Upload Dockerfile

Dockerfile

@@ -72,7 +72,6 @@ repo_id = os.getenv("HF_DATASET")
 token = os.getenv("HF_TOKEN")
 
 OPENCLAW_LOCAL  = "/root/.openclaw"
-SCATTER_REMOTE_PREFIX = "openclaw/"   # 散裝文件在 Dataset 中的前綴
 
 # ── 通用工具函數 ──────────────────────────────────────────────────────────────
 
@@ -121,12 +120,26 @@ def restore():
     if not repo_id or not token:
         print("Skip Restore: HF_DATASET or HF_TOKEN not set")
         return
+
+    # RESTORE_SKIP=all：不恢復任何 tar.gz 內文件到本地，直接返回
+    restore_skip_raw = os.getenv("RESTORE_SKIP", "").strip()
+    if restore_skip_raw == "all":
+        print("Restore skip: RESTORE_SKIP=all, skipping all restore.")
+        return
+
+    # 解析跳過列表（用於在解壓時逐條目跳過，不是解壓後刪除）
+    skip_set = _parse_skip_list("RESTORE_SKIP")
+
     try:
         all_files = list(api.list_repo_files(repo_id=repo_id, repo_type="dataset", token=token))
 
-        # ── 1. 恢復 tar.gz 備份 ──────────────────────────────────────────────
+        # ── 從 tar.gz 備份恢復 ───────────────────────────────────────────────
         # 備份範圍：/root/.openclaw 下所有文件及文件夾（保持完整目錄結構）
         # tar 包內 arcname 直接對應 .openclaw/ 下的相對路徑，解壓目標為 /root/.openclaw/
+        # RESTORE_SKIP：逗號分隔，填寫相對於 /root/.openclaw 的路徑
+        # 支持多級子目錄，例如：RESTORE_SKIP=agents/main/sessions,cron,extensions/foo/bar
+        # 留空或不設置 = 全部恢復（默認行為）
+        # RESTORE_SKIP=all = 跳過全部恢復（已在上方提前返回）
         now = datetime.now()
         for i in range(5):
             day = (now - timedelta(days=i)).strftime("%Y-%m-%d")
@@ -136,113 +149,18 @@ def restore():
                 path = hf_hub_download(repo_id=repo_id, filename=name, repo_type="dataset", token=token)
                 os.makedirs(OPENCLAW_LOCAL, exist_ok=True)
                 with tarfile.open(path, "r:gz") as tar:
-                    tar.extractall(path=OPENCLAW_LOCAL)
+                    for member in tar.getmembers():
+                        if _is_skipped(member.name, skip_set):
+                            print(f"Restore skip (RESTORE_SKIP): skipping {member.name}")
+                            continue
+                        tar.extract(member, path=OPENCLAW_LOCAL)
                 print(f"Success: Restored from {name}")
-
-                # ── RESTORE_SKIP：解壓後刪除指定條目，讓後續腳本用默認值重新生成 ──
-                # 格式：逗號分隔，填寫相對於 /root/.openclaw 的路徑
-                # 支持多級子目錄，例如：RESTORE_SKIP=agents/main/sessions,cron,extensions/foo/bar
-                # 若路徑不存在則忽略，不影響腳本繼續運行
-                # 留空或不設置 = 全部恢復（默認行為）
-                # 注意：用完後清空此變量，否則每次重啟都會跳過恢復
-                skip_set = _parse_skip_list("RESTORE_SKIP")
-                for entry in skip_set:
-                    target = os.path.join(OPENCLAW_LOCAL, entry)
-                    if os.path.isdir(target):
-                        shutil.rmtree(target)
-                        print(f"Restore skip (RESTORE_SKIP): removed dir  {entry}")
-                    elif os.path.isfile(target):
-                        os.remove(target)
-                        print(f"Restore skip (RESTORE_SKIP): removed file {entry}")
-                    else:
-                        print(f"Restore skip (RESTORE_SKIP): not found, skipped {entry}")
                 break
 
