MetaGPT

English

中文

English

中文

Appearance

Tutorial Assistant: Generate technology tutorial ​

Role Introduction ​

Function Description ​

Generate a technical tutorial document based on a single sentence input, with support for custom languages.

Design Concept ​

The design approach involves using the LLM (Large Language Model) to initially generate the tutorial's outline. Then, the outline is divided into sections based on secondary headings. For each section, detailed content is generated according to the headings. Finally, the titles and content are concatenated. The use of sections addresses the limitation of long text in the LLM model.

Source Code ​

metagpt/roles/tutorial_assistant.py

Role Definition ​

  1. Define the role class, inheriting from the Role base class, and override the __init__ initialization method. The __init__ method must include the parameters name, profile, goal, and constraints. The first line of code uses super().__init__(name, profile, goal, constraints) to call the constructor of the parent class, initializing the Role. Use self.set_actions([WriteDirectory(language=language)]) to add initial action and states. Here, the write directory action is added initially. Additionally, custom parameters can be defined; here, the language parameter is added to support custom languages. Use self._set_react_mode(react_mode="by_order") to set the execution order of actions in set_actions to sequential.

    python
    class TutorialAssistant(Role):
    """Tutorial assistant, input one sentence to generate a tutorial document in markup format.

    Args:
        name: The name of the role.
        profile: The role profile description.
        goal: The goal of the role.
        constraints: Constraints or requirements for the role.
        language: The language in which the tutorial documents will be generated.
    """

    def __init__(
        self,
        name: str = "Stitch",
        profile: str = "Tutorial Assistant",
        goal: str = "Generate tutorial documents",
        constraints: str = "Strictly follow Markdown's syntax, with neat and standardized layout",
        language: str = "Chinese",
    ):
        super().__init__(name, profile, goal, constraints)
        self.set_actions([WriteDirectory(language=language)])
        self.topic = ""
        self.main_title = ""
        self.total_content = ""
        self.language = language
        self._set_react_mode(react_mode="by_order")
    class TutorialAssistant(Role):
    """Tutorial assistant, input one sentence to generate a tutorial document in markup format.

    Args:
        name: The name of the role.
        profile: The role profile description.
        goal: The goal of the role.
        constraints: Constraints or requirements for the role.
        language: The language in which the tutorial documents will be generated.
    """

    def __init__(
        self,
        name: str = "Stitch",
        profile: str = "Tutorial Assistant",
        goal: str = "Generate tutorial documents",
        constraints: str = "Strictly follow Markdown's syntax, with neat and standardized layout",
        language: str = "Chinese",
    ):
        super().__init__(name, profile, goal, constraints)
        self.set_actions([WriteDirectory(language=language)])
        self.topic = ""
        self.main_title = ""
        self.total_content = ""
        self.language = language
        self._set_react_mode(react_mode="by_order")

  2. Override the react method. Use await super().react() to call the react method of the Role base class. According to the react_mode="by_order" set in the __init__ method, execute each action in states in order. The purpose of overriding here is to perform final operations after completing all actions, i.e., writing the concatenated tutorial content into a markdown file.

    python
    async def react(self) -> Message:
    msg = await super().react()
    root_path = TUTORIAL_PATH / datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
    await File.write(root_path, f"{self.main_title}.md", self.total_content.encode('utf-8'))
    return msg
    async def react(self) -> Message:
    msg = await super().react()
    root_path = TUTORIAL_PATH / datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
    await File.write(root_path, f"{self.main_title}.md", self.total_content.encode('utf-8'))
    return msg

  3. Override the _act method. The _act method is responsible for executing the action. Use todo = self.rc.todo to get the next action to be executed from the context, and then execute the run method of the action. Here, it first obtains the tutorial directory structure through WriteDirectory, then chunks the directory, generates a WriteContent action for each chunk, and initializes the newly added action. Here, calling await super().react() again is to execute all the newly added WriteContent actions from the beginning. The result of each action is used to generate a message Message(content=resp, role=self.profile), which can be placed in the context memory self.rc.memory. This role does not need to be stored.

