Merge pull request emrgnt-cmplxty#25 from maks-ivanov/babyagi-archite…

…cture

Babyagi architecture

config.py

@@ -10,3 +10,4 @@
 YOUR_TABLE_NAME = os.getenv("TABLE_NAME", "")
 OBJECTIVE = os.getenv("OBJECTIVE", "")
 DO_RETRY = bool(os.getenv("DO_RETRY", 1))
+PLANNER_AGENT_OUTPUT_STRING = "Planner Agent Output: "

custom_tools.py

@@ -5,9 +5,10 @@
 from github.Issue import Issue
 from github.PullRequest import PullRequest
 from github.Repository import Repository
-from langchain.agents import Tool
+from langchain.agents import Tool, tool
+import requests
 
-from utils import PassThroughBuffer
+from utils import PassThroughBuffer, remove_html_tags
 
 
 class GitToolBuilder:
@@ -59,50 +60,75 @@ def create_new_branch(self, branch_name: str) -> str:
         Creates and checks out a new branch in the specified repository. The only input is the branch name. For exmpale: "my-branch". Before creating a new branch, make sure to pick a name that is not taken."
         """
         # Create branch
-        self.pygit_repo.git.branch(branch_name)
-        # Checkout branch
-        self.pygit_repo.git.checkout(branch_name)
+        try:
+            self.pygit_repo.git.branch(branch_name)
+            # Checkout branch
+            self.pygit_repo.git.checkout(branch_name)
+            return f"Created and checked out branch {branch_name} in {self.github_repo.name} repository."
 
-        return f"Created and checked out branch {branch_name} in {self.github_repo.name} repository."
+        except Exception as e:
+            return f"Error: {e}"
 
     def checkout_branch(self, branch_name: str) -> str:
         """
         Creates and checks out a new branch in the specified repository. The only input is the branch name. For exmpale: "my-branch"
         """
         # Checkout branch
-        self.pygit_repo.git.checkout(branch_name)
-        self.pygit_repo.git.pull()
-        return f"Checked out an existing branch {branch_name} in {self.github_repo.name} repository."
+        try:
+            self.pygit_repo.git.checkout(branch_name)
+            self.pygit_repo.git.pull()
+            return f"Checked out an existing branch {branch_name} in {self.github_repo.name} repository."
+        except Exception as e:
+            return f"Error: {e}"
 
     def commit_to_git(self, file_names: str) -> str:
         """
         Takes a string of comma-separated file names and commits them to git. For example "file1.py,file2.py"
         """
-        file_names = file_names.split(",")
-        for file_name in file_names:
-            self.pygit_repo.git.add(file_name)
+        try:
+            file_names = file_names.split(",")
+            for file_name in file_names:
+                self.pygit_repo.git.add(file_name)
 
-        self.pygit_repo.git.commit(m="Committing changes")
-        self.pygit_repo.git.push(
-            "--set-upstream", "origin", self.pygit_repo.git.branch("--show-current")
-        )
-        return f"Committed {file_names} to {self.github_repo.name} repository."
+            self.pygit_repo.git.commit(m="Committing changes")
+            self.pygit_repo.git.push(
+                "--set-upstream", "origin", self.pygit_repo.git.branch("--show-current")
+            )
+            return f"Committed {file_names} to {self.github_repo.name} repository."
+        except Exception as e:
+            return f"Error: {e}"
 