-        # ── 2. 恢復散裝文件 ──────────────────────────────────────────────────
-        # 備份範圍：/root/.openclaw 下所有文件及文件夾（保持完整目錄結構）
-        # Dataset 中存儲路徑：openclaw/<相對於 .openclaw 的路徑>
-        scatter_files = [f for f in all_files if f.startswith(SCATTER_REMOTE_PREFIX)]
-        if scatter_files:
-            os.makedirs(OPENCLAW_LOCAL, exist_ok=True)
-            for remote_path in scatter_files:
-                rel = remote_path[len(SCATTER_REMOTE_PREFIX):]
-                if not rel:
-                    continue
-                local_path = os.path.join(OPENCLAW_LOCAL, rel)
-                print(f"Restoring file: {rel}")
-                dl = hf_hub_download(repo_id=repo_id, filename=remote_path, repo_type="dataset", token=token)
-                os.makedirs(os.path.dirname(local_path), exist_ok=True)
-                shutil.copy2(dl, local_path)
-            print(f"Scatter restore: {len(scatter_files)} file(s) restored.")
-        else:
-            print("No scatter files found in Dataset, skipping scatter restore.")
-
     except Exception as e:
         print(f"Restore Error: {e}")
 
-# ── restore_workspace() ───────────────────────────────────────────────────────
 
-def restore_workspace():
-    """
-    從 HuggingFace Dataset 的 openclaw/ 前綴下拉取最新文件到本地 /root/.openclaw/。
-    與 restore() 的區別：
-      - 不處理 tar.gz 備份
-      - 專門用於"用戶在 Dataset 網頁上編輯了文件後，讓容器立刻感知更新"的場景
-      - 由後台定時循環每 30 分鐘自動調用一次
-      - 也可以通過釘釘發指令手動觸發：python3 /usr/local/bin/sync.py restore_workspace
-    衝突保護邏輯：
-      - Dataset 上的版本始終優先（網頁手動編輯的版本會覆蓋本地）
-      - 這與 backup() 的邏輯相反：backup() 是本地新則上傳，restore_workspace() 是遠端新則下載
-      - 若本地文件比遠端更新（本地有未備份的修改），跳過下載，保留本地版本
-    """
-    if not repo_id or not token:
-        print("Skip restore_workspace: HF_DATASET or HF_TOKEN not set")
-        return
-    try:
-        all_files = list(api.list_repo_files(
-            repo_id=repo_id, repo_type="dataset", token=token
-        ))
-        scatter_files = [f for f in all_files if f.startswith(SCATTER_REMOTE_PREFIX)]
-        if not scatter_files:
-            print("restore_workspace: No scatter files found in Dataset, skipping.")
-            return
-        os.makedirs(OPENCLAW_LOCAL, exist_ok=True)
-        updated = 0
-        skipped = 0
-        for remote_path in scatter_files:
-            rel = remote_path[len(SCATTER_REMOTE_PREFIX):]
-            if not rel:
-                continue
-            local_path = os.path.join(OPENCLAW_LOCAL, rel)
-            try:
-                remote_info = api.list_repo_tree(
-                    repo_id=repo_id, repo_type="dataset", token=token,
-                    path_in_repo=SCATTER_REMOTE_PREFIX.rstrip("/"), recursive=True
-                )
-                remote_mtime = None
-                for item in remote_info:
-                    if hasattr(item, "path") and item.path == remote_path:
-                        if item.last_commit and item.last_commit.date:
-                            remote_mtime = item.last_commit.date.timestamp()
-                        break
-                if remote_mtime and os.path.exists(local_path):
-                    local_mtime = os.path.getmtime(local_path)
-                    if local_mtime >= remote_mtime:
-                        skipped += 1
-                        continue
-            except Exception:
-                pass
-            dl = hf_hub_download(
-                repo_id=repo_id, filename=remote_path,
-                repo_type="dataset", token=token
-            )
-            os.makedirs(os.path.dirname(local_path), exist_ok=True)
-            shutil.copy2(dl, local_path)
-            print(f"restore_workspace: updated {rel}")
-            updated += 1
-        print(f"restore_workspace: {updated} updated, {skipped} already up-to-date.")
-    except Exception as e:
-        print(f"restore_workspace Error: {e}")
 
