1

2

3

4

5

6

# 克隆项目

git clone https://github.com/openclaw/openclaw.git

cd openclaw

# 进入部署脚本目录

cd scripts/deploy

1

2

# 复制示例配置文件

cp config.example.yaml config.yaml

1

2

3

4

5

6

7

8

9

# config.yaml 配置示例

doubao:

  api_key: "your-doubao-api-key-here"  # 替换为您的真实 API Key

  model: "doubao-pro-256k"             # 选择豆包模型版本

  base_url: "https://ark.cn-beijing.volces.com/api/v3"  # API 端点

server:

  host: "0.0.0.0"

  port: 7860

  debug: false

1

2

3

4

5

# 赋予脚本执行权限

chmod +x deploy.sh

# 执行部署（根据提示操作）

./deploy.sh

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

import requests

import json

def chat_with_openclaw(use_stream=False):

    """

    与 Openclaw 服务进行对话的完整示例

    参数:

        use_stream (bool): 是否使用流式响应，默认为 False

    """

    # 1. 配置 API 端点

    url = "http://localhost:7860/api/v1/chat/completions"

    # 2. 设置请求头

    headers = {

        "Content-Type": "application/json",

        "Accept": "application/json"

    }

    # 3. 构建请求数据

    data = {

        "model": "doubao-pro-256k",  # 使用的模型名称

        "messages": [

            {"role": "system", "content": "你是一个专业的 AI 助手。"},  # 系统提示词

            {"role": "user", "content": "请用中文简要介绍一下 Openclaw 项目。"}  # 用户消息

        ],

        "temperature": 0.7,  # 控制回答的随机性 (0.0-1.0)

        "max_tokens": 500,   # 限制生成的最大 token 数

        "stream": use_stream  # 是否启用流式响应

    }

    try:

        # 4. 发送 POST 请求

        print(f"正在发送请求到 {url}...")

        print(f"使用流式响应: {use_stream}")

        if use_stream:

            # 流式响应处理

            print("开始接收流式响应:")

            print("-" * 40)

            response = requests.post(

                url,

                headers=headers,

                data=json.dumps(data),

                stream=True  # 启用流式传输

            )

            # 检查响应状态

            if response.status_code != 200:

                print(f"请求失败，状态码: {response.status_code}")

                print(f"错误信息: {response.text}")

                return None

            # 逐行处理流式响应

            full_response = ""

            for line in response.iter_lines():

                if line:

                    line_str = line.decode('utf-8')

                    # 跳过 SSE 格式中的 "data: " 前缀

                    if line_str.startswith("data: "):

                        line_str = line_str[6:]

                    # 检查是否为结束标记

                    if line_str == "[DONE]":

                        print("\n流式响应结束")

                        break

                    # 解析 JSON 数据

                    try:

                        chunk_data = json.loads(line_str)

                        if "choices" in chunk_data and len(chunk_data["choices"]) > 0:

                            delta = chunk_data["choices"][0].get("delta", {})

                            content = delta.get("content", "")

                            if content:

                                print(content, end="", flush=True)

                                full_response += content

                    except json.JSONDecodeError:

                        # 忽略非 JSON 数据行

                        continue

            print(f"\n{'='*40}")

            print(f"完整响应内容:\n{full_response}")

            return full_response

        else:

            # 非流式响应处理

            response = requests.post(

                url,

                headers=headers,

                data=json.dumps(data),

                timeout=30  # 设置超时时间（秒）

            )

            # 5. 检查响应状态

            response.raise_for_status()  # 如果状态码不是 200，抛出异常

            # 6. 解析响应数据

            result = response.json()

            print("请求成功!")

            print(f"状态码: {response.status_code}")

            print(f"响应时间: {response.elapsed.total_seconds():.2f}秒")

            # 7. 提取并显示回答内容

            if "choices" in result and len(result["choices"]) > 0:

                message = result["choices"][0].get("message", {})

                content = message.get("content", "无内容")

                print(f"\nAI 回答:")

                print("-" * 40)

                print(content)

                print("-" * 40)

                # 显示使用统计（如果存在）

                if "usage" in result:

                    usage = result["usage"]

                    print(f"\n使用统计:")

                    print(f"  提示词 tokens: {usage.get('prompt_tokens', 'N/A')}")

                    print(f"  完成 tokens: {usage.get('completion_tokens', 'N/A')}")

                    print(f"  总 tokens: {usage.get('total_tokens', 'N/A')}")

            return result

    except requests.exceptions.Timeout:

        print("错误: 请求超时，请检查网络连接或服务状态")

    except requests.exceptions.ConnectionError:

        print("错误: 连接失败，请确保 Openclaw 服务正在运行")

    except requests.exceptions.HTTPError as e:

        print(f"HTTP 错误: {e}")

        if response is not None:

            print(f"错误详情: {response.text}")

    except json.JSONDecodeError:

        print("错误: 响应不是有效的 JSON 格式")

        if response is not None:

            print(f"原始响应: {response.text}")

    except Exception as e:

        print(f"未知错误: {type(e).__name__}: {e}")

    return None

if __name__ == "__main__":

    print("=== Openclaw API 调用示例 ===")

    print("\n1. 非流式调用示例:")

    result1 = chat_with_openclaw(use_stream=False)

    print("\n\n2. 流式调用示例:")

    result2 = chat_with_openclaw(use_stream=True)

    print("\n=== 示例执行完成 ===")

1

2

3

# 使用国内镜像源

export PIP_INDEX_URL=https://pypi.tuna.tsinghua.edu.cn/simple

./deploy.sh

1

2

3

4

5

# 查看服务日志

journalctl -u openclaw.service -f

# 或直接查看输出日志

tail -f /var/log/openclaw.log