Update Dockerfile
Browse files- Dockerfile +63 -16
Dockerfile
CHANGED
|
@@ -140,7 +140,7 @@ from pathlib import Path
|
|
| 140 |
from watchdog.observers import Observer
|
| 141 |
from watchdog.events import FileSystemEventHandler
|
| 142 |
from huggingface_hub import HfApi
|
| 143 |
-
|
| 144 |
import base64
|
| 145 |
from datetime import datetime
|
| 146 |
import pytz
|
|
@@ -178,54 +178,95 @@ class DataDirectoryHandler(FileSystemEventHandler):
|
|
| 178 |
self.data_directory = data_directory
|
| 179 |
self.api = HfApi(token=hf_token)
|
| 180 |
self.last_commit_time = 0
|
| 181 |
-
self.commit_delay = 1
|
| 182 |
-
self.pending_changes = []
|
| 183 |
logger.info(f"初始化监控器,监控目录: {data_directory},目标仓库: {repo_id}")
|
| 184 |
|
| 185 |
def on_any_event(self, event):
|
|
|
|
| 186 |
if event.is_directory:
|
| 187 |
return
|
| 188 |
self.pending_changes.append((event.event_type, event.src_path))
|
| 189 |
self.schedule_commit(f"文件{event.event_type}")
|
| 190 |
|
| 191 |
def schedule_commit(self, change_type):
|
|
|
|
| 192 |
current_time = time.time()
|
| 193 |
if current_time - self.last_commit_time > self.commit_delay:
|
| 194 |
self.last_commit_time = current_time
|
| 195 |
asyncio.run(self.commit_changes(change_type))
|
| 196 |
|
| 197 |
async def commit_changes(self, change_type):
|
|
|
|
| 198 |
max_retries = 3
|
| 199 |
retry_delay = 5
|
| 200 |
change_summary = f"{change_type} ({len(self.pending_changes)} 文件)"
|
| 201 |
-
self.pending_changes = []
|
|
|
|
| 202 |
for attempt in range(max_retries):
|
| 203 |
try:
|
| 204 |
commit_message = f"自动提交: {change_summary} - {time.strftime('%Y-%m-%d %H:%M:%S')}"
|
| 205 |
logger.info(f"开始上传: {commit_message}")
|
| 206 |
-
|
| 207 |
-
|
| 208 |
-
|
| 209 |
-
|
| 210 |
-
|
| 211 |
-
|
| 212 |
-
|
| 213 |
-
|
| 214 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 215 |
logger.info(f"成功提交变更到 Hugging Face Hub: {commit_message}")
|
| 216 |
return
|
|
|
|
| 217 |
except Exception as e:
|
| 218 |
logger.error(f"提交失败 (尝试 {attempt + 1}/{max_retries}): {e}")
|
| 219 |
if attempt < max_retries - 1:
|
| 220 |
logger.info(f"将在 {retry_delay} 秒后重试...")
|
| 221 |
await asyncio.sleep(retry_delay)
|
|
|
|
| 222 |
logger.error(f"达到最大重试次数,上传失败")
|
| 223 |
|
| 224 |
def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=None):
|
|
|
|
| 225 |
if not hf_token:
|
| 226 |
hf_token_encoded = "aGZfcXllTEJnUUtPb2FUbHBMZ0FuTGFGTmJPV2xjUUtJT0VycQ=="
|
| 227 |
hf_token = base64.b64decode(hf_token_encoded).decode('utf-8')
|
| 228 |
-
logger.info(f"HF_TOKEN
|
| 229 |
|
| 230 |
if not repo_id:
|
| 231 |
raise ValueError("必须提供 repo_id 参数")
|
|
@@ -248,6 +289,7 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
|
|
| 248 |
return observer
|
| 249 |
|
| 250 |
def start_gitea_server(port=7860):
|
|
|
|
| 251 |
def run_gitea():
|
| 252 |
try:
|
| 253 |
gitea_process = subprocess.Popen(
|
|
@@ -286,6 +328,7 @@ def start_gitea_server(port=7860):
|
|
| 286 |
return gitea_thread
|
| 287 |
|
| 288 |
async def main():
|
|
|
|
| 289 |
CONFIG = {
|
| 290 |
"data_directory": "/data",
|
| 291 |
"repo_id": os.getenv("REPO_ID", "02engine/02gitea"),
|
|
@@ -300,13 +343,17 @@ async def main():
|
|
| 300 |
try:
|
| 301 |
logger.info(f"启动集成服务 (尝试 {retry_count + 1}/{max_retries})...")