 # ── backup() ──────────────────────────────────────────────────────────────────
 
@@ -251,7 +169,7 @@ def backup():
         print("Skip Backup: HF_DATASET or HF_TOKEN not set")
         return
 
-    # ── 1. tar.gz 打包備份 ───────────────────────────────────────────────────
+    # ── tar.gz 打包備份 ──────────────────────────────────────────────────────
     # 備份範圍：/root/.openclaw 下所有文件及文件夾（完整目錄結構）
     # tar 包解壓目標為 /root/.openclaw/，arcname 對應其內部相對路徑
     #
@@ -259,6 +177,11 @@ def backup():
     # 格式：逗號分隔，支持多級子目錄
     # 例如：BACKUP_TAR_SKIP=agents/main/sessions,cron,extensions/foo/bar
     # 留空或不設置 = 備份全部
+    # BACKUP_TAR_SKIP=all = 不執行 tar.gz 備份，不生成任何 tar.gz 備份文件
+    backup_tar_skip_raw = os.getenv("BACKUP_TAR_SKIP", "").strip()
+    if backup_tar_skip_raw == "all":
+        print("Backup tar.gz skip: BACKUP_TAR_SKIP=all, skipping tar.gz backup.")
+        return
     try:
         tar_skip_set = _parse_skip_list("BACKUP_TAR_SKIP")
         day = datetime.now().strftime("%Y-%m-%d")
@@ -272,76 +195,9 @@ def backup():
     except Exception as e:
         print(f"Backup tar.gz Error: {e}")
 
-    # ── 2. 散裝文件備份（逐個上傳，跳過 Dataset 上更新的文件）────────────────
-    # 備份範圍：/root/.openclaw 下所有文件及文件夾（保持完整目錄結構）
-    # Dataset 中存儲路徑：openclaw/<相對於 .openclaw 的路徑>
-    #
-    # BACKUP_SCATTER_SKIP：散裝備份時跳過的文件/目錄（相對於 /root/.openclaw）
-    # 格式：逗號分隔，支持多級子目錄
-    # 例如：BACKUP_SCATTER_SKIP=agents/main/sessions,cron,workspace/tmp
-    # 留空或不設置 = 備份全部
-    #
-    # 核心改動：原來逐個 upload_file 每個文件產生1個commit，
-    # 很快觸發 HF 免費帳號 128 commits/小時限制。
-    # 改為：把需要上傳的文件複製到臨時目錄，再用 upload_folder 一次性
-    # 合成單個 commit 上傳，徹底解決 429 rate limit 問題。
-    try:
-        scatter_skip_set = _parse_skip_list("BACKUP_SCATTER_SKIP")
-
-        # 獲取 Dataset 上已有文件的最後修改時間（用於跳過 Dataset 更新的文件）
-        remote_mtimes = {}
-        try:
-            repo_info = api.list_repo_tree(
-                repo_id=repo_id, repo_type="dataset", token=token,
-                path_in_repo=SCATTER_REMOTE_PREFIX.rstrip("/"), recursive=True
-            )
-            for item in repo_info:
-                if hasattr(item, "path") and hasattr(item, "last_commit"):
-                    rel = item.path[len(SCATTER_REMOTE_PREFIX):]
-                    if item.last_commit and item.last_commit.date:
-                        remote_mtimes[rel] = item.last_commit.date.timestamp()
-        except Exception:
-            pass
-
-        # 把需要上傳的文件複製到臨時目錄，保持目錄結構，再一次性 upload_folder
-        import tempfile
-        with tempfile.TemporaryDirectory() as tmpdir:
-            uploaded = 0
-            skipped = 0
-            for abs_path, rel_to_base in _walk_local(OPENCLAW_LOCAL, skip_set=scatter_skip_set):
-                local_mtime = os.path.getmtime(abs_path)
-                remote_mtime = remote_mtimes.get(rel_to_base)
-                if remote_mtime and remote_mtime > local_mtime:
-                    print(f"Scatter skip (Dataset newer): {rel_to_base}")
-                    skipped += 1
-                    continue
-                dst = os.path.join(tmpdir, rel_to_base)
-                os.makedirs(os.path.dirname(dst), exist_ok=True)
-                shutil.copy2(abs_path, dst)
-                uploaded += 1
-
-            if uploaded > 0:
-                # upload_folder 將 tmpdir 下所有文件合成單個 commit 上傳
-                # path_in_repo 指定遠端存放前綴（openclaw/）
-                api.upload_folder(
-                    folder_path=tmpdir,
-                    path_in_repo=SCATTER_REMOTE_PREFIX.rstrip("/"),
-                    repo_id=repo_id,
-                    repo_type="dataset",
-                    token=token,
-                    commit_message=f"Scatter backup {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
-                )
-                print(f"Scatter backup: {uploaded} file(s) uploaded in single commit, {skipped} skipped (Dataset newer).")
-            else:
-                print(f"Scatter backup: nothing to upload ({skipped} skipped, Dataset newer).")
-    except Exception as e:
-        print(f"Scatter Backup Error: {e}")
-
 if __name__ == "__main__":
     if len(sys.argv) > 1 and sys.argv[1] == "backup":
         backup()
-    elif len(sys.argv) > 1 and sys.argv[1] == "restore_workspace":
-        restore_workspace()
     else:
         restore()
 EOF
@@ -402,10 +258,26 @@ RUN cat <<'EOF' > /usr/local/bin/start-openclaw
 #!/bin/bash
 set -e
 
