Merge pull request #1 from Torantulino/master

Merging up to latest on Torantulino/Auto-GPT Master
2026-01-19 14:04:29 +01:00 · 2023-04-04 10:33:38 -05:00
parent 6adef8ed7c 62dfd84599
commit fa0ec78441
10 changed files with 82 additions and 83 deletions
--- a/README.md
+++ b/README.md
@@ -20,10 +20,11 @@ Your support is greatly appreciated
 Development of this free, open-source project is made possible by all the <a href="https://github.com/Torantulino/Auto-GPT/graphs/contributors">contributors</a> and <a href="https://github.com/sponsors/Torantulino">sponsors</a>. If you'd like to sponsor this project and have your avatar or company logo appear below <a href="https://github.com/sponsors/Torantulino">click here</a>. 💖
 <p align="center">
 <p align="center">
-<a href="https://github.com/nocodeclarity"><img src="https://github.com/nocodeclarity.png" width="50px" alt="nocodeclarity" /></a>&nbsp;&nbsp;<a href="https://github.com/tjarmain"><img src="https://github.com/tjarmain.png" width="50px" alt="tjarmain" /></a>&nbsp;&nbsp;<a href="https://github.com/tekelsey"><img src="https://github.com/tekelsey.png" width="50px" alt="tekelsey" /></a>&nbsp;&nbsp;<a href="https://github.com/robinicus"><img src="https://github.com/robinicus.png" width="50px" alt="robinicus" /></a>&nbsp;&nbsp;<a href="https://github.com/digisomni"><img src="https://github.com/digisomni.png" width="50px" alt="digisomni" /></a>&nbsp;&nbsp;
+<a href="https://github.com/thepok"><img src="https://github.com/thepok.png" width="50px" alt="thepok" /></a>&nbsp;&nbsp;<a href="https://github.com/SpacingLily"><img src="https://github.com/SpacingLily.png" width="50px" alt="SpacingLily" /></a>&nbsp;&nbsp;<a href="https://github.com/m"><img src="https://github.com/m.png" width="50px" alt="m" /></a>&nbsp;&nbsp;<a href="https://github.com/zkonduit"><img src="https://github.com/zkonduit.png" width="50px" alt="zkonduit" /></a>&nbsp;&nbsp;<a href="https://github.com/maxxflyer"><img src="https://github.com/maxxflyer.png" width="50px" alt="maxxflyer" /></a>&nbsp;&nbsp;<a href="https://github.com/tekelsey"><img src="https://github.com/tekelsey.png" width="50px" alt="tekelsey" /></a>&nbsp;&nbsp;<a href="https://github.com/nocodeclarity"><img src="https://github.com/nocodeclarity.png" width="50px" alt="nocodeclarity" /></a>&nbsp;&nbsp;<a href="https://github.com/tjarmain"><img src="https://github.com/tjarmain.png" width="50px" alt="tjarmain" /></a>&nbsp;&nbsp;<a href="https://github.com/alexisneuhaus"><img src="https://github.com/alexisneuhaus.png" width="50px" alt="alexisneuhaus" /></a>&nbsp;&nbsp;<a href="https://github.com/jaumebalust"><img src="https://github.com/jaumebalust.png" width="50px" alt="jaumebalust" /></a>&nbsp;&nbsp;<a href="https://github.com/robinicus"><img src="https://github.com/robinicus.png" width="50px" alt="robinicus" /></a>&nbsp;&nbsp;<a href="https://github.com/digisomni"><img src="https://github.com/digisomni.png" width="50px" alt="digisomni" /></a>&nbsp;&nbsp;
 </p>
+
 <p align="center">
-<a href="https://github.com/alexisneuhaus"><img src="https://github.com/alexisneuhaus.png" width="30px" alt="alexisneuhaus" /></a>&nbsp;&nbsp;<a href="https://github.com/iokode"><img src="https://github.com/iokode.png" width="30px" alt="iokode" /></a>&nbsp;&nbsp;<a href="https://github.com/jaumebalust"><img src="https://github.com/jaumebalust.png" width="30px" alt="jaumebalust" /></a>&nbsp;&nbsp;
+<a href="https://github.com/alexisneuhaus"><img src="https://github.com/alexisneuhaus.png" width="30px" alt="alexisneuhaus" /></a>&nbsp;&nbsp;<a href="https://github.com/iokode"><img src="https://github.com/iokode.png" width="30px" alt="iokode" /></a>&nbsp;&nbsp;<a href="https://github.com/jaumebalust"><img src="https://github.com/jaumebalust.png" width="30px" alt="jaumebalust" /></a>&nbsp;&nbsp;<a href="https://github.com/nova-land"><img src="https://github.com/nova-land.png" width="30px" alt="nova-land" /></a>&nbsp;&nbsp;<a href="https://github.com/robinicus"><img src="https://github.com/robinicus.png" width="30px" alt="robinicus" /></a>&nbsp;&nbsp;<a href="https://github.com/Void-n-Null"><img src="https://github.com/Void-n-Null.png" width="30px" alt="Void-n-Null" /></a>&nbsp;&nbsp;<a href="https://github.com/ritesh24"><img src="https://github.com/ritesh24.png" width="30px" alt="ritesh24" /></a>&nbsp;&nbsp;<a href="https://github.com/merwanehamadi"><img src="https://github.com/merwanehamadi.png" width="30px" alt="merwanehamadi" /></a>&nbsp;&nbsp;<a href="https://github.com/raulmarindev"><img src="https://github.com/raulmarindev.png" width="30px" alt="raulmarindev" /></a>&nbsp;&nbsp;<a href="https://github.com/siduppal"><img src="https://github.com/siduppal.png" width="30px" alt="siduppal" /></a>&nbsp;&nbsp;<a href="https://github.com/goosecubedaddy"><img src="https://github.com/goosecubedaddy.png" width="30px" alt="goosecubedaddy" /></a>&nbsp;&nbsp;<a href="https://github.com/pleabargain"><img src="https://github.com/pleabargain.png" width="30px" alt="pleabargain" /></a>&nbsp;&nbsp;
 </p>