     def create_pull_request(self, body) -> str:
         """
         Creates a pull request in the specified repository.
         """
         # get current branch name
-        assert type(self.work_item) == Issue
-        current_branch = self.pygit_repo.git.branch("--show-current")
-        title = "Fix for issue #" + str(self.work_item.number)
-        pull: github.PullRequest.PullRequest = self.github_repo.create_pull(
-            head=current_branch,
-            base=self.github_repo.default_branch,
-            issue=self.work_item,
-        )
-        if self.logger:
-            pull.create_issue_comment(self.logger.saved_output)
-        return (
-            f"Created pull request for  {title} in {self.github_repo.name} repository."
-        )
+        try:
+            assert type(self.work_item) == Issue
+            current_branch = self.pygit_repo.git.branch("--show-current")
+            title = "Fix for issue #" + str(self.work_item.number)
+            pull: github.PullRequest.PullRequest = self.github_repo.create_pull(
+                head=current_branch,
+                base=self.github_repo.default_branch_name,
+                issue=self.work_item,
+            )
+            if self.logger:
+                pull.create_issue_comment(self.logger.saved_output)
+            return f"Created pull request for  {title} in {self.github_repo.name} repository."
+        except Exception as e:
+            return f"Error: {e}"
+
+
+@tool
+def requests_get_clean(url: str) -> str:
+    """
+    Sends a get request to a specified URL and returns clean text in the response.
+    """
+    response = requests.get(url)
+    try:
+        if response.status_code == 200:
+            return remove_html_tags(response.text)
+        else:
+            raise Exception(f"Error: {response.status_code} {response.text}")
+    except Exception as e:
+        return f"Error: {e}"

main.py

@@ -3,13 +3,13 @@
 import traceback
 
 from git import Repo
-from github.Issue import Issue
-from github.PullRequest import PullRequest
 from langchain.agents import initialize_agent, load_tools, AgentType
 from langchain.chat_models import ChatOpenAI
+from langchain.memory import ConversationBufferMemory
 
 from config import *
-from custom_tools import GitToolBuilder
+from custom_tools import GitToolBuilder, requests_get_clean
+from prompts import make_planning_task, make_execution_task
 from utils import login_github, list_repositories, choose_work_item, PassThroughBuffer
 
 # Log into GitHub
@@ -29,74 +29,88 @@
 # create a repo object which represents the repository we are inside of
 pygit_repo = Repo(os.getcwd())
 
+default_branch_name = "babyagi-architecture"
 # reset to default branch if necessary
-if pygit_repo.active_branch.name != "main":
-    pygit_repo.git.checkout("main")
+if pygit_repo.active_branch.name != default_branch_name:
+    pygit_repo.git.checkout(default_branch_name)
 
-# checkout default branch and pull
-pygit_repo.git.checkout("main")
-pygit_repo.git.pull()
 
 work_item = choose_work_item(github_repo)
 
 
-llm = ChatOpenAI(temperature=0.01, model="gpt-4")
+llm = ChatOpenAI(temperature=0.01, model="gpt-3.5-turbo")
 # llm1 = OpenAI(temperature=0)
 pass_through_buffer = PassThroughBuffer(sys.stdout)
 assert pass_through_buffer.saved_output == ""
 sys.stdout = pass_through_buffer
-tools = load_tools(["python_repl", "terminal", "serpapi", "requests_get"], llm=llm)
-tools += GitToolBuilder(github_repo, pygit_repo, work_item).build_tools()
-
-exec_agent = initialize_agent(
-    tools, llm, agent=AgentType.CHAT_ZERO_SHOT_REACT_DESCRIPTION, verbose=True
+base_tools = load_tools(["python_repl", "terminal", "human"], llm=llm)
+base_tools += [requests_get_clean]
+exec_tools = (
+    base_tools + GitToolBuilder(github_repo, pygit_repo, work_item).build_tools()
 )
 
+memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
 
-task = (
-    f"You are a GPT-4-powered software engineer agent."
-    f" You are built with langchain, a framework for building language-based agents. "
-    f" You can read about it here: https://python.langchain.com/en/latest/modules/agents.html"
-    f" Your task is to contribute clean, high-quality code to the given codebase."
-    f" You are working in {os.getcwd()} on {github_repo.name} repository."
-    f" Feel free to look around this repo. You can browse the internet for documentation and examples."
-    f" You may need to create, modify, or delete one or more files in this repository."
+plan_agent = initialize_agent(
+    base_tools,
+    llm,
+    agent=AgentType.CHAT_CONVERSATIONAL_REACT_DESCRIPTION,
+    verbose=True,
+    memory=memory,
 )
 
