#!/usr/bin/env python3
"""Re-runnable Anthropic model discovery.

GET /v1/models (paginated, limit=100) -> sanitized sources/anthropic/models-api-raw.json
Diff against the previous file -> sources/anthropic/models-diff.json, prints NEW / REMOVED / CHANGED ids.

Usage:  python3 scripts/discover_anthropic_models.py [--limit N] [--dry-run]
Auth comes from .env via scripts/live.py (never printed). Each call is logged to reports/live-requests.jsonl.
"""
from __future__ import annotations

import argparse
import json
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parent))
import live  # noqa: E402  (scripts/live.py)

ROOT = Path(__file__).resolve().parents[1]
RAW = ROOT / "sources" / "anthropic" / "models-api-raw.json"
DIFF = ROOT / "sources" / "anthropic" / "models-diff.json"


def fetch_all(limit: int) -> dict:
    data: list[dict] = []
    after: str | None = None
    pages = 0
    while True:
        q = f"?limit={limit}" + (f"&after_id={after}" if after else "")
        st, body, _ = live.anthropic_request("GET", f"/v1/models{q}", note=f"discover models page {pages + 1}")
        if st != 200:
            raise SystemExit(f"GET /v1/models -> HTTP {st}: {json.dumps(body)[:300]}")
        data.extend(body.get("data", []))
        pages += 1
        if not body.get("has_more"):
            break
        after = body.get("last_id")
        if not after:
            break
    return {"data": data, "has_more": False, "first_id": data[0]["id"] if data else None,
            "last_id": data[-1]["id"] if data else None, "_pages": pages, "_retrieved_at": live.now()}


def index(doc: dict) -> dict[str, dict]:
    return {m["id"]: m for m in doc.get("data", [])}


def main() -> None:
    ap = argparse.ArgumentParser()
    ap.add_argument("--limit", type=int, default=100)
    ap.add_argument("--dry-run", action="store_true", help="fetch and diff, do not write")
    args = ap.parse_args()

    prev = json.loads(RAW.read_text()) if RAW.exists() else {"data": []}
    cur = fetch_all(args.limit)
    pi, ci = index(prev), index(cur)
    new = sorted(set(ci) - set(pi))
    removed = sorted(set(pi) - set(ci))
    changed = sorted(i for i in set(ci) & set(pi) if ci[i] != pi[i])
    diff = {"retrieved_at": cur["_retrieved_at"], "previous_count": len(pi), "current_count": len(ci),
            "new": new, "removed": removed, "changed": [{"id": i, "before": pi[i], "after": ci[i]} for i in changed],
            "pages_fetched": cur["_pages"]}
    print(f"models: {len(ci)} (previous {len(pi)}), pages={cur['_pages']}")
    for i in new:
        print(f"NEW      {i}  created_at={ci[i].get('created_at')}")
    for i in removed:
        print(f"REMOVED  {i}")
    for i in changed:
        print(f"CHANGED  {i}")
    if not (new or removed or changed):
        print("no changes")
    if args.dry_run:
        return
    out = {k: v for k, v in cur.items() if not k.startswith("_")}
    live.save_sanitized(out, RAW)
    live.save_sanitized(diff, DIFF)
    print(f"wrote {RAW.relative_to(ROOT)} and {DIFF.relative_to(ROOT)}")


if __name__ == "__main__":
    main()