#!/usr/bin/env python3
"""Run derive-workspace analyze locally and print LLM token usage from summary.llm."""

from __future__ import annotations

import argparse
import asyncio
import json
import os
import sys
from datetime import date
from pathlib import Path

_BACKEND = Path(__file__).resolve().parents[1]
if str(_BACKEND) not in sys.path:
    sys.path.insert(0, str(_BACKEND))

from dotenv import load_dotenv

load_dotenv(_BACKEND / ".env")
load_dotenv(_BACKEND / ".env.local", override=True)

from investibook.core import derive_workspace as dw
from investibook.db import queries as dbq


def _load_env_data_dir() -> None:
    data = (os.environ.get("DATA_DIR") or "").strip()
    if data:
        dbq.DB_PATH = Path(data) / "db" / "investibook.db"
    elif not Path(dbq.DB_PATH).is_file():
        dbq.DB_PATH = _BACKEND / "data" / "db" / "investibook.db"


async def main() -> int:
    parser = argparse.ArgumentParser(description="Measure Analyze LLM token usage")
    parser.add_argument("--day", default="2026-05-04", help="UTC+8 bucket start (YYYY-MM-DD)")
    parser.add_argument(
        "--metric",
        default="dossiers_created",
        choices=["dossiers_created", "dossiers_workspace_linked_new"],
    )
    parser.add_argument("--project", default="ALL")
    parser.add_argument("--granularity", default="week", choices=["day", "week", "month"])
    parser.add_argument("--limit", type=int, default=0, help="Max dossiers to analyze (0 = all in bucket)")
    args = parser.parse_args()

    _load_env_data_dir()
    if not Path(dbq.DB_PATH).is_file():
        print(f"DB not found: {dbq.DB_PATH}", file=sys.stderr)
        return 1

    day = date.fromisoformat(args.day)
    dd = dbq._db_admin_dashboard_ticket_trends_drill_down(
        day=day,
        metric=args.metric,
        project=args.project,
        granularity=args.granularity,
    )
    dossier_ids = [
        str(x.get("dossier_id") or x.get("id") or "").strip()
        for x in (dd.get("items") or [])
        if str(x.get("dossier_id") or x.get("id") or "").strip()
    ]
    if args.limit > 0:
        dossier_ids = dossier_ids[: args.limit]

    print(f"DB: {dbq.DB_PATH}")
    print(f"Bucket: {args.metric} · {args.granularity} of {args.day} · project={args.project}")
    print(f"Dossiers in drill-down: {len(dd.get('items') or [])}, analyzing: {len(dossier_ids)}")

    scope = dw.AnalyzeScope(
        dossier_ids=dossier_ids,
        metric=args.metric,
        day=args.day,
        project=args.project,
        granularity=args.granularity,
        range_from=None,
        range_to=None,
    )
    out = await dw.analyze(scope)
    summary = out.get("summary") or {}
    llm = summary.get("llm") or {}

    print("\n--- Analyze summary ---")
    print(json.dumps(summary, indent=2, ensure_ascii=False))
    print("\n--- Token usage ---")
    print(f"LLM calls (with usage): {llm.get('calls', 0)}")
    print(f"Input tokens:  {llm.get('input_tokens', 0):,}")
    print(f"Output tokens: {llm.get('output_tokens', 0):,}")
    print(f"Total tokens:  {llm.get('total_tokens', 0):,}")
    scanned = int(summary.get("dossiers_scanned") or 0)
    skipped = int(summary.get("dossiers_skipped_prior_state") or 0)
    if scanned:
        denom = scanned - skipped
        if denom > 0 and llm.get("calls"):
            print(
                f"Per analyzed dossier (excl. skipped): "
                f"~{int(llm.get('input_tokens', 0)) // denom:,} in / "
                f"~{int(llm.get('output_tokens', 0)) // denom:,} out"
            )
    return 0


if __name__ == "__main__":
    raise SystemExit(asyncio.run(main()))