LQAgentPlatform/utils_test/Chunk_Split_Test/test_real_scenario.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
模拟实际工作流场景测试
使用 DocumentWorkflow 完整流程测试
"""

import sys
import os
import asyncio
from pathlib import Path

project_root = Path(__file__).parent.parent.parent
os.chdir(project_root)
sys.path.insert(0, str(project_root))

from core.construction_review.workflows.document_workflow import DocumentWorkflow
from core.base.task_models import TaskFileInfo

TARGET_FILE = Path(__file__).parent / "330测试fa56f7a8-bd36-4140-8cb1-f9f973ce8745.pdf"

async def test_document_workflow():
    """测试完整文档工作流"""

    if not TARGET_FILE.exists():
        print(f"[ERROR] 文件不存在: {TARGET_FILE}")
        return 1

    # 读取文件内容
    with open(TARGET_FILE, "rb") as f:
        file_content = f.read()

    # 创建 TaskFileInfo
    task_info = TaskFileInfo({
        "file_id": "test_330_file",
        "user_id": "test_user",
        "callback_task_id": "test_callback_001",
        "file_name": TARGET_FILE.name,
        "file_type": "pdf",
        "file_content": file_content,
        "review_config": [],
        "review_item_config": [],
        "project_plan_type": "T梁运输及安装专项施工方案",
        "tendency_review_role": "",
        "launched_at": 0
    })

    # 创建工作流
    workflow = DocumentWorkflow(
        task_file_info=task_info,
        progress_manager=None,  # 不依赖进度管理器
        redis_duplicate_checker=None
    )

    print(f"\n[INFO] 开始处理文件: {TARGET_FILE.name}")
    print(f"       文件大小: {len(file_content)} bytes")

    try:
        result = await workflow.execute(
            file_content=file_content,
            file_type="pdf"
        )

        print("\n" + "=" * 80)
        print("文档处理结果")
        print("=" * 80)
        print(f"  file_id: {result.get('file_id')}")
        print(f"  document_name: {result.get('document_name')}")
        print(f"  total_chunks: {result.get('total_chunks')}")

        structured = result.get('structured_content', {})
        chunks = structured.get('chunks', [])

        # 分析章节分布
        print("\n" + "=" * 80)
        print("章节分布分析")
        print("=" * 80)

        first_levels = set()
        for c in chunks:
            label = c.get('section_label', '') or c.get('title', '')
            if '->' in label:
                first = label.split('->')[0].strip()
            else:
                first = label.strip()
            first_levels.add(first)

        print(f"  共 {len(first_levels)} 个一级章节:")
        for fl in sorted(first_levels):
            # 计算该章节下的chunks数量
            count = sum(1 for c in chunks if (c.get('section_label', '') or c.get('title', '')).startswith(fl))
            print(f"    - {fl}: {count} chunks")

        # 检查是否有缺失的章节
        expected_chapters = ["第一章", "第二章", "第三章", "第四章", "第五章", "第六章", "第七章", "第八章", "第九章", "第十章"]
        found_chapters = []
        missing_chapters = []

        for ec in expected_chapters:
            found = any(ec in fl for fl in first_levels)
            if found:
                found_chapters.append(ec)
            else:
                missing_chapters.append(ec)

        print(f"\n  找到的章节: {found_chapters}")
        if missing_chapters:
            print(f"  [WARNING] 缺失的章节: {missing_chapters}")

        # 详细信息
        print("\n" + "=" * 80)
        print("所有 Chunks 详情")
        print("=" * 80)
        for i, chunk in enumerate(chunks, 1):
            label = chunk.get('section_label', '') or chunk.get('title', '')
            content = chunk.get('content', '') or chunk.get('review_chunk_content', '')
            print(f"\n  {i}. {label}")
            print(f"     chunk_id: {chunk.get('chunk_id')}")
            print(f"     page: {chunk.get('page')}")
            print(f"     content_length: {len(content)}")
            if content:
                preview = content[:80].replace('\n', ' ')
                print(f"     preview: {preview}...")

        return 0

    except Exception as e:
        print(f"\n[ERROR] 处理失败: {e}")
        import traceback
        traceback.print_exc()
        return 1

if __name__ == "__main__":
    result = asyncio.run(test_document_workflow())
    sys.exit(result)