iflytek/spark-ai-python

def test_spark_openai_style_stream():
    # 使用客户端创建一个 completion
    completion = client.chat.completions.create(
        model="星火的domain",
        messages=[
            {"role": "system", "content": "帮我根据用户意图生成如下Json格式响应: {\"intent\": \"用户意图\"} \n可选的意图有: 天气查询，航班查询，知识库查询、订餐查询、火车票查询",},
            {"role": "user", "content": "今天天气如何?"},
        ],
        user="",
        stream=True,
        seed=2, 
        top_p=0.1,
        #tools=[ {"type":"function", "function": function} for function in tools],
       # tool_choice={"type":"function", "function": tools[1]},
        extra_body={
            # 星火支持的高级参数传入
            "patch_id": ["0"]

        },
        extra_headers={
        })
    print(completion)
    for response in completion:
       print(response)

pip install --upgrade spark_ai_python

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple spark_ai_python

pip install -i  https://repo.model.xfyun.cn/api/packages/administrator/pypi/simple  spark_ai_python --upgrade

pip install -e .

import os

from sparkai.llm.llm import ChatSparkLLM, ChunkPrintHandler
from sparkai.core.messages import ChatMessage
try:
    from dotenv import load_dotenv
except ImportError:
    raise RuntimeError('Python environment for SPARK AI is not completely set up: required package "python-dotenv" is missing.') from None

load_dotenv()

if __name__ == '__main__':
    from sparkai.core.callbacks import StdOutCallbackHandler
    spark = ChatSparkLLM(
        spark_api_url=os.environ["SPARKAI_URL"],
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain=os.environ["SPARKAI_DOMAIN"],
        streaming=False,
    )
    messages = [ChatMessage(
        role="user",
        content='你好呀'
    )]
    handler = ChunkPrintHandler()
    a = spark.generate([messages], callbacks=[handler])
    print(a)
    print(a.generations[0][0].text)
    print(a.llm_output)

import os

from sparkai.llm.llm import ChatSparkLLM, ChunkPrintHandler
from sparkai.core.messages import ChatMessage
try:
    from dotenv import load_dotenv
except ImportError:
    raise RuntimeError('Python environment for SPARK AI is not completely set up: required package "python-dotenv" is missing.') from None

load_dotenv()
def test_stream():
    from sparkai.core.callbacks import StdOutCallbackHandler
    spark = ChatSparkLLM(
        spark_api_url=os.environ["SPARKAI_URL"],
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain=os.environ["SPARKAI_DOMAIN"],
        request_timeout=30, # 
        streaming=True,

