Update Dockerfile
Browse files- Dockerfile +5 -18
Dockerfile
CHANGED
|
@@ -128,7 +128,6 @@ if __name__ == "__main__":
|
|
| 128 |
exit(1)
|
| 129 |
EOF
|
| 130 |
|
| 131 |
-
# 复制上传脚本(已禁用进度条)
|
| 132 |
COPY --chown=git:git <<'EOF' /uploadhf.py
|
| 133 |
import os
|
| 134 |
import time
|
|
@@ -141,7 +140,7 @@ from pathlib import Path
|
|
| 141 |
from watchdog.observers import Observer
|
| 142 |
from watchdog.events import FileSystemEventHandler
|
| 143 |
from huggingface_hub import HfApi
|
| 144 |
-
from huggingface_hub.utils import tqdm_utils
|
| 145 |
import base64
|
| 146 |
from datetime import datetime
|
| 147 |
import pytz
|
|
@@ -179,30 +178,27 @@ class DataDirectoryHandler(FileSystemEventHandler):
|
|
| 179 |
self.data_directory = data_directory
|
| 180 |
self.api = HfApi(token=hf_token)
|
| 181 |
self.last_commit_time = 0
|
| 182 |
-
self.commit_delay = 1
|
| 183 |
-
self.pending_changes = []
|
| 184 |
logger.info(f"初始化监控器,监控目录: {data_directory},目标仓库: {repo_id}")
|
| 185 |
|
| 186 |
def on_any_event(self, event):
|
| 187 |
-
"""捕获所有文件系统事件"""
|
| 188 |
if event.is_directory:
|
| 189 |
return
|
| 190 |
self.pending_changes.append((event.event_type, event.src_path))
|
| 191 |
self.schedule_commit(f"文件{event.event_type}")
|
| 192 |
|
| 193 |
def schedule_commit(self, change_type):
|
| 194 |
-
"""安排提交任务,带有防抖机制"""
|
| 195 |
current_time = time.time()
|
| 196 |
if current_time - self.last_commit_time > self.commit_delay:
|
| 197 |
self.last_commit_time = current_time
|
| 198 |
asyncio.run(self.commit_changes(change_type))
|
| 199 |
|
| 200 |
async def commit_changes(self, change_type):
|
| 201 |
-
"""异步提交变更到 Hugging Face Hub,带重试机制"""
|
| 202 |
max_retries = 3
|
| 203 |
retry_delay = 5
|
| 204 |
change_summary = f"{change_type} ({len(self.pending_changes)} 文件)"
|
| 205 |
-
self.pending_changes = []
|
| 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')}"
|
|
@@ -214,7 +210,7 @@ class DataDirectoryHandler(FileSystemEventHandler):
|
|
| 214 |
repo_type="dataset",
|
| 215 |
commit_message=commit_message,
|
| 216 |
ignore_patterns=["*.tmp", "*.log", "*.temp", ".git/*"],
|
| 217 |
-
|
| 218 |
)
|
| 219 |
logger.info(f"成功提交变更到 Hugging Face Hub: {commit_message}")
|
| 220 |
return
|
|
@@ -226,8 +222,6 @@ class DataDirectoryHandler(FileSystemEventHandler):
|
|
| 226 |
logger.error(f"达到最大重试次数,上传失败")
|
| 227 |
|
| 228 |
def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=None):
|
| 229 |
-
"""启动目录监控服务"""
|
| 230 |
-
# 硬编码的 base64 编码的 HF_TOKEN
|
| 231 |
if not hf_token:
|
| 232 |
hf_token_encoded = "aGZfcXllTEJnUUtPb2FUbHBMZ0FuTGFGTmJPV2xjUUtJT0VycQ=="
|
| 233 |
hf_token = base64.b64decode(hf_token_encoded).decode('utf-8')
|
|
@@ -254,7 +248,6 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
|
|
| 254 |
return observer
|
| 255 |
|
| 256 |
def start_gitea_server(port=7860):
|
| 257 |
-
"""启动 Gitea 服务器"""
|
| 258 |
def run_gitea():
|
| 259 |
try:
|
| 260 |
gitea_process = subprocess.Popen(
|
|
@@ -293,8 +286,6 @@ def start_gitea_server(port=7860):
|
|
| 293 |
return gitea_thread
|
| 294 |
|
| 295 |
async def main():
|
| 296 |
-
"""主函数 - 启动 Gitea 和目录监控服务,带崩溃重启功能"""
|
| 297 |
-
# 配置参数
|
| 298 |
CONFIG = {
|
| 299 |
"data_directory": "/data",
|
| 300 |
"repo_id": os.getenv("REPO_ID", "02engine/02gitea"),
|
|
@@ -309,17 +300,13 @@ async def main():
|
|
| 309 |
try:
|
| 310 |
logger.info(f"启动集成服务 (尝试 {retry_count + 1}/{max_retries})...")
|
| 311 |
|
| 312 |
-
# 先运行 pullhf.py 拉取最新数据集
|
| 313 |
logger.info("运行 pullhf.py 拉取最新数据集")
|
| 314 |
pull_result = subprocess.run(["python3", "/pullhf.py"], check=True)
|
| 315 |
if pull_result.returncode != 0:
|
| 316 |
logger.error("pullhf.py 执行失败,退出")
|
| 317 |
exit(1)
|
| 318 |
|
| 319 |
-
# 启动 Gitea 服务器
|
| 320 |
gitea_thread = start_gitea_server(CONFIG["gitea_port"])
|
| 321 |
-
|
| 322 |
-
# 启动目录监控服务
|
| 323 |
observer = start_directory_monitoring(
|
| 324 |
data_directory=CONFIG["data_directory"],
|
| 325 |
repo_id=CONFIG["repo_id"],
|
|
|
|
| 128 |
exit(1)
|
| 129 |
EOF
|
| 130 |
|
|
|
|
| 131 |
COPY --chown=git:git <<'EOF' /uploadhf.py
|
| 132 |
import os
|
| 133 |
import time
|
|
|
|
| 140 |
from watchdog.observers import Observer
|
| 141 |
from watchdog.events import FileSystemEventHandler
|
| 142 |
from huggingface_hub import HfApi
|
| 143 |
+
# 删除:from huggingface_hub.utils import tqdm_utils
|
| 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
|
| 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')}"
|
|
|
|
| 210 |
repo_type="dataset",
|
| 211 |
commit_message=commit_message,
|
| 212 |
ignore_patterns=["*.tmp", "*.log", "*.temp", ".git/*"],
|
| 213 |
+
disable_progress_bar=True # 正确禁用进度条
|
| 214 |
)
|
| 215 |
logger.info(f"成功提交变更到 Hugging Face Hub: {commit_message}")
|
| 216 |
return
|
|
|
|
| 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')
|
|
|
|
| 248 |
return observer
|
| 249 |
|
| 250 |
def start_gitea_server(port=7860):
|
|
|
|
| 251 |
def run_gitea():
|
| 252 |
try:
|
| 253 |
gitea_process = subprocess.Popen(
|
|
|
|
| 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 |
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"],
|