-if type(work_item) == Issue:
-    task += f" You must create a new branch, implement a solution and submit a pull request to address the following issue: \n\n Title: {work_item.title}.\n\n Body: {work_item.body}."
-if type(work_item) == PullRequest:
-    task += (
-        f" You must checkout the branch, understand the following pull request feedback make a commit with changes to address it:"
-        f" \n\n Title: {work_item.title}.\n\n Body: {work_item.body}. \n\n Files: {[f.filename for f in work_item.get_files()]}:"
-    )
-
-
-comments = work_item.get_comments()
-
-if list(comments):
-    task += f" Comments:"
-    for comment in comments:
-        task += f" {comment.body}"
-
-
-task += (
-    f"\n\nUseful tips: Don't use nano, vim or other text editors, but rather modify files directly either via python or terminal. "
-    f" Before creating a new branch, make sure to pick a name that is not taken."
+exec_agent = initialize_agent(
+    exec_tools, llm, agent=AgentType.CHAT_ZERO_SHOT_REACT_DESCRIPTION, verbose=True
 )
 
-try:
-    print("Task:", task)
-    x = exec_agent.run(task)
-except ValueError as e:
-    if DO_RETRY:
-        tb = traceback.format_exc()
-        task += f" This is your second attempt. During the previous attempt, you crashed with the following sequence: <run>{pass_through_buffer.saved_output}</run> Let's try again, avoiding previous mistakes."
-        pass_through_buffer.saved_output = ""
-        print("Failed to complete task with following error:", e, tb)
-        print("New task:", task)
-        print("Retrying...")
-        x = exec_agent.run(task)
-finally:
-    sys.stdout = pass_through_buffer.original_buffer
-    pygit_repo.git.checkout("main")
+# check if instrutions are already attached to the issue
+instructions = [
+    c.body
+    for c in work_item.get_comments()
+    if c.body.startswith(PLANNER_AGENT_OUTPUT_STRING)
+]
+if instructions:
+    instructions = instructions.pop()
+    instructions.replace(PLANNER_AGENT_OUTPUT_STRING, "")
+    print("Found instructions:", instructions)
+
+# ask user if they want to run planner agent, default is yes if no instructions
+
+do_plan = input("Do you want to run the planning agent? (y/n)")
+
+if do_plan == "y":
+    plan_task = make_planning_task(work_item, exec_tools, github_repo.name)
+    print("Planning task:", plan_task)
+    approved = False
+    while not approved:
+        instructions = plan_agent.run(plan_task)
+        print("Created new Instructions:", instructions)
+        feedback = input(
+            "Do you approve? If approved, type 'y'. If not approved, type why so the agent can try again: "
+        )
+        approved = feedback == "y"
+        plan_task = feedback
+
+    # save instructions to issue
+    work_item.create_comment(PLANNER_AGENT_OUTPUT_STRING + instructions)
+
+
+# ask user if they want to run exec agent
+do_exec = input("Do you want to run the execution agent? (y/n)")
+if do_exec == "y":
+    exec_task = make_execution_task(work_item, instructions, github_repo.name)
+    print("Execution task:", exec_task)
+    try:
+        exec_agent.run(exec_task)
+    except ValueError as e:
+        if DO_RETRY:
+            tb = traceback.format_exc()
+            exec_task += f" This is your second attempt. During the previous attempt, you crashed with the following sequence: <run>{pass_through_buffer.saved_output}</run> Let's try again, avoiding previous mistakes."
+            pass_through_buffer.saved_output = ""
+            print("Failed to complete execution task")
+            print("New task:", exec_task)
+            print("Retrying...")
+            exec_agent.run(exec_task)
+    finally:
+        sys.stdout = pass_through_buffer.original_buffer
+        pygit_repo.git.checkout(default_branch_name)