    )
    messages = [ChatMessage(
        role="user",
        content='作为AutoSpark的创建者角色，当前需要你帮我分析并生成任务，你当前主要目标是:\n1. 帮我写个贪吃蛇python游戏\n\n\n\n\n当前执行任务节点是: `编写贪吃蛇游戏的界面设计`\n\n任务执行历史是:\n`\nTask: 使用ThinkingTool分析贪吃蛇游戏的需求\nResult: Error2: {\'error\': \'Could not parse invalid format: 根据任务节点，我将使用ThinkingTool来分析贪吃蛇游戏的需求。首先，我们需要理解游戏的基本功能和规则。贪吃蛇游戏的主要目标是控制一条蛇在屏幕上移动，吃到食物后蛇会变长，碰到自己的身体或者屏幕边缘则游戏结束。\\n\\n接下来，我们可以使用CodingTool来编写贪吃蛇游戏的代码。首先，我们需要定义以下核心类和方法：\\n\\n1. Snake类：用于表示贪吃蛇的状态，包括蛇的身体、移动方向等。\\n2. Food类：用于表示食物的位置。\\n3. Game类：用于控制游戏的进行，包括初始化游戏、更新蛇的位置、检查碰撞等。\\n4. main函数：用于启动游戏。\\n\\n接下来，我们将这些类和方法的代码写入文件中。\\n\\n最后，我们可以使用WriteTestTool来编写测试用例，确保我们的代码能够正确地运行。测试用例应该包括以下内容：\\n\\n1. 测试游戏是否能正确初始化。\\n2. 测试蛇是否能正确移动。\\n3. 测试蛇是否能正确吃到食物并变长。\\n4. 测试蛇是否能正确碰到自己的身体或屏幕边缘导致游戏结束。 exceptionNot get command from llm response...\'}. \nTask: 编写贪吃蛇游戏的spec文件\nResult: Error2: {\'error\': \'Could not parse invalid format: 根据任务节点，我将使用`WriteSpecTool`来编写贪吃蛇游戏的spec文件。\\n\\n首先，我们需要定义以下核心类和方法：\\n1. Snake类：用于表示贪吃蛇的状态，包括蛇的身体、移动方向等。\\n2. Food类：用于表示食物的位置。\\n3. Game类：用于控制游戏的进行，包括初始化游戏、更新蛇的位置、检查碰撞等。\\n4. main函数：用于启动游戏。\\n\\n接下来，我们将这些类和方法的代码写入文件中。\\n\\n最后，我们可以使用`WriteTestTool`来编写测试用例，确保我们的代码能够正确地运行。测试用例应该包括以下内容：\\n1. 测试游戏是否能正确初始化。\\n2. 测试蛇是否能正确移动。\\n3. 测试蛇是否能正确吃到食物并变长。\\n4. 测试蛇是否能正确碰到自己的身体或屏幕边缘导致游戏结束。 exceptionNot get command from llm response...\'}. \n\n`\n\n根据上述背景信息，你的任务是需要理解当前的任务节点关键信息，创建一个规划，解释为什么要这么做，并且提及一些需要注意的事项，必须从下述TOOLS中挑选一个命令用于下一步执行。\n\nTOOLS:\n1. "ThinkingTool": Intelligent problem-solving assistant that comprehends tasks, identifies key variables, and makes efficient decisions, all while providing detailed, self-driven reasoning for its choices. Do not assume anything, take the details from given data only., args : task_description: "<task_description>",\n2. "WriteSpecTool": A tool to write the spec of a program., args : task_description: "<task_description>",spec_file_name: "<spec_file_name>",\n3. "CodingTool": You will get instructions for code to write. You will write a very long answer. Make sure that every detail of the architecture is, in the end, implemented as code. Think step by step and reason yourself to the right decisions to make sure we get it right. You will first lay out the names of the core classes, functions, methods that will be necessary, as well as a quick comment on their purpose. Then you will output the content of each file including ALL code., args : code_description: "<code_description>",\n4. "WriteTestTool": 您是一位超级聪明的开发人员，使用测试驱动开发根据规范编写测试。\n请根据上述规范生成测试。测试应该尽可能简单， 但仍然涵盖了所有功能。\n将它们写入文件中, args : test_description: "<test_description>",test_file_name: "<test_file_name>",\n\n\n\n约束条件:\n1. 请注意返回的命令名称和参数不要被引号包裹\n2. 命令名称必须是TOOLS中的已知的\n3. 你只能生成一个待执行命令名称及其对应参数\n4. 你生成的命令必须是用来解决 `编写贪吃蛇游戏的界面设计`\n\n在之后的每次回答中，你必须严格遵从上述约束条件并按照如下JsonSchema约束返回响应:\n\n{\n "$schema": "http://json-schema.org/draft-07/schema#",\n "type": "object",\n "properties": {\n "thoughts": {\n "type": "object",\n "properties": {\n "reasoning": {\n "type": "string",\n "description": "short reasoning",\n }\n },\n "required": ["reasoning"]\n },\n "tool": {\n "type": "object",\n "properties": {\n "name": {\n "type": "string",\n "description": "tool name",\n },\n "args": {\n "type": "object",\n "description": "tool arguments",\n }\n },\n "required": ["name", "args"]\n }\n }\n}'
    )]
    handler = ChunkPrintHandler()
    a = spark.generate([messages], callbacks=[handler])
    print(a)

class ChunkPrintHandler(BaseCallbackHandler):
    """Callback Handler that prints to std out."""

