# SPDX-License-Identifier: AGPL-3.0-or-later
"""
PDF -> Word (.docx) for ToolDabba's "PDF to Word" tool (https://tooldabba.com/pdf-to-word/).

Uses pdf2docx (MIT) and PyMuPDF (AGPL-3.0, Artifex). Because PyMuPDF is AGPL, this program is AGPL too and its
complete source is published at https://tooldabba.com/source/pdf_to_docx.py. Neither library is modified.

    python pdf_to_docx.py input.pdf output.docx

Prints one line of JSON on success, e.g. {"pages": 3, "scanned": false}. Exit codes:
3 = needs a password, 4 = too many pages, 5 = not a readable PDF, 1 = conversion failed.
"""
import json
import sys

import pymupdf
from pdf2docx import Converter

MAX_PAGES = 100


def done(code, **info):
    print(json.dumps(info))
    sys.exit(code)


def main(src, dst):
    try:
        doc = pymupdf.open(src)
    except Exception:
        done(5, error="invalid")
    if doc.needs_pass:
        done(3, error="password")
    pages = doc.page_count
    if pages > MAX_PAGES:
        done(4, error="too-many-pages")
    # A scan has (almost) no text layer: Word will get pictures of the pages, not editable text
    chars = sum(len(page.get_text("text").strip()) for page in doc)
    doc.close()

    try:
        converter = Converter(src)
        converter.convert(dst, multi_processing=False)
        converter.close()
    except Exception:
        done(1, error="failed")
    done(0, pages=pages, scanned=chars < 20 * pages)


if __name__ == "__main__":
    main(sys.argv[1], sys.argv[2])