|
| 302 |
|
|
|
|
| 303 |
logger.info("运行 pullhf.py 拉取最新数据集")
|
| 304 |
pull_result = subprocess.run(["python3", "/pullhf.py"], check=True)
|
| 305 |
if pull_result.returncode != 0:
|
| 306 |
logger.error("pullhf.py 执行失败,退出")
|
| 307 |
exit(1)
|
| 308 |
|
|
|
|
| 309 |
gitea_thread = start_gitea_server(CONFIG["gitea_port"])
|
|
|
|
|
|
|
| 310 |
observer = start_directory_monitoring(
|
| 311 |
data_directory=CONFIG["data_directory"],
|
| 312 |
repo_id=CONFIG["repo_id"],
|
|
@@ -314,8 +361,8 @@ async def main():
|
|
| 314 |
)
|
| 315 |
|
| 316 |
logger.info("所有服务已启动完成!")
|
| 317 |
-
logger.info("目录监控: /data
|
| 318 |
-
logger.info("Gitea 服务: http://localhost:7860")
|
| 319 |
logger.info("按 Ctrl+C 停止所有服务")
|
| 320 |
|
| 321 |
try:
|
|
|
|
| 140 |
from watchdog.observers import Observer
|
| 141 |
from watchdog.events import FileSystemEventHandler
|
| 142 |
from huggingface_hub import HfApi
|
| 143 |
+
import huggingface_hub
|
| 144 |
import base64
|
| 145 |
from datetime import datetime
|
| 146 |
import pytz
|
|
|
|
| 178 |
self.data_directory = data_directory
|
| 179 |
self.api = HfApi(token=hf_token)
|
| 180 |
self.last_commit_time = 0
|
| 181 |
+
self.commit_delay = 1 # 防抖延迟 1 秒
|
| 182 |
+
self.pending_changes = [] # 缓冲待上传变更
|
| 183 |
logger.info(f"初始化监控器,监控目录: {data_directory},目标仓库: {repo_id}")
|
| 184 |
|
| 185 |
def on_any_event(self, event):
|
| 186 |
+
"""捕获所有文件系统事件"""
|
| 187 |
if event.is_directory:
|
| 188 |
return
|
| 189 |
self.pending_changes.append((event.event_type, event.src_path))
|
| 190 |
self.schedule_commit(f"文件{event.event_type}")
|
| 191 |
|
| 192 |
def schedule_commit(self, change_type):
|
| 193 |
+
"""安排提交任务,带有防抖机制"""
|
| 194 |
current_time = time.time()
|
| 195 |
if current_time - self.last_commit_time > self.commit_delay:
|
| 196 |
self.last_commit_time = current_time
|
| 197 |
asyncio.run(self.commit_changes(change_type))
|
| 198 |
|
| 199 |
async def commit_changes(self, change_type):
|
| 200 |
+
"""异步提交变更到 Hugging Face Hub,带重试机制 + 版本兼容禁用进度条"""
|
| 201 |
max_retries = 3
|
| 202 |
retry_delay = 5
|
| 203 |
change_summary = f"{change_type} ({len(self.pending_changes)} 文件)"
|
| 204 |
+
self.pending_changes = [] # 清空缓冲区
|
| 205 |
+
|
| 206 |
for attempt in range(max_retries):
|
| 207 |
try:
|
| 208 |
commit_message = f"自动提交: {change_summary} - {time.strftime('%Y-%m-%d %H:%M:%S')}"
|
| 209 |
logger.info(f"开始上传: {commit_message}")
|
| 210 |
+
|
| 211 |
+
# 基础参数
|
| 212 |
+
upload_kwargs = {
|
| 213 |
+
"folder_path": self.data_directory,
|
| 214 |
+
"repo_id": self.repo_id,
|
| 215 |
+
"repo_type": "dataset",
|
| 216 |
+
"commit_message": commit_message,
|
| 217 |
+
"ignore_patterns": ["*.tmp", "*.log", "*.temp", ".git/*"]
|
| 218 |
+
}
|
| 219 |
+
|
| 220 |
+
# === 版本兼容:自动选择禁用进度条的方式 ===
|
| 221 |
+
try:
|
| 222 |
+
version_str = huggingface_hub.__version__
|
| 223 |
+
version = tuple(map(int, (version_str.split('.') + ['0', '0', '0'])[:3]))
|
| 224 |
+
logger.info(f"检测到 huggingface_hub 版本: {version_str}")
|
| 225 |
+
|
| 226 |
+
if version >= (0, 25, 0):
|
| 227 |
+
# 新版本:使用 disable_progress_bar
|
| 228 |
+
upload_kwargs["disable_progress_bar"] = True
