初次提交

backend/simple_files_check.py

@@ -0,0 +1,168 @@
+#!/usr/bin/env python3
+"""
+简化版文件存储检查脚本
+"""
+
+import mysql.connector
+import os
+import hashlib
+
+def check_files_storage():
+    """检查文件存储情况"""
+
+    print("=== 文件存储情况检查 ===")
+
+    try:
+        # 连接数据库
+        conn = mysql.connector.connect(
+            host="101.126.85.76",
+            user="mytest_db",
+            password="mytest_db",
+            database="mytest_db"
+        )
+        cursor = conn.cursor()
+
+        # 查询所有文件
+        cursor.execute("""
+            SELECT id, user_id, original_filename, filename, file_path, file_size,
+                   file_hash, mime_type, created_at
+            FROM files
+            ORDER BY created_at DESC
+        """)
+
+        db_files = cursor.fetchall()
+        print(f"数据库中的文件记录数: {len(db_files)}")
+        print()
+
+        print("=== 数据库中的文件记录 ===")
+        for file in db_files:
+            (id, user_id, original_filename, filename, file_path,
+             file_size, file_hash, mime_type, created_at) = file
+
+            print(f"ID: {id}")
+            print(f"  原始文件名: {original_filename}")
+            print(f"  存储文件名: {filename}")
+            print(f"  文件大小: {file_size} bytes")
+            print(f"  MIME类型: {mime_type}")
+            print(f"  创建时间: {created_at}")
+            print("-" * 40)
+
+        # 检查实际文件存在情况
+        print("\n=== 文件存在性检查 ===")
+        existing_count = 0
+        for file in db_files:
+            (id, user_id, original_filename, filename, file_path,
+             file_size, file_hash, mime_type, created_at) = file
+
+            full_path = os.path.join("uploads", filename)
+            if os.path.exists(full_path):
+                existing_count += 1
+                print(f"[存在] ID {id}: {original_filename}")
+            else:
+                print(f"[缺失] ID {id}: {original_filename}")
+
+        print(f"\n实际存在的文件数: {existing_count}")
+        print(f"缺失的文件数: {len(db_files) - existing_count}")
+
+        # 检查uploads目录详情
+        print("\n=== uploads目录详情 ===")
+        uploads_dir = "uploads"
+        if os.path.exists(uploads_dir):
+            files = os.listdir(uploads_dir)
+            print(f"uploads目录中的文件数: {len(files)}")
+
+            for file in files:
+                file_path = os.path.join(uploads_dir, file)
+                file_size = os.path.getsize(file_path)
+                print(f"文件: {file}, 大小: {file_size} bytes")
+        else:
+            print("uploads目录不存在")
+
+    except Exception as e:
+        print(f"检查出错: {e}")
+    finally:
+        if 'conn' in locals() and conn.is_connected():
+            cursor.close()
+            conn.close()
+
+def check_file_integrity():
+    """检查文件完整性"""
+
+    print("\n=== 文件完整性检查 ===")
+
+    try:
+        conn = mysql.connector.connect(
+            host="101.126.85.76",
+            user="mytest_db",
+            password="mytest_db",
+            database="mytest_db"
+        )
+        cursor = conn.cursor()
+
+        cursor.execute("SELECT id, filename, file_hash, file_size FROM files")
+        db_files = cursor.fetchall()
+
+        integrity_ok = True
+
+        for (id, filename, expected_hash, expected_size) in db_files:
+            full_path = os.path.join("uploads", filename)
+
+            if os.path.exists(full_path):
+                actual_size = os.path.getsize(full_path)
+                if actual_size != expected_size:
+                    print(f"ID {id}: 文件大小不匹配 (期望: {expected_size}, 实际: {actual_size})")
+                    integrity_ok = False
+                    continue
+
+                try:
+                    with open(full_path, 'rb') as f:
+                        content = f.read()
+                        actual_hash = hashlib.sha256(content).hexdigest()
+
+                    if actual_hash != expected_hash:
+                        print(f"ID {id}: 文件哈希不匹配")
+                        print(f"  期望: {expected_hash}")
+                        print(f"  实际: {actual_hash}")
+                        integrity_ok = False
+                    else:
+                        print(f"ID {id}: 完整性检查通过")
+                except Exception as e:
+                    print(f"ID {id}: 无法计算哈希 - {e}")
+                    integrity_ok = False
+            else:
+                print(f"ID {id}: 文件不存在")
+                integrity_ok = False
+
+        if integrity_ok:
+            print("所有文件完整性检查通过!")
+        else:
+            print("发现文件完整性问题!")
+
+    except Exception as e:
+        print(f"完整性检查出错: {e}")
+    finally:
+        if 'conn' in locals() and conn.is_connected():
+            cursor.close()
+            conn.close()
+
+if __name__ == "__main__":
+    check_files_storage()
+    check_file_integrity()
+
+    print("\n=== 文件存储架构总结 ===")
+    print("1. 数据库(files表): 存储文件元数据")
+    print("   - 文件ID、用户ID、原始文件名")
+    print("   - 存储文件名(UUID格式)")
+    print("   - 文件大小、MIME类型")
+    print("   - SHA-256哈希值")
+    print("   - 创建时间等")
+    print()
+    print("2. 文件系统(uploads目录): 存储实际文件")
+    print("   - 文件使用UUID命名确保唯一性")
+    print("   - 文件内容与数据库记录一一对应")
+    print("   - 通过file_hash验证完整性")
+    print()
+    print("3. 回答您的问题:")
+    print("   是的，数据库中存储的是文件元数据，")
+    print("   实际文件内容存储在服务器的uploads目录中。")
+    print("   两者通过file_hash保持关联和完整性验证。")