Create sync_data.sh

sync_data.sh

@@ -0,0 +1,153 @@
+#!/bin/sh
+
+# 检查环境变量
+if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
+    echo "未检测到 HF_TOKEN 或 DATASET_ID，备份功能不可用"
+    exit 1
+fi
+
+# 生成同步脚本
+cat > hf_sync.py << 'EOL'
+# HuggingFace 同步脚本
+from huggingface_hub import HfApi
+import sys
+import os
+import tarfile
+import tempfile
+
+# 管理备份文件数量，超出最大数量则自动删除最旧的备份
+def manage_backups(api, repo_id, max_files=50):
+    files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
+    backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
+    backup_files.sort()
+    if len(backup_files) >= max_files:
+        files_to_delete = backup_files[:(len(backup_files) - max_files + 1)]
+        for file_to_delete in files_to_delete:
+            try:
+                api.delete_file(path_in_repo=file_to_delete, repo_id=repo_id, repo_type="dataset")
+                print(f'已删除旧备份: {file_to_delete}')
+            except Exception as e:
+                print(f'删除 {file_to_delete} 时出错: {str(e)}')
+
+# 上传备份文件到 HuggingFace
+def upload_backup(file_path, file_name, token, repo_id):
+    api = HfApi(token=token)
+    try:
+        api.upload_file(
+            path_or_fileobj=file_path,
+            path_in_repo=file_name,
+            repo_id=repo_id,
+            repo_type="dataset"
+        )
+        print(f"成功上传 {file_name}")
+        manage_backups(api, repo_id)
+    except Exception as e:
+        print(f"上传文件出错: {str(e)}")
+
+
+# 下载最新备份
+def download_latest_backup(token, repo_id, extract_path):
+    try:
+        api = HfApi(token=token)
+        files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
+        backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
+        if not backup_files:
+            print("未找到任何备份文件")
+            return
+        latest_backup = sorted(backup_files)[-1]
+        with tempfile.TemporaryDirectory() as temp_dir:
+            filepath = api.hf_hub_download(
+                repo_id=repo_id,
+                filename=latest_backup,
+                repo_type="dataset",
+                local_dir=temp_dir
+            )
+            if filepath and os.path.exists(filepath):
+                with tarfile.open(filepath, 'r:gz') as tar:
+                    tar.extractall(extract_path)
+                print(f"已成功恢复备份: {latest_backup}")
+    except Exception as e:
+        print(f"下载备份出错: {str(e)}")
+
+# 合并历史提交
+def super_squash_history(token, repo_id):
+    try:
+        api = HfApi(token=token)
+        api.super_squash_history(repo_id=repo_id, repo_type="dataset")
+        print("历史合并完成。")
+    except Exception as e:
+        print(f"合并历史出错: {str(e)}")
+
+# 主函数
+if __name__ == "__main__":
+    action = sys.argv[1]
+    token = sys.argv[2]
+    repo_id = sys.argv[3]
+    if action == "upload":
+        file_path = sys.argv[4]
+        file_name = sys.argv[5]
+        upload_backup(file_path, file_name, token, repo_id)
+    elif action == "download":
+        extract_path = sys.argv[4] if len(sys.argv) > 4 else '.'
+        download_latest_backup(token, repo_id, extract_path)
+    elif action == "super_squash":
+        super_squash_history(token, repo_id)
+EOL
+# 首次启动时从 HuggingFace 下载最新备份（解压到应用目录）
+echo "正在从 HuggingFace 下载最新备份..."
+python hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}" "/OpenList"  
+
+# 同步函数
+sync_data() {
+    while true; do
+        echo "同步进程启动于 $(date)"
+        
+        # 确保数据目录存在（请根据实际路径修改）
+        STORAGE_PATH="/OpenList/data"
+        if [ -d "${STORAGE_PATH}" ]; then
+            # 创建备份
+            timestamp=$(date +%Y%m%d_%H%M%S)
+            backup_file="backup_${timestamp}.tar.gz"
+            
+            # 压缩目录（使用-C避免包含父路径）
+            tar -czf "/tmp/${backup_file}" -C "$(dirname "${STORAGE_PATH}")" "$(basename "${STORAGE_PATH}")"
+            
+            # 上传到 HuggingFace
+            echo "正在上传备份到 HuggingFace..."
+            python hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "/tmp/${backup_file}" "${backup_file}"
+            
+            # 合并历史提交
+        SQUASH_FLAG_FILE="/tmp/last_squash_time"
+        NOW=$(date +%s)
+        SEVEN_DAYS=$((7*24*60*60))
+        if [ ! -f "$SQUASH_FLAG_FILE" ]; then
+            echo $NOW > "$SQUASH_FLAG_FILE"
+            echo "首次合并历史提交..."
+            python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
+        else
+            LAST=$(cat "$SQUASH_FLAG_FILE")
+            DIFF=$((NOW - LAST))
+            if [ $DIFF -ge $SEVEN_DAYS ]; then
+                echo $NOW > "$SQUASH_FLAG_FILE"
+                echo "距离上次合并已超过7天，正在合并历史提交..."
+                python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
+            else
+                echo "距离上次合并未满7天，本次跳过合并历史提交。"
+            fi
+        fi
+
+            # 清理临时文件
+            rm -f "/tmp/${backup_file}"
+        else
+            echo "存储目录 ${STORAGE_PATH} 不存在，等待中..."
+        fi
+        
+        # 同步间隔
+        SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
+        echo "下次同步将在 ${SYNC_INTERVAL} 秒后进行..."
+        sleep $SYNC_INTERVAL
+    done
+}
+
+# 启动同步进程
+sync_data &