"""Extract text from original PDF files using Claude Opus Vision OCR."""

import asyncio
import sys
import time
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parent.parent / "mcp-server" / "src"))

from dotenv import load_dotenv
load_dotenv(Path.home() / ".env")

from legal_mcp.services import extractor

ORIGINALS_DIR = Path("/home/chaim/legal-ai/data/cases/1130-25/documents/originals")
OUTPUT_DIR = ORIGINALS_DIR / "extracted"


async def main():
    OUTPUT_DIR.mkdir(exist_ok=True)

    pdfs = sorted(ORIGINALS_DIR.glob("*.pdf"))
    print(f"Found {len(pdfs)} PDFs\n")

    total_cost = 0.0
    total_pages = 0
    total_time = 0.0

    for pdf in pdfs:
        print(f"{'=' * 60}")
        print(f"Processing: {pdf.name}")
        print(f"  Size: {pdf.stat().st_size:,} bytes")

        t0 = time.time()
        text, page_count = await extractor.extract_text(str(pdf))
        elapsed = time.time() - t0

        total_pages += page_count
        total_time += elapsed

        # Estimate cost (Opus: $15/M input, $75/M output, ~1000 tokens per image)
        # Rough: ~$0.05 per page for image input + output
        est_cost = page_count * 0.05
        total_cost += est_cost

        # Save extracted text
        out_file = OUTPUT_DIR / f"{pdf.stem}.md"
        out_file.write_text(text, encoding="utf-8")

        print(f"  Pages: {page_count}")
        print(f"  Extracted: {len(text):,} chars, {len(text.split()):,} words")
        print(f"  Time: {elapsed:.1f}s ({elapsed/max(page_count,1):.1f}s/page)")
        print(f"  Est. cost: ${est_cost:.3f}")
        print(f"  Saved to: {out_file.name}")
        print()

    print(f"{'=' * 60}")
    print(f"TOTAL")
    print(f"  Documents: {len(pdfs)}")
    print(f"  Pages: {total_pages}")
    print(f"  Time: {total_time:.1f}s")
    print(f"  Est. cost: ${total_cost:.3f}")


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