update prepare_environment

utils/prepare_environment.py

@@ -16,8 +16,14 @@ from tqdm import tqdm
 import tempfile
 
 DOWNLOADS = [
-    "http://172.16.15.118:9557/repository/atalink-hf-models/backups/data_cache_backup.tar.gz",
-    "http://172.16.15.118:9557/repository/atalink-hf-models/backups/data_gfpgan_backup.tar.gz",
+    (
+        "http://172.16.15.118:9557/repository/atalink-hf-models/backups/data_cache_backup.tar.gz",
+        3,
+    ),
+    (
+        "http://172.16.15.118:9557/repository/atalink-hf-models/backups/data_gfpgan_backup.tar.gz",
+        4,
+    ),
     # "https://huggingface.co/hynt/F5-TTS-Vietnamese-ViVoice/resolve/main/config.json"
 ]
 
@@ -34,7 +40,7 @@ def download_file(url, dest_folder):
         print(
             f"✅ [SKIP] {filename} already exists at \033[96m{dest_path}\033[0m, skipping download."
         )
-        return dest_path
+        return dest_path, False  # not downloaded this run
     print(
         f"\n📥 [START] Downloading: \033[1;36m{filename}\033[0m → \033[96m{dest_path}\033[0m"
     )
@@ -52,12 +58,34 @@ def download_file(url, dest_folder):
     print(
         f"✅ [DONE] Downloaded: \033[1;32m{filename}\033[0m → \033[96m{dest_path}\033[0m\n"
     )
-    return dest_path
+    return dest_path, True  # downloaded in this run
+
+
+def volume_exists(volume_name: str) -> bool:
+    """Return True if a Docker volume exists."""
+    try:
+        subprocess.run(
+            [
+                "docker",
+                "volume",
+                "inspect",
+                volume_name,
+            ],
+            check=True,
+            capture_output=True,
+        )
+        return True
+    except subprocess.CalledProcessError:
+        return False
 
 
-def create_volume_and_extract(tar_path, volume_name):
-    # Create Docker volume
-    subprocess.run(["docker", "volume", "create", volume_name], check=True)
+def create_volume_and_extract(tar_path, volume_name, extract_path=1):
+    # Create Docker volume only if it doesn't exist
+    if volume_exists(volume_name):
+        print(f"✅ [VOLUME] Exists: \033[1;33m{volume_name}\033[0m — skipping create")
+    else:
+        subprocess.run(["docker", "volume", "create", volume_name], check=True)
+        print(f"🚀 [VOLUME] Created: \033[1;33m{volume_name}\033[0m")
 
     # Convert host path để Docker hiểu cả Windows & Linux
     host_path = os.path.abspath(os.path.dirname(tar_path))
@@ -76,10 +104,10 @@ def create_volume_and_extract(tar_path, volume_name):
             f"{volume_name}:/data",  # Mount Docker volume (volume_name) vào thư mục /data trong container
             "-v",
             f"{host_path}:/tmpdata",  # Mount thư mục chứa file tar.gz trên host vào /tmpdata trong container
-            "busybox",  # Image dùng để chạy container (ở đây là Ubuntu 22.04)
+            "busybox",  # Sử dụng busybox, tar hỗ trợ --strip-components
             "sh",
-            "-c",  # Chạy lệnh bash trong container
-            f"tar -xzvf /tmpdata/{os.path.basename(tar_path)} --strip 1 -C /data",  # Lệnh giải nén file tar.gz từ /tmpdata vào /data
+            "-c",  # Chạy lệnh shell trong container
+            f"tar -xzvf /tmpdata/{os.path.basename(tar_path)} --strip {extract_path} -C /data",  # Lệnh giải nén file tar.gz từ /tmpdata vào /data
         ],
         check=True,
     )
@@ -99,16 +127,19 @@ def cleanup_tmp(files):
 
 def main():
     downloaded_files = []
-    for url in DOWNLOADS:
-        tar_path = download_file(url, TMP_DIR)
+    for url, extract_path in DOWNLOADS:
+        tar_path, downloaded = download_file(url, TMP_DIR)
         downloaded_files.append(tar_path)
         volume_name = (
             VOLUME_PREFIX
             + os.path.splitext(os.path.splitext(os.path.basename(tar_path))[0])[0]
         )
         print(f"🚀 [VOLUME] Name: \033[1;33m{volume_name}\033[0m")
-        create_volume_and_extract(tar_path, volume_name)
-    cleanup_tmp(downloaded_files)
+        create_volume_and_extract(tar_path, volume_name, extract_path)
+    if downloaded_files:
+        cleanup_tmp(downloaded_files)
+    else:
+        print("🧹 [CLEANUP] No new files downloaded — nothing to remove.")
 
 
 if __name__ == "__main__":
@@ -116,9 +147,9 @@ if __name__ == "__main__":
 
 
 # if __name__ == "__main__":
-    # Test create_volume_and_extract với file test_data_backup.tar.gz
-    # test_tar = os.path.join(TMP_DIR, "data_backup.tar.gz")
-    # if os.path.exists(test_tar):
-    #     create_volume_and_extract(test_tar, "atalink_data_backup")
-    # else:
-    #     print("File tmp/test_data_backup.tar.gz không tồn tại để test.")
+# Test create_volume_and_extract với file test_data_backup.tar.gz
+# test_tar = os.path.join(TMP_DIR, "data_backup.tar.gz")
+# if os.path.exists(test_tar):
+#     create_volume_and_extract(test_tar, "atalink_data_backup")
+# else:
+#     print("File tmp/test_data_backup.tar.gz không tồn tại để test.")