-mkdir -p /root/.openclaw/sessions
+# ── restore() 觸發條件判斷 ──────────────────────────────────────────────────
+# HF Space 磁盤是 ephemeral 的，每次容器重啟（首次部署、factory rebuild、普通 restart）
+# /root/.openclaw 都從空狀態開始。平台本身無法區分三種重啟場景。
+# 注意：start-openclaw-code-server 在調用本腳本之前會執行
+#   mkdir -p /root/.openclaw/logs
+# 因此不能以目錄是否存在或為空作為判定依據。
+#
+# 改用 openclaw.json 作為判定標誌：
+#   - 該文件由本腳本在 restore() 之後的配置寫入步驟生成
+#   - mkdir -p /root/.openclaw/logs 不會創建它
+#   - 首次部署 / 任何形式的容器重建：文件不存在 → 執行 restore()
+#   - 運行中再次調用本腳本（文件已存在）           → 跳過 restore()
+if [ ! -f /root/.openclaw/openclaw.json ]; then
+    echo "=== First deploy / rebuild detected: running restore() ==="
+    python3 /usr/local/bin/sync.py restore
+else
+    echo "=== openclaw.json exists, skipping restore() ==="
+fi
 
-# 執行恢復邏輯
-python3 /usr/local/bin/sync.py restore
+mkdir -p /root/.openclaw/sessions
 
 # provider1 model2~5 fallback 到自身 model1
 provide1_models_id2="${provide1_models_id2:-$provide1_models_id1}"
@@ -705,9 +577,6 @@ fi
 # 增量備份循環（每 30 分鐘後台運行）
 (while true; do sleep 1800; python3 /usr/local/bin/sync.py backup; done) &
 
-# Workspace 同步循環（每 30 分鐘後台運行）
-(while true; do sleep 1800; python3 /usr/local/bin/sync.py restore_workspace; done) &
-
 # 把 HF Space Secrets 裡的 GEMINI key 寫入 auth-profiles.json
 mkdir -p /root/.openclaw/agents/main/agent
 python3 - << 'PYEOF'
@@ -901,7 +770,7 @@ CS_PORT          = int(os.environ.get("CODE_SERVER_PORT", "13337"))
 CS_PASSWORD      = os.environ.get("CODE_SERVER_PASSWORD", "changeme123!")
 CS_USER_DATA_DIR = "/root/.code-server"
 CS_EXTENSIONS_DIR= "/root/.code-server/extensions"
-CS_WORKSPACE     = "/"
+CS_WORKSPACE     = "/root/.openclaw"
 CS_LOG           = "/root/.openclaw/logs/code-server.log"
 IDLE_MINUTES     = int(os.environ.get("IDE_IDLE_MINUTES", "30"))
 LAST_ACCESS_FILE = "/tmp/cs-last-access"