100d1a06eb
- 后端 API（Flask + Gunicorn） - RAG 引擎（混合检索 + 云端 Reranker + 引用溯源） - 文档解析（MinerU + 多格式支持） - Docker 生产部署配置 - 排除前端项目、敏感配置、模型文件
98 lines
2.9 KiB
Python
98 lines
2.9 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""
|
|
阶段 1 验证脚本：检查向量库中的 metadata 是否包含图片信息
|
|
|
|
运行方式：
|
|
.\venv\Scripts\python.exe scripts/test_image_metadata.py
|
|
"""
|
|
|
|
import sys
|
|
import os
|
|
|
|
# 添加项目根目录到路径
|
|
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
|
|
|
|
from knowledge.manager import get_kb_manager
|
|
|
|
|
|
def test_image_metadata():
|
|
"""检查向量库中的 metadata 是否包含图片信息"""
|
|
print("=" * 60)
|
|
print("阶段 1 验证：检查 metadata 中的图片信息")
|
|
print("=" * 60)
|
|
|
|
kb = get_kb_manager()
|
|
coll = kb.get_collection('public_kb')
|
|
|
|
if not coll:
|
|
print("❌ public_kb 不存在")
|
|
return
|
|
|
|
# 获取所有切片
|
|
result = coll.get(include=['metadatas', 'documents'])
|
|
|
|
if not result or not result.get('ids'):
|
|
print("❌ 向量库为空，请先上传一个包含图片的 PDF")
|
|
return
|
|
|
|
total = len(result['ids'])
|
|
print(f"\n📊 向量库总切片数: {total}")
|
|
|
|
# 统计图片相关字段
|
|
has_images_json = 0
|
|
has_image_path = 0
|
|
image_chunks = 0
|
|
table_chunks = 0
|
|
|
|
print("\n📋 切片详情（前 20 个）:")
|
|
print("-" * 60)
|
|
|
|
for i, (chunk_id, meta, doc) in enumerate(zip(result['ids'], result['metadatas'], result['documents'])):
|
|
chunk_type = meta.get('chunk_type', 'text')
|
|
images_json = meta.get('images_json')
|
|
image_path = meta.get('image_path')
|
|
|
|
if images_json:
|
|
has_images_json += 1
|
|
if image_path:
|
|
has_image_path += 1
|
|
if chunk_type in ('image', 'chart'):
|
|
image_chunks += 1
|
|
if chunk_type == 'table':
|
|
table_chunks += 1
|
|
|
|
# 打印前 20 个切片的详情
|
|
if i < 20:
|
|
print(f"[{i+1}] chunk_type: {chunk_type}")
|
|
if images_json:
|
|
print(f" images_json: {images_json[:100]}...")
|
|
if image_path:
|
|
print(f" image_path: {image_path}")
|
|
print()
|
|
|
|
print("-" * 60)
|
|
print("\n📈 统计结果:")
|
|
print(f" - 包含 images_json 的切片: {has_images_json}")
|
|
print(f" - 包含 image_path 的切片: {has_image_path}")
|
|
print(f" - 图片类型切片 (image/chart): {image_chunks}")
|
|
print(f" - 表格类型切片: {table_chunks}")
|
|
|
|
# 验证结果
|
|
print("\n✅ 验证结果:")
|
|
if has_images_json > 0:
|
|
print(f" ✅ images_json 字段已正确写入 ({has_images_json} 个切片)")
|
|
else:
|
|
print(" ⚠️ images_json 字段未找到，可能需要重新上传 PDF")
|
|
|
|
if has_image_path > 0:
|
|
print(f" ✅ image_path 字段已正确写入 ({has_image_path} 个切片)")
|
|
else:
|
|
print(" ⚠️ image_path 字段未找到，可能没有独立图片切片")
|
|
|
|
|
|
if __name__ == "__main__":
|
|
if sys.platform == 'win32':
|
|
sys.stdout.reconfigure(encoding='utf-8')
|
|
|
|
test_image_metadata()
|