提交文件

extract_all_doc.py

@@ -0,0 +1,198 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+批量提取参考文档目录中所有 .doc 文件的内容
+输出为 JSON 格式，方便后续处理
+"""
+
+import os
+import sys
+import json
+import olefile
+import re
+from pathlib import Path
+from datetime import datetime
+
+
+def extract_text_from_doc(filepath):
+    """从 .doc 文件提取文本内容"""
+    try:
+        ole = olefile.OleFileIO(filepath)
+        
+        if not ole.exists('WordDocument'):
+            return None, "不是有效的Word文档"
+        
+        # 读取WordDocument流
+        word_stream = ole.openstream('WordDocument')
+        data = word_stream.read()
+        
+        # 尝试读取表格流
+        table_name = '1Table' if ole.exists('1Table') else '0Table'
+        table_data = b''
+        if ole.exists(table_name):
+            table_stream = ole.openstream(table_name)
+            table_data = table_stream.read()
+        
+        # 提取文本
+        text_parts = []
+        
+        # 从WordDocument流提取
+        try:
+            decoded = data.decode('utf-16-le', errors='ignore')
+            # 提取中文和英文文本
+            matches = re.findall(r'[\u4e00-\u9fff\u0020-\u007e\n\r\t\.,;:!?()（）。，；：！？、""''【】\[\]｛｝{}<>《》\-+×÷=±≡≈≠≤≥∞∩∪∈∉⊂⊃⊆⊇∠⊥∥→←↑↓↔\d]+', decoded)
+            for m in matches:
+                if len(m.strip()) > 1:
+                    text_parts.append(m.strip())
+        except:
+            pass
+        
+        # 从表格流提取
+        try:
+            decoded = table_data.decode('utf-16-le', errors='ignore')
+            matches = re.findall(r'[\u4e00-\u9fff\u0020-\u007e\n\r\t\.,;:!?()（）。，；：！？、""''【】\[\]｛｝{}<>《》\-+×÷=±≡≈≠≤≥∞∩∪∈∉⊂⊃⊆⊇∠⊥∥→←↑↓↔\d]+', decoded)
+            for m in matches:
+                if len(m.strip()) > 1:
+                    text_parts.append(m.strip())
+        except:
+            pass
+        
+        # 简单提取（从整个文件）
+        with open(filepath, 'rb') as f:
+            raw_data = f.read()
+        
+        try:
+            decoded = raw_data.decode('utf-16-le', errors='ignore')
+            matches = re.findall(r'[\u4e00-\u9fff\u0020-\u007e\n\r\t\.,;:!?()（）。，；：！？、""''【】\[\]｛｝{}<>《》\-+×÷=±≡≈≠≤≥∞∩∪∈∉⊂⊃⊆⊇∠⊥∥→←↑↓↔\d]+', decoded)
+            for m in matches:
+                if len(m.strip()) > 1:
+                    text_parts.append(m.strip())
+        except:
+            pass
+        
+        # 去重
+        seen = set()
+        unique_parts = []
+        for part in text_parts:
+            if part not in seen and len(part) > 1:
+                seen.add(part)
+                unique_parts.append(part)
+        
+        # 合并文本
+        full_text = '\n'.join(unique_parts)
+        
+        ole.close()
+        return full_text, None
+        
+    except Exception as e:
+        return None, str(e)
+
+
+def extract_tables_from_doc(filepath):
+    """尝试提取表格内容"""
+    try:
+        ole = olefile.OleFileIO(filepath)
+        
+        tables = []
+        
+        # 表格信息存储在表格流中
+        # 这是一个简化的方法，实际需要解析复杂的二进制结构
+        
+        # 从整个文件中寻找表格特征
+        with open(filepath, 'rb') as f:
+            data = f.read()
+        
+        # 尝试提取表格单元格内容
+        decoded = data.decode('utf-16-le', errors='ignore')
+        
+        # 寻找表格模式（连续的短文本行）
+        lines = [l.strip() for l in decoded.split('\n') if l.strip()]
+        
+        # 检测可能的表格行
+        table_rows = []
+        for line in lines:
+            if len(line) > 2 and len(line) < 200:  # 表格单元格通常是短文本
+                # 检查是否是表格分隔符
+                if not re.match(r'^[\s\-\|]+$', line):
+                    table_rows.append(line)
+        
+        ole.close()
+        return table_rows
+        
+    except Exception as e:
+        return []
+
+
+def process_all_doc_files(directory):
+    """处理目录下所有 .doc 文件"""
+    results = {}
+    
+    doc_files = []
+    for root, dirs, files in os.walk(directory):
+        for file in files:
+            if file.lower().endswith('.doc') and not file.lower().endswith('.docx'):
+                doc_files.append((file, os.path.join(root, file)))
+    
+    print(f"发现 {len(doc_files)} 个 .doc 文件")
+    
+    for i, (filename, filepath) in enumerate(doc_files):
+        print(f"处理 [{i+1}/{len(doc_files)}]: {filename}")
+        
+        text, error = extract_text_from_doc(filepath)
+        tables = extract_tables_from_doc(filepath)
+        
+        results[filename] = {
+            'filepath': filepath,
+            'text': text if text else '',
+            'tables': tables,
+            'error': error
+        }
+    
+    return results
+
+
+def main():
+    """主函数"""
+    # 默认处理参考文档目录
+    doc_dir = r"D:\医院绩效系统\参考文档"
+    
+    if len(sys.argv) > 1:
+        doc_dir = sys.argv[1]
+    
+    print(f"开始处理目录: {doc_dir}")
+    print("=" * 60)
+    
+    # 提取所有 .doc 文件
+    results = process_all_doc_files(doc_dir)
+    
+    # 保存结果
+    output_file = r"D:\医院绩效系统\doc_extracted_content.json"
+    
+    with open(output_file, 'w', encoding='utf-8') as f:
+        json.dump(results, f, ensure_ascii=False, indent=2)
+    
+    print(f"\n结果已保存到: {output_file}")
+    
+    # 统计
+    success_count = sum(1 for v in results.values() if v.get('text'))
+    error_count = sum(1 for v in results.values() if v.get('error'))
+    
+    print(f"成功提取: {success_count} 个文件")
+    print(f"提取失败: {error_count} 个文件")
+    
+    # 显示部分结果
+    print("\n" + "=" * 60)
+    print("部分提取结果预览:")
+    print("=" * 60)
+    
+    for filename, data in list(results.items())[:3]:
+        print(f"\n【{filename}】")
+        text = data.get('text', '')
+        if text:
+            print(text[:500] + "..." if len(text) > 500 else text)
+        else:
+            print(f"  错误: {data.get('error', '未知错误')}")
+
+
+if __name__ == '__main__':
+    main()