--- a/requirements-alternative.txt
+++ b/requirements-alternative.txt
@@ -1,14 +0,0 @@
-# I wasn't having any luck installing the requirements.txt file in Mac or Linux
-# But this seems to work.
-# The biggest difference is docker 5 instead of 6, because of this silliness:
-#
-# The conflict is caused by:
-#     The user requested requests>=2.26.0
-#     docker 6.0.1 depends on requests>=2.26.0
-#     googlesearch-python 1.1.0 depends on requests==2.25.1
-docker==5.0.3
-
-# I'd love to fix this in a cleaner way
-
-# Now go ahead and install the rest of what requirements.txt says:
-r requirements.txt
--- a/requirements.txt
+++ b/requirements.txt
@@ -7,7 +7,7 @@ pyyaml==6.0
 readability-lxml==0.8.1
 requests
 tiktoken==0.3.3
+gTTS==2.3.1
 docker
-googlesearch-python
+duckduckgo-search
 google-api-python-client #(https://developers.google.com/custom-search/v1/overview) 
-# Googlesearch python seems to be a bit cursed, anyone good at fixing thigns like this?
--- a/scripts/browse.py
+++ b/scripts/browse.py
@@ -74,30 +74,25 @@ def split_text(text, max_length=8192):
        yield "\n".join(current_chunk)


-def summarize_text(text, is_website=True):
-    if text == "":
+def create_message(chunk, question):
+    return {
+        "role": "user",
+        "content": f"\"\"\"{chunk}\"\"\" Using the above text, please answer the following question: \"{question}\" -- if the question cannot be answered using the text, please summarize the text."
+    }
+
+def summarize_text(text, question):
+    if not text:
        return "Error: No text to summarize"

-    print("Text length: " + str(len(text)) + " characters")
+    text_length = len(text)
+    print(f"Text length: {text_length} characters")
+
    summaries = []
    chunks = list(split_text(text))

    for i, chunk in enumerate(chunks):
-        print("Summarizing chunk " + str(i + 1) + " / " + str(len(chunks)))
-        if is_website:
-            messages = [
-                {
-                    "role": "user",
-                    "content": "Please summarize the following website text, do not describe the general website, but instead concisely extract the specific information this subpage contains.: " +
-                    chunk},
-            ]
-        else:
-            messages = [
-                {
-                    "role": "user",
-                    "content": "Please summarize the following text, focusing on extracting concise and specific information: " +
-                    chunk},
-            ]
+        print(f"Summarizing chunk {i + 1} / {len(chunks)}")
+        messages = [create_message(chunk, question)]

        summary = create_chat_completion(
            model=cfg.fast_llm_model,
@@ -105,25 +100,11 @@ def summarize_text(text, is_website=True):
            max_tokens=300,
        )
        summaries.append(summary)
-    print("Summarized " + str(len(chunks)) + " chunks.")
+
+    print(f"Summarized {len(chunks)} chunks.")

