AiDOPWarehouse/doc/_tmp_toc_verify.py

# -*- coding: utf-8 -*-
"""核对目录里的页码与正文标题实际所在页是否一致。"""
import io
import os
import re
import sys

import fitz

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

HERE = os.path.dirname(os.path.abspath(__file__))
PDF = os.path.join(HERE, "_tmp_inspect", "current.pdf")

d = fitz.open(PDF)
print(f"PDF 共 {d.page_count} 页")

# 目录只在正文开始前（响应函在第 9 页），限定在前 8 页里找
HEAD = 8
toc_pages = [i for i in range(1, HEAD + 1)
             if len(re.findall(r"[^\s]\s+\d{1,3}\s*$",
                              d[i - 1].get_text(), re.M)) >= 5]
print("目录所在页：", toc_pages)

entries = []
for i in toc_pages:
    for line in d[i - 1].get_text().split("\n"):
        m = re.match(r"^(.*?)[\s.\u2026]+(\d{1,3})$", line.strip())
        if m and len(m.group(1).strip()) >= 3:
            entries.append((m.group(1).strip(), int(m.group(2))))
print(f"解析出目录条目 {len(entries)} 条\n")

# 正文里每个标题实际出现的页（跳过目录页本身）
body_text = {i: d[i - 1].get_text() for i in range(1, d.page_count + 1)}
bad = 0
for title, claim in entries:
    key = re.sub(r"\s+", "", title)
    hit = None
    for i in range(HEAD + 1, d.page_count + 1):
        if key and key in re.sub(r"\s+", "", body_text[i]):
            hit = i
            break
    if hit is None:
        print(f"  ? 正文未找到：{title[:44]}（目录写 {claim}）")
        bad += 1
    elif hit != claim:
        print(f"  × {title[:44]}　目录 {claim} ≠ 实际 {hit}")
        bad += 1
print(f"\n不一致 {bad} 条 / 共 {len(entries)} 条")
d.close()