    python
    async def _act(self) -> Message:
    todo = self.rc.todo
    if type(todo) is WriteDirectory:
        msg = self.rc.memory.get(k=1)[0]
        self.topic = msg.content
        resp = await todo.run(topic=self.topic)
        logger.info(resp)
        await self._handle_directory(resp)
        return await super().react()
    resp = await todo.run(topic=self.topic)
    logger.info(resp)
    if self.total_content != "":
        self.total_content += "\n\n\n"
    self.total_content += resp
    return Message(content=resp, role=self.profile)

async def _handle_directory(self, titles: Dict) -> Message:
    """Handle the directories for the tutorial document.

    Args:
        titles: A dictionary containing the titles and directory structure,
                such as {"title": "xxx", "directory": [{"dir 1": ["sub dir 1", "sub dir 2"]}]}

    Returns:
        A message containing information about the directory.
    """
    self.main_title = titles.get("title")
    directory = f"{self.main_title}\n"
    self.total_content += f"# {self.main_title}"
    actions = list()
    for first_dir in titles.get("directory"):
        actions.append(WriteContent(language=self.language, directory=first_dir))
        key = list(first_dir.keys())[0]
        directory += f"- {key}\n"
        for second_dir in first_dir[key]:
            directory += f"  - {second_dir}\n"
    self.set_actions(actions)
    async def _act(self) -> Message:
    todo = self.rc.todo
    if type(todo) is WriteDirectory:
        msg = self.rc.memory.get(k=1)[0]
        self.topic = msg.content
        resp = await todo.run(topic=self.topic)
        logger.info(resp)
        await self._handle_directory(resp)
        return await super().react()
    resp = await todo.run(topic=self.topic)
    logger.info(resp)
    if self.total_content != "":
        self.total_content += "\n\n\n"
    self.total_content += resp
    return Message(content=resp, role=self.profile)

async def _handle_directory(self, titles: Dict) -> Message:
    """Handle the directories for the tutorial document.

    Args:
        titles: A dictionary containing the titles and directory structure,
                such as {"title": "xxx", "directory": [{"dir 1": ["sub dir 1", "sub dir 2"]}]}

    Returns:
        A message containing information about the directory.
    """
    self.main_title = titles.get("title")
    directory = f"{self.main_title}\n"
    self.total_content += f"# {self.main_title}"
    actions = list()
    for first_dir in titles.get("directory"):
        actions.append(WriteContent(language=self.language, directory=first_dir))
        key = list(first_dir.keys())[0]
        directory += f"- {key}\n"
        for second_dir in first_dir[key]:
            directory += f"  - {second_dir}\n"
    self.set_actions(actions)

Action Definition ​

  1. Define an action, where each action corresponds to a class object. Inherit from the Action base class and override the __init__ initialization method. The __init__ method includes the name parameter. The first line of code uses super().__init__(name, *args, **kwargs) to call the constructor of the parent class, initializing the action. Here, use args and kwargs to pass other parameters to the parent class constructor, such as context and llm.

    python
    #!/usr/bin/env python3
# _*_ coding: utf-8 _*_
"""
@Time    : 2023/9/4 15:40:40
@Author  : Stitch-z
@File    : tutorial_assistant.py
@Describe : Actions of the tutorial assistant, including writing directories and document content.
"""

from typing import Dict

from metagpt.actions import Action
from metagpt.prompts.tutorial_assistant import DIRECTORY_PROMPT, CONTENT_PROMPT
from metagpt.utils.common import OutputParser


class WriteDirectory(Action):
    """Action class for writing tutorial directories.