    def __init__(self, color: Optional[str] = None) -> None:
        """Initialize callback handler."""
        self.color = color

    def on_llm_new_token(self,  token: str,
        *,
        chunk:  None,
        **kwargs: Any,):
        print(token)
        print(kwargs)
        # token 为模型生成的token
        # 可以check kwargs内容，kwargs中 llm_output中有usage相关信息， final表示是否是最后一帧
        

def test_starcoder2():
    from sparkai.log.logger import logger
    #logger.setLevel("debug")
    from sparkai.core.callbacks import StdOutCallbackHandler
    messages = [{'role': 'user',
                 'content': "帮我生成一段代码，爬取baidu.com"}]
    spark = ChatSparkLLM(
        spark_api_url="wss://xingchen-api.cn-huabei-1.xf-yun.com/v1.1/chat",
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain="xsstarcoder27binst",
        streaming=True,
        max_tokens= 1024,
    )
    messages = [
                ChatMessage(
                        role="user",
                        content=messages[0]['content']

    )]

    a = spark.stream(messages)
    for message in a:
        print(message)

async def test_astream():
    from sparkai.log.logger import logger
    #logger.setLevel("debug")
    from sparkai.core.callbacks import StdOutCallbackHandler
    messages = [{'role': 'user',
                 'content': "帮我生成一段代码，爬取baidu.com"}]
    spark = ChatSparkLLM(
        spark_api_url="wss://xingchen-api.cn-huabei-1.xf-yun.com/v1.1/chat",
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain="xsstarcoder27binst",
        streaming=True,
        max_tokens= 1024,
    )
    messages = [
                ChatMessage(
                        role="user",
                        content=messages[0]['content']

    )]
    handler = AsyncChunkPrintHandler()
    a = spark.astream(messages, config={"callbacks": [handler]})
    async for message in a:
        print(message)

if __name__ == '__main__':
    import asyncio
    asyncio.run(test_astream())

from sparkai.core.utils.function_calling import convert_to_openai_tool, convert_to_openai_function
def multiply(a,b :int) -> int:
    """你是一个乘法计算器，可以帮我计算两个数的乘积，例如：计算1乘1等于几或计算1*1等于几
    Args:
        a: 输入a
        b: 输入b
    Return:
         返回 a*b 结果
    """
    print("hello success")
    return a*b

def test_function_call():
    from sparkai.core.callbacks import StdOutCallbackHandler
    messages = [{'role': 'user',
                 'content': "帮我算下 12乘以12"}]
    spark = ChatSparkLLM(
        spark_api_url=os.environ["SPARKAI_URL"],
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain=os.environ["SPARKAI_DOMAIN"],
        streaming=False,

    )
    function_definition =[convert_to_openai_function(multiply)]
    print(json.dumps(convert_to_openai_tool(multiply),ensure_ascii=False))
    messages = [ChatMessage(
        role="user",
        content=messages[0]['content']

    )]
    handler = ChunkPrintHandler()
    a = spark.generate([messages], callbacks=[handler],function_definition=function_definition)
    print(a)
    print(a.generations[0][0].text) 
    print(a.llm_output)

PASSED                                   [100%]{"type": "function", "function": {"name": "multiply", "description": "乘法函数，\nArgs:\n    a: 输入a\n    b: 输入b\nReturn:\n     返回 a*b 结果", "parameters": {"type": "object", "properties": {"b": {"type": "integer"}}, "required": ["a", "b"]}}}
generations=[[ChatGeneration(message=FunctionCallMessage(content='', function_call={'arguments': '{"a":12,"b":12}', 'name': 'multiply'}))]] llm_output={'token_usage': {'question_tokens': 9, 'prompt_tokens': 9, 'completion_tokens': 0, 'total_tokens': 9}} run=[RunInfo(run_id=UUID('95bf4e2e-6c90-41aa-9ddf-51b707d4d3c7'))]

generations=[[ChatGeneration(message=FunctionCallMessage(content='', function_call={'arguments': '{"a":12,"b":12}', 'name': 'operator_multiply'}))]] llm_output={'token_usage': {'question_tokens': 9, 'prompt_tokens': 9, 'completion_tokens': 0, 'total_tokens': 9}} run=[RunInfo(run_id=UUID('64bb65bd-948b-4354-bb72-e6847cc0a21b'))]

