LQAgentPlatform/utils_test/Chunk_Split_Test/check_leak_detail.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
检查特定文件的章节泄漏详情
"""

import sys
from pathlib import Path

project_root = Path(__file__).parent.parent.parent

from test_chunk_split_fix import build_test_facade
from core.construction_review.component.doc_worker.interfaces import DocumentSource


def check_file(file_path: Path):
    print(f"\nChecking: {file_path.name}")
    print("=" * 80)

    source = DocumentSource(path=str(file_path))
    facade = build_test_facade()
    result = facade.process(source)

    chunks = result.get("chunks", [])

    # 找到第九章和第十章的 chunks
    chapter9_chunks = []
    chapter10_chunks = []

    for chunk in chunks:
        label = chunk.get("section_label", "")
        if label.startswith("第九章"):
            chapter9_chunks.append(chunk)
        elif label.startswith("第十章"):
            chapter10_chunks.append(chunk)

    print(f"\n第九章 chunks: {len(chapter9_chunks)}")
    for chunk in chapter9_chunks:
        print(f"  - {chunk.get('chunk_id')}: {chunk.get('section_label')}")

    print(f"\n第十章 chunks: {len(chapter10_chunks)}")
    for chunk in chapter10_chunks:
        print(f"  - {chunk.get('chunk_id')}: {chunk.get('section_label')}")

    # 检查第九章最后一个 chunk 的内容是否包含"第十章"
    if chapter9_chunks:
        last_chunk = chapter9_chunks[-1]
        content = last_chunk.get("review_chunk_content", "")
        print(f"\n第九章最后一个 chunk: {last_chunk.get('chunk_id')}")
        print(f"  section_label: {last_chunk.get('section_label')}")
        print(f"  content长度: {len(content)}")

        # 搜索"第十章"
        idx = content.find("第十章")
        if idx >= 0:
            print(f"\n  [WARNING] 发现'第十章'在位置 {idx}")
            # 显示上下文
            start = max(0, idx - 100)
            end = min(len(content), idx + 100)
            print(f"  上下文: ...{content[start:end]}...")
        else:
            print(f"\n  [OK] 未包含'第十章'")

    # 检查第十章第一个 chunk 的内容（应该包含"第十章"）
    if chapter10_chunks:
        first_chunk = chapter10_chunks[0]
        content = first_chunk.get("review_chunk_content", "")
        print(f"\n第十章第一个 chunk: {first_chunk.get('chunk_id')}")
        print(f"  section_label: {first_chunk.get('section_label')}")
        idx = content.find("第十章")
        if idx >= 0:
            print(f"  [OK] 包含'第十章'在位置 {idx}（这是正常的，是章节标题）")


if __name__ == "__main__":
    test_file = Path("D:/wx_work/sichuan_luqiao/lu_sgsc_testfile/测试模版-四川路桥专项施工方案框架以及编制说明（2025修订第三版）- v0.2.pdf")
    if test_file.exists():
        check_file(test_file)
    else:
        print(f"File not found: {test_file}")