AiDOPWarehouse/doc/_word_toc_one.py

# -*- coding: utf-8 -*-
"""Refresh TOC fields in one docx via a fresh Word process."""
from __future__ import annotations

import hashlib
import shutil
import sys
import tempfile
import time
from pathlib import Path

WD_COLOR_BLACK = 0
WD_STAT_PAGES = 2


def retry(fn, times=8, wait=0.8):
    last = None
    for i in range(times):
        try:
            return fn()
        except Exception as exc:
            last = exc
            time.sleep(wait * (i + 1))
    raise last


def main() -> int:
    if len(sys.argv) < 2:
        return 2
    src = Path(sys.argv[1])
    tmp_dir = Path(tempfile.gettempdir()) / "aidop_toc_word"
    tmp_dir.mkdir(exist_ok=True)
    tmp = tmp_dir / f"{hashlib.md5(str(src).encode('utf-8')).hexdigest()}_{int(time.time())}.docx"
    shutil.copy2(src, tmp)

    import pythoncom
    import win32com.client as win32

    pythoncom.CoInitialize()
    word = None
    doc = None
    try:
        word = retry(lambda: win32.DispatchEx("Word.Application"))
        word.Visible = False
        try:
            word.DisplayAlerts = 0
        except Exception:
            pass
        try:
            word.AutomationSecurity = 3
        except Exception:
            pass
        time.sleep(1)
        doc = retry(
            lambda: word.Documents.Open(
                str(tmp),
                ConfirmConversions=False,
                ReadOnly=False,
                AddToRecentFiles=False,
            )
        )
        time.sleep(1)
        retry(lambda: doc.Fields.Update())
        def update_tocs():
            for i in range(1, doc.TablesOfContents.Count + 1):
                doc.TablesOfContents(i).Update()
        retry(update_tocs)
        try:
            doc.Repaginate()
        except Exception:
            pass
        retry(update_tocs)
        for name in ("TOC 1", "TOC 2", "TOC 3", "TOC 4", "Hyperlink"):
            try:
                doc.Styles(name).Font.Color = WD_COLOR_BLACK
            except Exception:
                pass
        pages = retry(lambda: doc.ComputeStatistics(WD_STAT_PAGES))
        lines = 0
        if doc.TablesOfContents.Count:
            lines = len([x for x in doc.TablesOfContents(1).Range.Text.split("\r") if x.strip()])
        retry(doc.Save)
        retry(lambda: doc.Close(False))
        doc = None
        retry(word.Quit)
        word = None
        time.sleep(0.5)
        shutil.copy2(tmp, src)
        print(f"pages={pages} toc={lines}", flush=True)
        return 0
    finally:
        try:
            if doc is not None:
                doc.Close(False)
        except Exception:
            pass
        try:
            if word is not None:
                word.Quit()
        except Exception:
            pass
        try:
            pythoncom.CoUninitialize()
        except Exception:
            pass
        try:
            tmp.unlink(missing_ok=True)
        except Exception:
            pass


if __name__ == "__main__":
    raise SystemExit(main())