        )
+    print_catalog_rows(display_rows)
     choice = click.prompt(
         click.style("selection", fg=PROMPT_COLOR),
         type=click.IntRange(1, len(display_rows)),
@@ -314,18 +376,13 @@ def prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queu
         click.echo(line)
     if suggestions:
         click.echo(f"{len(suggestions)} catalog_name suggestions found:")
-        for index, suggestion in enumerate(suggestions, start=1):
-            click.echo(
-                f" [{index}] {suggestion['catalog_name']}, {suggestion.get('product_type', '')}, "
-                f"{suggestion.get('category', '')} ({suggestion['linked_normalized_items']} items, "
-                f"{suggestion['linked_purchase_rows']} rows)"
-            )
+        print_catalog_rows(suggestions)
     else:
         click.echo("no catalog_name suggestions found")
     prompt_bits = []
     if suggestions:
         prompt_bits.append("[#] link to suggestion")
-    prompt_bits.extend(["[l]ink existing", "[n]ew", "[s]kip", "e[x]clude", "[q]uit"])
+    prompt_bits.extend(["[s]earch", "[n]ew", "e[x]clude", "[q]uit"])
     click.secho("  ".join(prompt_bits) + " >", fg=PROMPT_COLOR)
     action = click.prompt("", type=str, prompt_suffix=" ").strip().lower()
     if action.isdigit() and suggestions:
@@ -346,14 +403,42 @@ def prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queu
     if action == "q":
         return None, None
     if action == "s":
-        return {
-            "normalized_item_id": queue_row["normalized_item_id"],
-            "catalog_id": "",
-            "resolution_action": "skip",
-            "status": "pending",
-            "resolution_notes": queue_row.get("resolution_notes", ""),
-            "reviewed_at": str(date.today()),
-        }, None
+        while True:
+            query = click.prompt(click.style("search", fg=PROMPT_COLOR), default="", show_default=False).strip()
+            if not query:
+                return prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queue_index, queue_total)
+            search_rows = search_catalog_rows(
+                query,
+                catalog_rows,
+                purchase_rows,
+                queue_row["normalized_item_id"],
+            )
+            if not search_rows:
+                click.echo("no matches found")
+                retry = click.prompt(
+                    click.style("search again? [enter=yes, q=no]", fg=PROMPT_COLOR),
+                    default="",
+                    show_default=False,
+                ).strip().lower()
+                if retry == "q":
+                    return prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queue_index, queue_total)
+                continue
+            click.echo(f"{len(search_rows)} search results found:")
+            print_catalog_rows(search_rows)
+            choice = click.prompt(
+                click.style("selection", fg=PROMPT_COLOR),
+                type=click.IntRange(1, len(search_rows)),
+            )
+            chosen_row = search_rows[choice - 1]
+            notes = click.prompt(click.style("link notes", fg=PROMPT_COLOR), default="", show_default=False)
+            return {
+                "normalized_item_id": queue_row["normalized_item_id"],
+                "catalog_id": chosen_row["catalog_id"],
+                "resolution_action": "link",
+                "status": "approved",
+                "resolution_notes": notes,
+                "reviewed_at": str(date.today()),
+            }, None
     if action == "x":
         notes = click.prompt(click.style("exclude notes", fg=PROMPT_COLOR), default="", show_default=False)
         return {
@@ -364,45 +449,6 @@ def prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queu
             "resolution_notes": notes,
             "reviewed_at": str(date.today()),
         }, None
-    if action == "l":
-        display_rows = suggestions or [
-            {
-                "catalog_id": row["catalog_id"],
-                "catalog_name": row["catalog_name"],
-                "reason": "catalog sample",
-                "product_type": row.get("product_type", ""),
-                "category": row.get("category", ""),
-                "linked_normalized_items": 0,
-                "linked_purchase_rows": 0,
-            }
-            for row in catalog_rows[:10]
-            if row.get("catalog_id")
-        ]
-        while True:
-            catalog_id, outcome = choose_existing_catalog(display_rows, normalized_name, matched_count)
-            if outcome == "skip":
-                return {
-                    "normalized_item_id": queue_row["normalized_item_id"],
-                    "catalog_id": "",
-                    "resolution_action": "skip",
-                    "status": "pending",
-                    "resolution_notes": queue_row.get("resolution_notes", ""),
-                    "reviewed_at": str(date.today()),
-                }, None
-            if outcome == "quit":
-                return None, None
-            if outcome == "back":
-                continue
-            break
-        notes = click.prompt(click.style("link notes", fg=PROMPT_COLOR), default="", show_default=False)
-        return {
-            "normalized_item_id": queue_row["normalized_item_id"],
-            "catalog_id": catalog_id,
-            "resolution_action": "link",
-            "status": "approved",
-            "resolution_notes": notes,
-            "reviewed_at": str(date.today()),
-        }, None
     if action != "n":
         click.secho("invalid action", fg=WARNING_COLOR)
         return prompt_resolution(queue_row, related_rows, purchase_rows, catalog_rows, queue_index, queue_total)