AiDOPWarehouse/doc/_bid_preview.py

# -*- coding: utf-8 -*-
"""技术方案排版自检：导出 PDF、渲染指定页预览、报告版面留白过多的页。

用法：python doc/_bid_preview.py <文件名关键字> [页码 ...] [--export]
例：  python doc/_bid_preview.py 第2章 1 5 9
"""
import io
import os
import sys
import time

import fitz

HERE = os.path.dirname(os.path.abspath(__file__))
PREV = os.path.join(HERE, "_bid_preview_pages")
FILL_WARN = 0.62

sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8")


def find_docx(keyword):
    hits = [n for n in os.listdir(HERE)
            if n.endswith(".docx") and not n.startswith("~$") and keyword in n]
    if not hits:
        raise SystemExit(f"未找到含「{keyword}」的 docx")
    if len(hits) > 1:
        print("匹配到多个，取第一个：", hits)
    return os.path.join(HERE, hits[0])


def export_pdf(src, pdf):
    import win32com.client as win32
    for attempt in range(4):
        try:
            word = win32.gencache.EnsureDispatch("Word.Application")
            word.Visible = False
            d = word.Documents.Open(src, ReadOnly=True)
            d.ExportAsFixedFormat(pdf, 17)
            d.Close(False)
            word.Quit()
            return
        except Exception as e:                    # Word COM 偶发拒绝调用
            print(f"  导出重试 {attempt + 1}/4：{e}")
            time.sleep(4)
    raise SystemExit("Word 导出 PDF 失败")


def report_gaps(doc):
    print(f"版面留白检查（内容占比 < {FILL_WARN:.0%}）：")
    hit = False
    for i, pg in enumerate(doc, 1):
        r = pg.rect
        bottoms = [b[3] for b in pg.get_text("blocks") if b[3] < r.height - 40]
        bottoms += [bb.y1 for bb in
                    (pg.get_image_bbox(x) for x in pg.get_images(full=True)) if bb]
        if not bottoms:
            print(f"  P{i:<3} 空白页")
            hit = True
            continue
        used = (max(bottoms) - 60) / (r.height - 120)
        if used < FILL_WARN:
            print(f"  P{i:<3} 内容占比 {used * 100:5.1f}%")
            hit = True
    if not hit:
        print("  无")


def main():
    args = sys.argv[1:]
    if not args:
        raise SystemExit(__doc__)
    keyword = args[0]
    pages = [int(a) for a in args[1:] if a.isdigit()]
    src = find_docx(keyword)
    # PDF 按源文档命名，避免不同章节复用同一个缓存文件
    stem = os.path.splitext(os.path.basename(src))[0]
    pdf = os.path.join(HERE, "_bid_preview", stem + ".pdf")
    os.makedirs(os.path.dirname(pdf), exist_ok=True)
    os.makedirs(PREV, exist_ok=True)
    print("文档：", os.path.basename(src))

    stale = (not os.path.exists(pdf)
             or os.path.getmtime(pdf) < os.path.getmtime(src))
    if "--export" in args or stale:
        export_pdf(src, pdf)

    doc = fitz.open(pdf)
    print("PDF 页数：", doc.page_count)
    for p in pages:
        if 1 <= p <= doc.page_count:
            doc[p - 1].get_pixmap(dpi=110).save(
                os.path.join(PREV, f"p{p:02d}.png"))
    if pages:
        print("已渲染预览页：", pages)
    report_gaps(doc)


if __name__ == "__main__":
    main()