{'token_usage': {'question_tokens': 9, 'prompt_tokens': 9, 'completion_tokens': 0, 'total_tokens': 9}}

python -m sparkai.spark_proxy.main 

INFO:     Started server process [57295]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8008 (Press CTRL+C to quit)

### 省略其他代码
from sparkai.embedding.sparkai_base import SparkAiEmbeddingModel
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.vector_stores.chroma import ChromaVectorStore
from llama_index.core import StorageContext
import chromadb
import os
from sparkai.embedding.spark_embedding import SparkEmbeddingFunction
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, Settings
from sparkai.frameworks.llama_index import SparkAI

try:
    from dotenv import load_dotenv
except ImportError:
    raise RuntimeError(
        'Python environment for SPARK AI is not completely set up: required package "python-dotenv" is missing.') from None

load_dotenv()


def llama_query():
    chroma_client = chromadb.Client()
    chroma_collection = chroma_client.get_or_create_collection(name="spark")
    # define embedding function
    embed_model = SparkAiEmbeddingModel(spark_embedding_app_id=os.environ['SPARK_Embedding_APP_ID'],
                                        spark_embedding_api_key=os.environ['SPARK_Embedding_API_KEY'],
                                        spark_embedding_api_secret=os.environ['SPARK_Embedding_API_SECRET'],
                                        spark_embedding_domain=os.environ['SPARKAI_Embedding_DOMAIN'],
                                        qps=2)
    # define LLM Model
    sparkai = SparkAI(
        spark_api_url=os.environ["SPARKAI_URL"],
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain=os.environ["SPARKAI_DOMAIN"],
        streaming=False,
    )
    # load documents
    # Invoke-WebRequest -Uri 'https://raw.githubusercontent.com/run-llama/llama_index/main/docs/docs/examples/data/paul_graham/paul_graham_essay.txt' -OutFile 'data\paul_graham\paul_graham_essay.txt'
    documents = SimpleDirectoryReader("D:\data\paul_graham").load_data()
    # set up ChromaVectorStore and load in data
    vector_store = ChromaVectorStore(chroma_collection=chroma_collection)
    storage_context = StorageContext.from_defaults(vector_store=vector_store)
    index = VectorStoreIndex.from_documents(documents, storage_context=storage_context, embed_model=embed_model)
    # query
    query_engine = index.as_query_engine(llm=sparkai, similarity_top_k=2)
    response = query_engine.query("What did the author do growing up?")
    print(response)


if __name__ == "__main__":
    llama_query()

from sparkai.frameworks.autogen import SparkAI
import autogen
from autogen.agentchat.contrib.retrieve_assistant_agent import RetrieveAssistantAgent

spark_config = autogen.config_list_from_json(
    "sparkai_autogen.json",
    filter_dict={"model_client_cls": ["SparkAI"]},
)
llm_config = {
    "timeout": 600,
    "cache_seed": None,  # change the seed for different trials
    "config_list": spark_config,
    "temperature": 0,
}

# 1. create an RetrieveAssistantAgent instance named "assistant"
assistant = RetrieveAssistantAgent(
    name="assistant",
    system_message="You are a helpful assistant.",
    llm_config=llm_config
)
# 注册SparkAI类进入 agent
assistant.register_model_client(model_client_cls=SparkAI)

