instant-bom/bom_assistant/suppliers/lcsc/make_enrich_input.py

#!/usr/bin/env python3
"""
Generate a compact input file for web-UI LLM enrichment.

Outputs: enrich_input.json — only subcategories missing description or examples.

Usage:
    python -m bom_assistant.suppliers.lcsc.make_enrich_input                         # all
    python -m bom_assistant.suppliers.lcsc.make_enrich_input --limit 100             # batch 1
    python -m bom_assistant.suppliers.lcsc.make_enrich_input --skip 100 --limit 100  # batch 2
    python -m bom_assistant.suppliers.lcsc.make_enrich_input --skip 200 --limit 100  # batch 3
    # attach each enrich_input.json to claude.ai, apply output, move to next batch
"""
from __future__ import annotations

import argparse
import json
from pathlib import Path

_CATS_PATH = Path(__file__).parent / "categories.json"
_OUT_PATH = Path(__file__).parent / "enrich_input.json"


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("--limit", type=int, default=None, help="max subcategories to include")
    parser.add_argument("--skip", type=int, default=0, help="skip first N subcategories")
    args = parser.parse_args()

    cats = json.loads(_CATS_PATH.read_text(encoding="utf-8"))

    needed: list[dict] = []
    for cat in cats:
        for sub in cat["subcategories"]:
            missing_desc = not sub.get("description")
            missing_ex = not sub.get("examples")
            if not missing_desc and not missing_ex:
                continue
            entry: dict = {
                "id": sub["id"],
                "category": cat["category"],
                "name": sub["name"],
            }
            if sub.get("examples"):
                entry["examples"] = sub["examples"]
            if missing_ex:
                entry["need_examples"] = True
            needed.append(entry)

    needed = needed[args.skip :]
    if args.limit:
        needed = needed[: args.limit]

    _OUT_PATH.write_text(
        json.dumps(needed, indent=2, ensure_ascii=False), encoding="utf-8"
    )

    need_ex = sum(1 for e in needed if e.get("need_examples"))
    total_needed = sum(
        1 for c in cats for s in c["subcategories"]
        if not s.get("description") or not s.get("examples")
    )
    remaining = total_needed - args.skip - len(needed)
    print(f"Wrote {len(needed)} subcategories to {_OUT_PATH.name}")
    print(f"  {len(needed)} need description, {need_ex} also need examples")
    if remaining > 0:
        print(f"  {remaining} more remaining after this batch")


if __name__ == "__main__":
    main()