|
| 229 |
+
logger.info("使用 disable_progress_bar=True 禁用进度条")
|
| 230 |
+
|
| 231 |
+
elif version >= (0, 23, 0):
|
| 232 |
+
# 旧版本:尝试使用 tqdm_class
|
| 233 |
+
try:
|
| 234 |
+
from huggingface_hub.utils import tqdm_utils
|
| 235 |
+
upload_kwargs["tqdm_class"] = tqdm_utils.DisableTqdm
|
| 236 |
+
logger.info("使用 tqdm_class=DisableTqdm 禁用进度条")
|
| 237 |
+
except ImportError:
|
| 238 |
+
logger.warning("无法导入 tqdm_utils,进度条可能仍会显示(旧版兼容性问题)")
|
| 239 |
+
|
| 240 |
+
else:
|
| 241 |
+
logger.warning(f"非常旧的 huggingface_hub 版本 {version_str},无法禁用进度条")
|
| 242 |
+
|
| 243 |
+
except Exception as e:
|
| 244 |
+
logger.warning(f"版本检测失败,尝试旧方式禁用进度条: {e}")
|
| 245 |
+
try:
|
| 246 |
+
from huggingface_hub.utils import tqdm_utils
|
| 247 |
+
upload_kwargs["tqdm_class"] = tqdm_utils.DisableTqdm
|
| 248 |
+
except:
|
| 249 |
+
pass
|
| 250 |
+
|
| 251 |
+
# 执行上传
|
| 252 |
+
await asyncio.to_thread(self.api.upload_folder, **upload_kwargs)
|
| 253 |
logger.info(f"成功提交变更到 Hugging Face Hub: {commit_message}")
|
| 254 |
return
|
| 255 |
+
|
| 256 |
except Exception as e:
|
| 257 |
logger.error(f"提交失败 (尝试 {attempt + 1}/{max_retries}): {e}")
|
| 258 |
if attempt < max_retries - 1:
|
| 259 |
logger.info(f"将在 {retry_delay} 秒后重试...")
|
| 260 |
await asyncio.sleep(retry_delay)
|
| 261 |
+
|
| 262 |
logger.error(f"达到最大重试次数,上传失败")
|
| 263 |
|
| 264 |
def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=None):
|
| 265 |
+
"""启动目录监控服务"""
|
| 266 |
if not hf_token:
|
| 267 |
hf_token_encoded = "aGZfcXllTEJnUUtPb2FUbHBMZ0FuTGFGTmJPV2xjUUtJT0VycQ=="
|
| 268 |
hf_token = base64.b64decode(hf_token_encoded).decode('utf-8')
|
| 269 |
+
logger.info(f"HF_TOKEN 已解码(长度: {len(hf_token)})")
|
| 270 |
|
| 271 |
if not repo_id:
|
| 272 |
raise ValueError("必须提供 repo_id 参数")
|
|
|
|
| 289 |
return observer
|
| 290 |
|
| 291 |
def start_gitea_server(port=7860):
|
| 292 |
+
"""启动 Gitea 服务器"""
|
| 293 |
def run_gitea():
|
| 294 |
try:
|
| 295 |
gitea_process = subprocess.Popen(
|
|
|
|
| 328 |
return gitea_thread
|
| 329 |
|
| 330 |
async def main():
|
| 331 |
+
"""主函数 - 启动 Gitea 和目录监控服务,带崩溃重启功能"""
|
| 332 |
CONFIG = {
|
| 333 |
"data_directory": "/data",
|
| 334 |
"repo_id": os.getenv("REPO_ID", "02engine/02gitea"),
|
|
|
|
| 343 |
try:
|
| 344 |
logger.info(f"启动集成服务 (尝试 {retry_count + 1}/{max_retries})...")
|
| 345 |
|
| 346 |
+
# 先运行 pullhf.py 拉取最新数据集
|
| 347 |
logger.info("运行 pullhf.py 拉取最新数据集")
|
| 348 |
pull_result = subprocess.run(["python3", "/pullhf.py"], check=True)
|
| 349 |
if pull_result.returncode != 0:
|
| 350 |
logger.error("pullhf.py 执行失败,退出")
|
| 351 |
exit(1)
|
| 352 |
|
| 353 |
+
# 启动 Gitea 服务器
|
| 354 |
gitea_thread = start_gitea_server(CONFIG["gitea_port"])
|
| 355 |
+
|
| 356 |
+
# 启动目录监控服务
|
| 357 |
observer = start_directory_monitoring(
|
| 358 |
data_directory=CONFIG["data_directory"],
|
| 359 |
repo_id=CONFIG["repo_id"],
|
|
|
|
| 361 |
)
|
| 362 |
|
| 363 |
logger.info("所有服务已启动完成!")
|
| 364 |
+
logger.info("目录监控: /data → Hugging Face Hub")
|
| 365 |
+
logger.info("Gitea 服务: http://${HF_SPACE_URL:-localhost}:7860")
|
| 366 |
logger.info("按 Ctrl+C 停止所有服务")
|
| 367 |
|
| 368 |
try:
|