    Args:
        name: The name of the action.
        language: The language to output, default is "Chinese".
    """

    def __init__(self, name: str = "", language: str = "Chinese", *args, **kwargs):
        super().__init__(name, *args, **kwargs)
        self.language = language
    #!/usr/bin/env python3
# _*_ coding: utf-8 _*_
"""
@Time    : 2023/9/4 15:40:40
@Author  : Stitch-z
@File    : tutorial_assistant.py
@Describe : Actions of the tutorial assistant, including writing directories and document content.
"""

from typing import Dict

from metagpt.actions import Action
from metagpt.prompts.tutorial_assistant import DIRECTORY_PROMPT, CONTENT_PROMPT
from metagpt.utils.common import OutputParser


class WriteDirectory(Action):
    """Action class for writing tutorial directories.

    Args:
        name: The name of the action.
        language: The language to output, default is "Chinese".
    """

    def __init__(self, name: str = "", language: str = "Chinese", *args, **kwargs):
        super().__init__(name, *args, **kwargs)
        self.language = language

  2. Override the run method. The run method is the main function for action execution, using the self._aask(prompt=prompt) method to query the LLM model.

    python
    async def run(self, topic: str, *args, **kwargs) -> Dict:
    """Execute the action to generate a tutorial directory according to the topic.

    Args:
        topic: The tutorial topic.

    Returns:
        The tutorial directory information, including {"title": "xxx", "directory": [{"dir 1": ["sub dir 1", "sub dir 2"]}]}.
    """
    prompt = DIRECTORY_PROMPT.format(topic=topic, language=self.language)
    resp = await self._aask(prompt=prompt)
    return OutputParser.extract_struct(resp, dict)
    async def run(self, topic: str, *args, **kwargs) -> Dict:
    """Execute the action to generate a tutorial directory according to the topic.

    Args:
        topic: The tutorial topic.

    Returns:
        The tutorial directory information, including {"title": "xxx", "directory": [{"dir 1": ["sub dir 1", "sub dir 2"]}]}.
    """
    prompt = DIRECTORY_PROMPT.format(topic=topic, language=self.language)
    resp = await self._aask(prompt=prompt)
    return OutputParser.extract_struct(resp, dict)

  3. Other action writing is similar.

    python
    class WriteContent(Action):
    """Action class for writing tutorial content.

    Args:
        name: The name of the action.
        directory: The content to write.
        language: The language to output, default is "Chinese".
    """

    def __init__(self, name: str = "", directory: str = "", language: str = "Chinese", *args, **kwargs):
        super().__init__(name, *args, **kwargs)
        self.language = language
        self.directory = directory

    async def run(self, topic: str, *args, **kwargs) -> str:
        """Execute the action to write document content according to the directory and topic.

        Args:
            topic: The tutorial topic.

        Returns:
            The written tutorial content.
        """
        prompt = CONTENT_PROMPT.format(topic=topic, language=self.language, directory=self.directory)
        return await self._aask(prompt=prompt)
    class WriteContent(Action):
    """Action class for writing tutorial content.

    Args:
        name: The name of the action.
        directory: The content to write.
        language: The language to output, default is "Chinese".
    """

    def __init__(self, name: str = "", directory: str = "", language: str = "Chinese", *args, **kwargs):
        super().__init__(name, *args, **kwargs)
        self.language = language
        self.directory = directory

    async def run(self, topic: str, *args, **kwargs) -> str:
        """Execute the action to write document content according to the directory and topic.

        Args:
            topic: The tutorial topic.

        Returns:
            The written tutorial content.
        """
        prompt = CONTENT_PROMPT.format(topic=topic, language=self.language, directory=self.directory)
        return await self._aask(prompt=prompt)

Role Execution Results ​

Input Examples ​

  • MySQL Tutorial
  • Redis Tutorial
  • Hive Tutorial

Execution Command Examples ​

bash
python examples/write_tutorial.py
python examples/write_tutorial.py

Execution Results ​

The generated tutorial documents are located in the project's /data/tutorial_docx directory. Screenshots are provided below:

image

image

Note ​

This role currently does not support internet search capabilities. Content generation relies on data trained by the LLM large model.

Released under the MIT License.

Copyright © 2023-present alexanderwu