AiDOPWarehouse/doc/_extend_delivery_front_matter.py

# -*- coding: utf-8 -*-
"""Apply S0 blueprint front matter to all S0-S4 delivery docx (recursive)."""
from __future__ import annotations

import importlib.util
import shutil
import tempfile
from pathlib import Path

from docx import Document
from docx.oxml.ns import qn
from docx.text.paragraph import Paragraph

BASE = Path(
    r"C:\Users\skygu\OneDrive\Projects\AIDOP\项目\项目管理\产互联项目管理\交付文档"
)
MODULES = ["S0", "S1", "S2", "S3", "S4"]


def load_module(name: str, filename: str):
    path = Path(__file__).with_name(filename)
    spec = importlib.util.spec_from_file_location(name, path)
    mod = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(mod)
    return mod


def iter_delivery_docx() -> list[Path]:
    files: list[Path] = []
    for module in MODULES:
        mod_dir = BASE / module
        if not mod_dir.is_dir():
            continue
        for path in sorted(mod_dir.rglob("*.docx")):
            if path.name.startswith("~$"):
                continue
            files.append(path)
    return files


def has_toc(doc: Document) -> bool:
    return any(p.text.strip() == "目录" for p in doc.paragraphs)


def has_page_break(p: Paragraph) -> bool:
    return bool(p._element.findall(".//" + qn("w:br")))


def front_matter_status(doc: Document) -> str:
    if not has_toc(doc):
        return "skip-no-toc"

    has_control = any(p.text.strip() == "文档控制" for p in doc.paragraphs)
    has_change = any(p.text.strip() == "更改记录" for p in doc.paragraphs)
    has_version = any(p.text.strip() == "版本记录" for p in doc.paragraphs)
    has_old_meta = any(
        p.text.strip().startswith(prefix)
        for p in doc.paragraphs[:20]
        for prefix in ("文档作者：", "创建日期：", "当前版本：")
    )
    has_cover_table = any(
        table.rows
        and table.rows[0].cells[0].text.strip() == "文档编号"
        and len(table.rows) == 5
        for table in doc.tables
    )

    version_break = False
    toc_break = False
    for p in doc.paragraphs:
        t = p.text.strip()
        if t == "版本记录":
            version_break = has_page_break(p)
        if t == "目录":
            toc_break = has_page_break(p)

    if has_control or has_change or has_old_meta or not has_cover_table:
        return "rebuild"
    if not has_version:
        return "rebuild"
    if not version_break or not toc_break:
        return "adjust"
    return "ok"


def infer_module_kind(path: Path) -> tuple[str, str]:
    rel = path.relative_to(BASE)
    module = rel.parts[0]
    kind = "blue" if "蓝图" in path.name else "req"
    return module, kind


def save_with_fallback(doc: Document, path: Path) -> None:
    try:
        doc.save(path)
    except PermissionError:
        tmp = Path(tempfile.gettempdir()) / f"docfix_{path.name}"
        doc.save(tmp)
        try:
            shutil.copy2(tmp, path)
            tmp.unlink(missing_ok=True)
        except PermissionError:
            print(f"  !! {path.name} 被占用，已写入: {tmp}")


def process_file(path: Path) -> None:
    rel = path.relative_to(BASE)
    doc = Document(path)
    status = front_matter_status(doc)
    if status == "skip-no-toc":
        print(f"{rel}: skip (无目录)")
        return
    if status == "ok":
        print(f"{rel}: ok")
        return

    module, kind = infer_module_kind(path)
    adjust = load_module("adjust", "_adjust_front_matter.py")

    if status == "rebuild":
        unify = load_module("unify", "_unify_doc_format.py")
        restore = load_module("restore", "_restore_cover_format.py")
        versions = load_module("versions", "_unify_version_records.py")

        unify.unify_document(path, module, kind)
        restore.restore_file(path)
        versions.process_file(path)
        adjust.adjust_file(path)
        print(f"{rel}: rebuilt + formatted")
        return

    adjust.adjust_file(path)
    print(f"{rel}: adjusted page breaks")


def main() -> None:
    for path in iter_delivery_docx():
        process_file(path)


if __name__ == "__main__":
    main()