    combined_summary = "\n".join(summaries)
-
-    # Summarize the combined summary
-    if is_website:
-        messages = [
-            {
-                "role": "user",
-                "content": "Please summarize the following website text, do not describe the general website, but instead concisely extract the specific information this subpage contains.: " +
-                combined_summary},
-        ]
-    else:
-        messages = [
-            {
-                "role": "user",
-                "content": "Please summarize the following text, focusing on extracting concise and specific infomation: " +
-                combined_summary},
-        ]
+    messages = [create_message(combined_summary, question)]

    final_summary = create_chat_completion(
        model=cfg.fast_llm_model,
@@ -131,4 +112,4 @@ def summarize_text(text, is_website=True):
        max_tokens=300,
    )

-    return final_summary
+    return final_summary
--- a/scripts/commands.py
+++ b/scripts/commands.py
@@ -9,7 +9,7 @@ import ai_functions as ai
 from file_operations import read_file, write_to_file, append_to_file, delete_file
 from execute_code import execute_python_file
 from json_parser import fix_and_parse_json
-from googlesearch import search
+from duckduckgo_search import ddg
 from googleapiclient.discovery import build
 from googleapiclient.errors import HttpError

@@ -72,7 +72,7 @@ def execute_command(command_name, arguments):
        elif command_name == "delete_agent":
            return delete_agent(arguments["key"])
        elif command_name == "get_text_summary":
-            return get_text_summary(arguments["url"])
+            return get_text_summary(arguments["url"], arguments["question"])
        elif command_name == "get_hyperlinks":
            return get_hyperlinks(arguments["url"])
        elif command_name == "read_file":
@@ -84,7 +84,7 @@ def execute_command(command_name, arguments):
        elif command_name == "delete_file":
            return delete_file(arguments["file"])
        elif command_name == "browse_website":
-            return browse_website(arguments["url"])
+            return browse_website(arguments["url"], arguments["question"])
        # TODO: Change these to take in a file rather than pasted code, if
        # non-file is given, return instructions "Input should be a python
        # filepath, write your code to file and try again"
@@ -112,7 +112,7 @@ def get_datetime():

 def google_search(query, num_results=8):
    search_results = []
-    for j in search(query, num_results=num_results):
+    for j in ddg(query, max_results=num_results):
        search_results.append(j)

    return json.dumps(search_results, ensure_ascii=False, indent=4)
@@ -152,8 +152,8 @@ def google_official_search(query, num_results=8):
    # Return the list of search result URLs
    return search_results_links

-def browse_website(url):
-    summary = get_text_summary(url)
+def browse_website(url, question):
+    summary = get_text_summary(url, question)
    links = get_hyperlinks(url)

    # Limit links to 5
@@ -165,9 +165,9 @@ def browse_website(url):
    return result


-def get_text_summary(url):
+def get_text_summary(url, question):
    text = browse.scrape_text(url)
-    summary = browse.summarize_text(text)
+    summary = browse.summarize_text(text, question)
    return """ "Result" : """ + summary


--- a/scripts/config.py
+++ b/scripts/config.py
@@ -61,7 +61,7 @@ class Config(metaclass=Singleton):
        self.smart_token_limit = value

    def set_openai_api_key(self, value: str):
-        self.apiopenai_api_key_key = value
+        self.openai_api_key = value
    
    def set_elevenlabs_api_key(self, value: str):
        self.elevenlabs_api_key = value
--- a/scripts/data.py
+++ b/scripts/data.py
@@ -1,6 +1,6 @@
 import os
 from pathlib import Path
-
+SRC_DIR = Path(__file__).parent

 def load_prompt():
    try:
@@ -9,7 +9,7 @@ def load_prompt():
        data_dir = file_dir / "data"
        prompt_file = data_dir / "prompt.txt"
        # Load the promt from data/prompt.txt
-        with open(prompt_file, "r") as prompt_file:
+        with open(SRC_DIR/ "data/prompt.txt", "r") as prompt_file:
            prompt = prompt_file.read()

        return prompt
--- a/scripts/data/prompt.txt
+++ b/scripts/data/prompt.txt
@@ -1,7 +1,8 @@
 CONSTRAINTS:

-1. ~4000 word limit for memory. Your memory is short, so immidiately save important information to long term memory and code to files.
+1. ~4000 word limit for memory. Your memory is short, so immediately save important information to long term memory and code to files.
 2. No user assistance
+3. Exclusively use the commands listed in double quotes e.g. "command name"

 COMMANDS:

