summarize_book.py

import argparse
import os
from pathlib import Path

from dotenv import load_dotenv

from epub_extractor import extract
from epub_processor import process_epub
from send_prompts import PromptProcessor


def summarize_book(epub_path: str, out_dir: str | None):
    # Defaults optimizados para capítulos largos
    MODEL = os.getenv("OPENAI_MODEL", "gpt-4o-mini")
    OUTPUT_TOKENS = int(os.getenv("OPENAI_OUTPUT_MAX_TOKENS", "800"))
    REQ_DELAY = float(os.getenv("OPENAI_REQUEST_DELAY_SEC", "0.5"))  # Más delay para estabilidad

    # 1) Extract
    extract(epub_path)

    # 2) Process to text
    epub = Path(epub_path)
    books_folder = Path('books')
    book_folder = books_folder / "".join([c for c in epub.stem if c.isalpha() or c.isdigit() or c==' ']).rstrip()
    if not book_folder.exists():
        raise FileNotFoundError(f"Processed book folder not found: {book_folder}")
    process_epub(book_folder, character_limit=999999999)

    # 3) Summarize chapters only into a flat responses folder
    api_key = os.getenv("OPENAI_API_KEY")
    if not api_key:
        raise RuntimeError("OPENAI_API_KEY not set. Add it to .env or env vars.")

    # Determine output dir
    responses_dir = Path(out_dir) if out_dir else (book_folder / "responses")
    responses_dir.mkdir(parents=True, exist_ok=True)

    processor = PromptProcessor(
        book_path=str(book_folder),
        api_key=api_key,
        model=MODEL,
        output_max_tokens=OUTPUT_TOKENS,
        request_delay_sec=REQ_DELAY,
        responses_dir_override=str(responses_dir),
    )
    processor.process_files()

    # 4) Global book summary from chapter summaries
    pieces = []
    for md in sorted(responses_dir.glob('ch*_summary.md')):
        try:
            text = md.read_text(encoding='utf-8')
        except Exception:
            continue
        pieces.append(f"\n\n## Source: {md.name}\n\n{text}\n")

    if pieces:
        big_prompt = "\n\n".join(pieces)
        # Reuse processor client with capped tokens
        content = processor.send_prompt(big_prompt)
        (responses_dir / 'BOOK_summary.md').write_text(content, encoding='utf-8')
        print(f"Global summary generated at: {responses_dir / 'BOOK_summary.md'}")


def main():
    load_dotenv()
    parser = argparse.ArgumentParser(description="Summarize an EPUB into Spanish chapter summaries + global summary.")
    parser.add_argument("epub_path", help="Path to .epub file")
    parser.add_argument("out_dir", nargs='?', default=None, help="Optional output directory for summaries")
    args = parser.parse_args()

    summarize_book(
        epub_path=args.epub_path,
        out_dir=args.out_dir,
    )


if __name__ == "__main__":
    main()