Skip to main content

test package for yorg

Project description

YORG Test Package

this is a test package

test version

how to use

  • Set up test version package
pip install -i https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple yorg-test
  • Set up openapikey (If you are in China, please set up a proxy to ensure that you can connect to openai.)
import os
#os.environ['http_proxy'] = 'http://127.0.0.1:10809'  # 这里设置自己的代理端口号
#os.environ['https_proxy'] = 'http://127.0.0.1:10809'  # 这里设置自己的代理端口号
os.environ['OPENAI_CHAT_API_KEY'] = 'sk-br3j7Gxxxxxxxxvt8r'
  • set up yaml file We have some tools built in eg.code_interpreter,swe_tool

tools.yaml

YORG: 0.0.1
info:
  title: yorg_tools_document
  description: yorg tool define document.
  version: 'v1'
tools:
  code_interpreter:
    name: code_interpreter
    entity_name: code_interpreter
    summary: Run the code through code_interpreter and return the result of the code run. If your query is about math, computer science, or data science, you can use this tool to get the result of the code run.
    parameters:
      - name: code
        description: code text that requires code_interpreter to run
        required: true
        parameter_schema:
          type: string
    responses:
      success:
        description: OK
        content:
          result:
            type: string
            description: the result of the code run
  example_stateful_tool:
    name: example_stateful_tool
    entity_name: example_stateful_tool
    summary: This tool is an example of a stateful tool. It will get two number from user input and return the sum of the two numbers.
    parameters: []
    responses: {}  
  swe_tool:
    name: sew_tool
    entity_name: swe_tool
    summary: SoftWare Engineer Agent(swe_tool) specializes in working with code files.
    parameters: []
    responses: {}  

Tools are categorized into stateful tools and function tools, Function tools can describe parameters and return values directly in tools.yaml Functions can be registered using decorators

from yorg_test.core.assistant.tools.tools import register_function_tool
@register_function_tool
def code_test(code: str):
    return {
        "type": "success",
        "content": {
            "result": "Hello, World!",
        },
    }

If it's a stateful tool you need to write an additional yaml file with a stateful description

start_stage: "init"
finish_stage: "finish"
all_stages:
  init:
    name: "init"
    next_stage_entry: 
      stage_1:
        - name: x
          required: true
          parameter_schema:
            type: number
            description: "input value x"
    need_llm_generate_parameters: false
    need_llm_generate_response: false
  stage_1:
    name: "stage_1"
    next_stage_entry: 
      stage_2:
        - name: y 
          required: true
          parameter_schema:
            type: number
            description: "input value y"
    need_llm_generate_parameters: false 
    need_llm_generate_response: false
  stage_2:
    name: "stage_2"
    next_stage_entry: 
      stage_3: []
    need_llm_generate_parameters: false
    need_llm_generate_response: false
  stage_3:
    name: "stage_3"
    next_stage_entry: 
      finish: []
    need_llm_generate_parameters: false
    need_llm_generate_response: false
  finish:
    name: "finish"
    next_stage_entry: {}
    need_llm_generate_parameters: false
    need_llm_generate_response: false

Stateful tools can also be registered using decorators. The yaml file is registered in init.

from yorg_test.core.assistant.tools.tools import register_stateful_tool
from yorg_test.core.assistant.tools.stateful_tool_entity import StatefulToolEntity
@register_stateful_tool
class ExampleStatefulToolEntity(StatefulToolEntity):
    """
    This example tool entity is stateful, and it has 3 inner stages.

    stage1: take integer x as input
    stage2: take integer y as input
    stage3: no input, return x + y
    """
    def __init__(self):
        super().__init__("example_stateful_tool.yaml")
    def _call(self, **kwargs):
        if "goto" not in kwargs:
            if self.current_stage.name == self.config.start_stage:
                return {
                    "type": "success",
                    "content": {"message": "stateful tool is started"},
                }
            else:
                return {
                    "type": "error",
                    "content": {"message": "please provide `goto` parameter"},
                }
        request_next_stage = kwargs["goto"]
        if request_next_stage not in self.config.all_stages:
            return {
                "type": "error",
                "content": {"message": f"stage {request_next_stage} not found"},
            }
        self.current_stage = self.config.all_stages[request_next_stage]
        match self.current_stage.name:
            case "stage_1":
                return self._stage1(kwargs["x"])
            case "stage_2":
                return self._stage2(kwargs["y"])
            case "stage_3":
                return self._stage3()
            case self.config.finish_stage:
                return self._finish()
            case _:
                return {
                    "type": "error",
                    "content": {
                        "message": f"stage {self.current_stage.name} not found"
                    },
                }
    def _stage1(self, x: int):
        self.x = x
        return {"type": "success", "content": {"message": "stage1 done"}}
    def _stage2(self, y: int):
        self.y = y
        return {"type": "success", "content": {"message": "stage2 done"}}
    def _stage3(self):
        return {"type": "success", "content": {"result": self.x + self.y}}
    def _finish(self):
        return {"type": "success", "content": {"message": "stateful tool is finished"}}
  • run example
import yorg_test
threads = yorg_test.Threads.create('tools.yaml')
assistant = yorg_test.Assistants.create(name="Test Assistant", model="gpt-4-1106-preview", instructions="Use swe tool auto fix code files", tools=[{'type':'swe_tool'}])
print(assistant.id)
# 运行 Threads 对象
result = threads.run(assistant.id, "Use SoftWare Engineer Agent swe tool auto fix code files.")
print(result)

result = threads.run(assistant.id, "the repo url is https://github.com/YORG-AI/Open-Assistant",goto="stage_1")
print(result)

result = threads.run(assistant.id, "add helloworld feature to readme",  goto="stage_2")
print(result)

result = threads.run(assistant.id, "focus_files_name_list = [README.md]", goto="stage_3")
print(result)

result = threads.run(assistant.id, "action=3", goto="stage_4")
print(result)

result = threads.run(assistant.id, "", goto="stage_5")
print(result)

result = threads.run(assistant.id, "action=0,action_idx=0", goto="stage_6")
print(result)

result = threads.run(assistant.id, "", goto="finish")
print(result)

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fireact-0.0.1.tar.gz (50.3 kB view details)

Uploaded Source

Built Distribution

fireact-0.0.1-py3-none-any.whl (74.1 kB view details)

Uploaded Python 3

File details

Details for the file fireact-0.0.1.tar.gz.

File metadata

  • Download URL: fireact-0.0.1.tar.gz
  • Upload date:
  • Size: 50.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.12

File hashes

Hashes for fireact-0.0.1.tar.gz
Algorithm Hash digest
SHA256 3ffc98d722aedbd549eaf405d217cc0906d021b9bcceb49631c1858cc3ebfc5e
MD5 9a191b456f6cd524e2c310bb3a38832b
BLAKE2b-256 cc456fdff83995b7c22084e12995eb678073f5d67ce93bfe305ac06af1c7df51

See more details on using hashes here.

File details

Details for the file fireact-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: fireact-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 74.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.10.12

File hashes

Hashes for fireact-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 7cfddc5639ed615556423efc7e4dfb12250126f591a5f49d09f048ac048abcdc
MD5 8b6dfde1e206890c57e11506542f931f
BLAKE2b-256 102f32c42c4b563fe61acc03417c2f078fe6d930e408c7d51673aa5dad9090cf

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page