ai大模型平台横向试了 3 个：4 类产物只有 1 个落盘成文件｜PRIVO AI智能操作助手 三合一·网页+桌面+IDE

import zipfile, pathlib
from docx import Document
from openpyxl import load_workbook
from pptx import Presentation
from lxml import html as LH

def audit(p: pathlib.Path) -> dict:
    r = {"file": p.name, "kb": round(p.stat().st_size / 1024, 1)}
    # 1) 魔数：docx/xlsx/pptx 本质都是 zip
    r["magic"] = p.open("rb").read(4) == b"PK\x03\x04"
    if p.suffix in {".docx", ".xlsx", ".pptx"}:
        with zipfile.ZipFile(p) as z:
            r["zip"] = z.testzip() is None
            r["parts"] = len(z.namelist())
    if p.suffix == ".docx":
        d = Document(p)
        r["paras"] = len([x for x in d.paragraphs if x.text.strip()])
        r["heading"] = len([x for x in d.paragraphs if x.style.name.startswith("Heading")])
    elif p.suffix == ".xlsx":
        wb = load_workbook(p)          # 不带 data_only，读公式原文
        ws = wb.active
        r["formula"] = sum(1 for row in ws.iter_rows() for c in row
                           if isinstance(c.value, str) and c.value.startswith("="))
    elif p.suffix == ".pptx":
        r["slides"] = len(Presentation(p).slides)
    elif p.suffix == ".html":
        doc = LH.fromstring(p.read_text(encoding="utf-8", errors="ignore"))
        r["title"] = bool(doc.xpath("//title/text()"))
        r["block"] = len(doc.xpath("//section|//div[@class]"))
    return r

# 每份交付一个日期目录，四类产物各一个子目录，版本递增不进文件名
mkdir -p D:/deliver/2026-11-08/{doc,sheet,deck,site}

import shutil, pathlib, datetime

def archive(src: pathlib.Path, root: pathlib.Path, sub: str) -> pathlib.Path:
    day = datetime.date.today().isoformat()
    dst_dir = root / day / sub
    dst_dir.mkdir(parents=True, exist_ok=True)
    dst = dst_dir / src.name            # 版本靠目录区分，不靠文件名
    shutil.move(str(src), dst)
    return dst