抱抱脸🤗部署reader服务器版

6 天前(已编辑)
/

抱抱脸🤗部署reader服务器版

时隔许久，更新一下发现的白嫖部署阅读服务器版的方法，该方法基于huggingface

第一步当然是注册账号啦（这里提醒，这个网站是处于被墙状态，自行解决哈

https://huggingface.co/

创建一个space

https://huggingface.co/new-space?sdk=docker

自己输入spacename，名字随便，类型选择docker，空间类型选择公开的话后续建议reader，私有空间后续可以用cf workers反代暴露出来

image
image
​

下面新建程序文件

首先点开files界面然后选择新建文件

image
image
image
image

新建下面的文件

Dockerfile

# 使用官方Ubuntu基础镜像
FROM ubuntu:22.04

# 设置环境变量
ENV TZ=Asia/Shanghai \
    JAVA_HOME=/usr/lib/jvm/java-17-openjdk-amd64 \
    APP_USER=reader \
    APP_HOME=/app
    

# 安装依赖并配置环境
RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime \
    && echo $TZ > /etc/timezone \
    && apt-get update \
    && apt-get install -y --no-install-recommends \
        openjdk-17-jdk \
        ca-certificates \
        curl \
        unzip \
        bash \
        jq \
        python3 \
        python3-pip \
        python3-venv \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/* \
    && update-ca-certificates \
    && useradd -r -u 1000 -m -d $APP_HOME -s /bin/bash $APP_USER \
    && chown -R $APP_USER:$APP_USER $APP_HOME

ENV VIRTUAL_ENV=/app/venv
RUN python3 -m venv $VIRTUAL_ENV
ENV PATH="$VIRTUAL_ENV/bin:$PATH"
RUN pip install --no-cache-dir huggingface_hub
    

# 配置Java环境路径
ENV PATH=$JAVA_HOME/bin:$PATH
ENV LANG C.UTF-8
ENV LC_ALL C.UTF-8

# 设置工作目录和用户
WORKDIR $APP_HOME
USER $APP_USER

# 复制安装脚本并设置权限
COPY --chown=reader:reader install_reader.sh ./
COPY --chown=reader:reader sync_data.sh ./
RUN chmod +x install_reader.sh \ 
    && sed -i 's|./bin/startup.sh -m single|exec &|' install_reader.sh
RUN chmod +x sync_data.sh

EXPOSE 8080
# 容器入口
CMD ["/bin/sh", "-c", "./sync_data.sh"]

‍

install_reader.sh

#!/bin/bash

set -euo pipefail

# 获取最新版本的重定向URL
echo "正在获取最新版本信息..."
redirect_url=$(curl -Ls -o /dev/null -w '%{url_effective}' 'https://github.com/hectorqin/reader/releases/latest')

# 提取版本标签
tag=$(basename "$redirect_url")
if [[ ! "$tag" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
    echo "错误：无效的版本标签 '$tag'"
    exit 1
fi
version="${tag#v}"
echo "检测到最新版本: $version"

# 构造下载链接
download_url="https://github.com/hectorqin/reader/releases/download/${tag}/reader-server-${version}.zip"
echo "开始下载: $download_url"

# 下载文件
if ! curl -LO "$download_url"; then
    echo "错误：文件下载失败"
    exit 1
fi

# 解压文件
zip_file="reader-server-${version}.zip"
echo "正在解压文件..."
unzip "$zip_file"

#+x

cd target
chmod +x reader-pro-${version}.jar

cd ../bin
# 执行启动脚本
if [ -f "./startup.sh" ]; then
    echo "正在启动服务..."
    chmod +x "./startup.sh"
    ./startup.sh -m single
    echo "服务已启动！"
    tail -f /app/logs/start.out
else
    echo "错误：启动脚本不存在"
    exit 1
fi

‍

sync_data.sh

#!/bin/sh

# 检查环境变量
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
    echo "Starting without backup functionality - missing HF_TOKEN or DATASET_ID"
    exit 1
fi

# 激活虚拟环境
. /app/venv/bin/activate

# 上传备份
cat > hf_sync.py << 'EOL'
from huggingface_hub import HfApi
import sys
import os
import tarfile
import tempfile

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'Deleted old backup: {file_to_delete}')
            except Exception as e:
                print(f'Error deleting {file_to_delete}: {str(e)}')

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"Successfully uploaded {file_name}")
        
        manage_backups(api, repo_id)
    except Exception as e:
        print(f"Error uploading file: {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("No backup files found")
            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)  # 解压到指定路径
                print(f"Successfully restored backup from {latest_backup}")
                
    except Exception as e:
        print(f"Error downloading backup: {str(e)}")

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 '.'  # 默认为当前目录
        download_latest_backup(token, repo_id, extract_path)
EOL

# 首次启动时从HuggingFace下载最新备份（解压到应用目录）
echo "Downloading latest backup from HuggingFace..."
python hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}" "./"  

# 同步函数
sync_data() {
    while true; do
        echo "Starting sync process at $(date)"
        
        # 确保数据目录存在（选择你的实际路径）
        STORAGE_PATH="./storage"  # 或改为"./storage"
        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 "Uploading backup to HuggingFace..."
            python hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "/tmp/${backup_file}" "${backup_file}"
            
            # 清理临时文件
            rm -f "/tmp/${backup_file}"
        else
            echo "Storage directory ${STORAGE_PATH} does not exist, waiting..."
        fi
        
        # 同步间隔
        SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
        echo "Next sync in ${SYNC_INTERVAL} seconds..."
        sleep $SYNC_INTERVAL
    done
}

# 启动同步进程
sync_data &

# 启动主应用（根据实际路径调整）
exec bash install_reader.sh  # 或改为你的启动命令

‍

并且修改README.md，在最后加入一行app_port: 8080​

‍

创建dataset数据库

image
image
image
image

记住你的数据库地址，格式：hugging名称/dataset名称

创建HF-token

image
image
image
image
image
image
完成后保存并记住你的token

在后台添加环境变量

​HF_TOKEN #你的token DATASET_ID # 用户名/数据集名称 SYNC_INTERVAL # 同步时间（秒钟）例如3600​

image
image

访问

现在在看到你的空间状态变为running后访问app选项卡应该就可以看到你的阅读服务器网页了

公开空间

下面，如果你的空间是公开空间，访问https://用户名-空间名.hf.space 就能访问到你的应用了

如果想要使用自己的域名，可以使用cf workers反代

export default {
  async fetch(request, env) {
    const url = new URL(request.url);
    url.host = '你的地址'; 
    return fetch(new Request(url, request))
  }
}

私有空间

参见https://linux.do/t/topic/282505

使用社交账号登录

  • Loading...
  • Loading...
  • Loading...
  • Loading...
  • Loading...