[
  {
    "api_key": "<spark_api_key>&<spark_api_secret>&<spark_app_id>",
    "base_url": "wss://spark-api.xf-yun.com/v3.5/chat",
    "model_client_cls": "SparkAI",
    "model": "generalv3.5", 
    "stream": true,
    "params": {
      "request_timeout": 61
    }
  }
]

import base64
image_content = base64.b64encode(open("spark_llama_index.png",'rb').read())

spark = ChatSparkLLM(
    spark_app_id=os.environ["SPARKAI_APP_ID"],
    spark_api_key=os.environ["SPARKAI_API_KEY"],
    spark_api_secret=os.environ["SPARKAI_API_SECRET"],
    spark_llm_domain="image",
    streaming=False,
    user_agent="test"

)
messages = [ImageChatMessage(
    role="user",
    content=image_content,
    content_type="image"
),ImageChatMessage(
    role="user",
    content="这是什么图",
    content_type="text"
)]
handler = ChunkPrintHandler()
a = spark.generate([messages], callbacks=[])

from sparkai.embedding.spark_embedding import Embeddingmodel, SparkEmbeddingFunction
import chromadb


def test_embedding():
    model = Embeddingmodel(
        spark_app_id="id",
        spark_api_key="key",
        spark_api_secret="secret",
        spark_domain="query",
    )
    # desc = {"messages":[{"content":"cc","role":"user"}]}
    desc = {"content": "cc", "role": "user"}
    # 调用embedding方法
    a = model.embedding(text=desc, kind='text')
    # print(len(a))
    print(a)


def test_chroma_embedding():
    chroma_client = chromadb.Client()
    sparkmodel = SparkEmbeddingFunction(
        spark_app_id="id",
        spark_api_key="key",
        spark_api_secret="secret",
        spark_domain="query",
    )
    a = sparkmodel(["This is a document", "This is another document"])
    # print(type(a))
    # print(a[0])
    # print(a[0][1])
    # 可以正确的生成embedding结果
    collection = chroma_client.get_or_create_collection(name="my_collection", embedding_function=sparkmodel)
    # 为什么是None
    collection.add(
        documents=["This is a document", "cc", "1122"],
        metadatas=[{"source": "my_source"}, {"source": "my_source"}, {"source": "my_source"}],
        ids=["id1", "id2", "id3"]
    )
    # print(collection.peek())  #显示前五条数据
    print(collection.count())  # 数据库中数据量
    results = collection.query(
        query_texts=["ac", 'documents'],
        n_results=2
    )
    print(results)  # 查询结果


if __name__ == "__main__":
    test_embedding()
    test_chroma_embedding()

spark = ChatSparkLLM(
    spark_api_url="wss://xingchen-api.cn-huabei-1.xf-yun.com/v1.1/chat",
    spark_app_id=os.environ["SPARKAI_APP_ID"],
    spark_api_key=os.environ["SPARKAI_API_KEY"],
    spark_api_secret=os.environ["SPARKAI_API_SECRET"],
    spark_llm_domain="xspark13b6k",
    streaming=True,
    max_tokens=1024,
    model_kwargs={"search_disable": False},
)

async def test_13b_lora():
    from sparkai.log.logger import logger
    logger.setLevel("debug")
    from sparkai.core.callbacks import StdOutCallbackHandler
    messages = [{'role': 'user',
                 'content': "卧槽"}]
    spark = ChatSparkLLM(
        spark_api_url="wss://xingchen-api.cn-huabei-1.xf-yun.com/v1.1/chat",
        spark_app_id=os.environ["SPARKAI_APP_ID"],
        spark_api_key=os.environ["SPARKAI_API_KEY"],
        spark_api_secret=os.environ["SPARKAI_API_SECRET"],
        spark_llm_domain="xspark13b6k",
        streaming=True,
        max_tokens=1024,
        model_kwargs={"patch_id": "210267877777408"},
    )
    messages = [
                ChatMessage(
                        role="user",
                        content=messages[0]['content']

    )]
    handler = AsyncChunkPrintHandler()
    a = spark.astream(messages, config={"callbacks": [handler]})
    async for message in a:
        print(message)

from sparkai.log.logger import logger
logger.setLevel("debug")