RagFlow_Plugs/test/verify_pdf_splitter.py

"""PDF拆分功能验证脚本"""

import sys
import os
from pathlib import Path

sys.path.append(os.path.dirname(os.path.abspath(__file__)))

def check_pymupdf():
    """检查PyMuPDF是否安装"""
    print("检查PyMuPDF是否安装...")
    try:
        import fitz
        print(f"✓ PyMuPDF已安装")
        print(f"  版本: {fitz.__version__}")
        return True
    except ImportError:
        print("✗ 未安装PyMuPDF，请运行: pip install PyMuPDF")
        return False
    except Exception as e:
        print(f"✗ 检查PyMuPDF时出错: {str(e)}")
        return False

def check_pdf_file(pdf_path):
    """检查PDF文件是否存在"""
    print(f"检查PDF文件: {pdf_path}")
    pdf = Path(pdf_path)
    if pdf.exists():
        print(f"✓ PDF文件存在，大小: {pdf.stat().st_size} 字节")
        return True
    else:
        print(f"✗ PDF文件不存在: {pdf_path}")
        return False

def test_pdf_splitter():
    """测试PDF拆分功能"""
    print("=" * 50)
    print("PDF拆分功能验证")
    print("=" * 50)
    
    # 检查PyMuPDF
    pymupdf_ok = check_pymupdf()
    print()
    
    # 检查示例PDF文件
    pdf_path = r"D:\project\work\ragflow_plugs\book\不一样的卡梅拉1-我想去看海.pdf"
    pdf_ok = check_pdf_file(pdf_path)
    print()
    
    if not pymupdf_ok:
        print("=" * 50)
        print("PyMuPDF安装指南:")
        print("1. 运行命令安装: pip install PyMuPDF")
        print("2. 安装完成后重试")
        print("=" * 50)
        return False
    
    if not pdf_ok:
        print("=" * 50)
        print(r"请确保PDF文件存在: D:\project\work\ragflow_plugs\book\不一样的卡梅拉1-我想去看海.pdf")
        print("或修改脚本中的pdf_path变量为实际的PDF文件路径")
        print("=" * 50)
        return False
    
    # 测试PDF拆分功能
    print("开始测试PDF拆分功能...")
    try:
        from services.pdf_parser.pdf_splitter import PDFSplitter
        
        splitter = PDFSplitter()
        print(f"正在拆分PDF: {pdf_path}")
        pages = splitter.split_pdf(pdf_path)
        
        print(f"✓ PDF拆分成功，共 {len(pages)} 页")
        for page in pages[:3]:  # 只显示前3页
            print(f"  - 页码: {page['page_number']}, 图像大小: {page['image'].size}")
        
        if len(pages) > 3:
            print(f"  ... 以及 {len(pages) - 3} 页")
        
        return True
    except Exception as e:
        print(f"✗ PDF拆分失败: {str(e)}")
        print("可能的解决方案:")
        print("1. 确保poppler已正确安装并在PATH中")
        print("2. 检查PDF文件是否损坏")
        print("3. 检查pdf2image库版本是否兼容")
        return False

def main():
    """主函数"""
    test_pdf_splitter()
    
    print("\n" + "=" * 50)
    print("验证完成")
    print("=" * 50)

if __name__ == "__main__":
    main()