prompts.py

@@ -0,0 +1,74 @@
+import os
+from typing import List, Union
+
+from github.Issue import Issue
+from github.PullRequest import PullRequest
+from langchain.tools import BaseTool
+
+from config import PLANNER_AGENT_OUTPUT_STRING
+
+
+def make_planning_task(
+    work_item: Union[Issue, PullRequest],
+    exec_tools: List[BaseTool],
+    github_repo_name: str,
+):
+    pr_or_issue_str = (
+        " submit a pull request" if isinstance(work_item, Issue) else " make a commit "
+    )
+    return (
+        f"You are a GPT-4 software engineering lead agent."
+        f" You plan out software engineering work for developer agents."
+        f" You are built with langchain, a framework for building language-based agents."
+        f" You can read about it here: https://python.langchain.com/en/latest/modules/agents.html"
+        f" Assume you have the code locally on your machine."  # todo
+        f" You are working in {os.getcwd()} on {github_repo_name} repository."
+        f" Your task is to thoroughly understand the following work item and "
+        f" create simple and thorough step-by-step instructions for a developer to implement the solution."
+        f" You should not make the changes yourself, but rather output instructions for a developer to make the changes."
+        f" \n\nTitle: {work_item.title}"
+        f" \n\nBody: {work_item.body}"
+        f" \n\nComments: {[c.body for c in work_item.get_comments() if not c.body.startswith(PLANNER_AGENT_OUTPUT_STRING)]}"
+        f" \n\n The developer will use your instructions to make changes to the repository and"
+        f" {pr_or_issue_str} with working, clean, and documented code."
+        f" Your developer is also a GPT-4-powered agent, so keep that in mind when creating instructions."
+        f" You should acquire an excellent understanding of the current state of the repository and the code within it."
+        f" You should also look up documentation on the internet whenever necessary."
+        f" Your instructions should be clear and concise, and should not contain any typos or grammatical errors."
+        f" You should tell the developer which files to create/modify/delete, and what to write in them."
+        f" You should also tell the developer which external libraries to use, if any."
+        f" For external libraries, you should provide a link to the documentation."
+        f" Make sure not to regress any existing functionality."
+        f" The developer agent will have access to the following tools: {[(tool.name, tool.description) for tool in exec_tools]}, so keep that in mind when creating instructions."
+        f" Begin."
+    )
+
+
+def make_execution_task(
+    work_item: Union[Issue, PullRequest],
+    solution_instructions: str,
+    github_repo_name: str,
+):
+    pr_or_issue_str = (
+        " create a pull request with your changes."
+        if isinstance(work_item, Issue)
+        else f" make a commit with your changes to the appropriate branch."
+    )
+    return (
+        f"You are a GPT-4-powered coding agent."
+        f" You are built with langchain, a framework for building language-based agents. "
+        f" You can read about it here: https://python.langchain.com/en/latest/modules/agents.html"
+        f" Your task is to contribute clean, high-quality code to the given codebase."
+        f" You are working in {os.getcwd()} on {github_repo_name} repository."
+        f" You are working on the following work item: "
+        f"\n\nTitle: {work_item.title};"
+        f"\n\nBody: {work_item.body};"
+        f"\n\nComments: {[c.body for c in work_item.get_comments() if not c.body.startswith(PLANNER_AGENT_OUTPUT_STRING)]};"
+        f"\n\n A planning agent has created the following step-by-step instructions for you: <instructions>{solution_instructions}</instructions>"
+        f" Execute the instructions thoroughly and"
+        f" {pr_or_issue_str}"
+        f" Some of the instructions may be high level, so it's up to you to understand what exactly needs to be done."
+        f" Make sure not to regress any existing functionality."
+        f"\n\nUseful tips: Do NOT use nano, vim or other text editors, but rather modify files directly either via python or terminal. "
+        f" Important: when following git-create-branch instructions, make sure to use a branch name that's not taken. "
+    )

requirements.txt

@@ -1,6 +1,11 @@
 GitPython==3.1.31
-langchain==0.0.134
+langchain~=0.0.137
 openai==0.25.0
 pinecone==0.1.0
 PyGithub==1.58.1
 python-dotenv==1.0.0
+
+regex~=2022.10.31
+typing~=3.10.0.0
+beautifulsoup4~=4.12.0
+requests~=2.28.1

utils.py

@@ -1,7 +1,9 @@
 # New functions
 from typing import Union
 import github
+from bs4 import BeautifulSoup
 from github import Github
+import regex as re
 
 
 def login_github(token):
@@ -56,3 +58,11 @@ def write(self, message):
 
     def __getattr__(self, attr):
         return getattr(self.original_buffer, attr)
+
+
+def remove_html_tags(text) -> str:
+    clean = re.compile("<.*?>")
+    soup = BeautifulSoup(text, "html.parser")
+    raw_text = soup.get_text(strip=True, separator="\n")
+    clean_text = re.sub(clean, "", raw_text)
+    return clean_text