@@ -9,7 +10,7 @@ COMMANDS:
 2. Memory Add: "memory_add", args: "string": "<string>"
 3. Memory Delete: "memory_del", args: "key": "<key>"
 4. Memory Overwrite: "memory_ovr", args: "key": "<key>", "string": "<string>"
-5. Browse Website: "browse_website", args: "url": "<url>"
+5. Browse Website: "browse_website", args: "url": "<url>", "question": "<what_you_want_to_find_on_website>"
 6. Start GPT Agent: "start_agent",  args: "name": <name>, "task": "<short_task_desc>", "prompt": "<prompt>"
 7. Message GPT Agent: "message_agent", args: "key": "<key>", "message": "<message>"
 8. List GPT Agents: "list_agents", args: ""
@@ -34,9 +35,9 @@ RESOURCES:
 PERFORMANCE EVALUATION:

 1. Continuously review and analyze your actions to ensure you are performing to the best of your abilities. 
-2. Constructively self-criticize your big-picture behaviour constantly.
+2. Constructively self-criticize your big-picture behavior constantly.
 3. Reflect on past decisions and strategies to refine your approach.
-4. Every command has a cost, so be smart and efficent. Aim to complete tasks in the least number of steps.
+4. Every command has a cost, so be smart and efficient. Aim to complete tasks in the least number of steps.

 You should only respond in JSON format as described below

@@ -58,4 +59,4 @@ RESPONSE FORMAT:
    }
 }

-Ensure the response can be parsed by Python json.loads
+Ensure the response can be parsed by Python json.loads
--- a/scripts/main.py
+++ b/scripts/main.py
@@ -52,6 +52,14 @@ def print_assistant_thoughts(assistant_reply):
        # Parse and print Assistant response
        assistant_reply_json = fix_and_parse_json(assistant_reply)

+        # Check if assistant_reply_json is a string and attempt to parse it into a JSON object
+        if isinstance(assistant_reply_json, str):
+            try:
+                assistant_reply_json = json.loads(assistant_reply_json)
+            except json.JSONDecodeError as e:
+                print_to_console("Error: Invalid JSON\n", Fore.RED, assistant_reply)
+                assistant_reply_json = {}
+
        assistant_thoughts_reasoning = None
        assistant_thoughts_plan = None
        assistant_thoughts_speak = None
@@ -303,7 +311,7 @@ while True:
            Fore.CYAN,
            f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL}  ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")
        print(
-            "Enter 'y' to authorise command or 'n' to exit program...",
+            f"Enter 'y' to authorise command or 'n' to exit program, or enter feedback for {ai_name}...",
            flush=True)
        while True:
            console_input = input(Fore.MAGENTA + "Input:" + Style.RESET_ALL)
@@ -314,16 +322,18 @@ while True:
                user_input = "EXIT"
                break
            else:
-                continue
+                user_input = console_input
+                command_name = "human_feedback"
+                break

-        if user_input != "GENERATE NEXT COMMAND JSON":
-            print("Exiting...", flush=True)
-            break
-
-        print_to_console(
+        if user_input == "GENERATE NEXT COMMAND JSON":
+            print_to_console(
            "-=-=-=-=-=-=-= COMMAND AUTHORISED BY USER -=-=-=-=-=-=-=",
            Fore.MAGENTA,
            "")
+        elif user_input == "EXIT":
+            print("Exiting...", flush=True)
+            break
    else:
        # Print command
        print_to_console(
@@ -332,10 +342,12 @@ while True:
            f"COMMAND = {Fore.CYAN}{command_name}{Style.RESET_ALL}  ARGUMENTS = {Fore.CYAN}{arguments}{Style.RESET_ALL}")

    # Execute command
-    if command_name.lower() != "error":
-        result = f"Command {command_name} returned: {cmd.execute_command(command_name, arguments)}"
-    else:
+    if command_name.lower() == "error":
        result = f"Command {command_name} threw the following error: " + arguments
+    elif command_name == "human_feedback":
+        result = f"Human feedback: {user_input}"
+    else:
+        result = f"Command {command_name} returned: {cmd.execute_command(command_name, arguments)}"

    # Check if there's a result from the command append it to the message
    # history
@@ -347,3 +359,4 @@ while True:
            chat.create_chat_message(
                "system", "Unable to execute command"))
        print_to_console("SYSTEM: ", Fore.YELLOW, "Unable to execute command")
+
--- a/scripts/speak.py
+++ b/scripts/speak.py
@@ -3,6 +3,8 @@ from playsound import playsound
 import requests
 from config import Config
 cfg = Config()
+import gtts
+

 # TODO: Nicer names for these ids
 voices = ["ErXwobaYiN019PkySvjV", "EXAVITQu4vr4xnSDxMaL"]
@@ -12,10 +14,9 @@ tts_headers = {
    "xi-api-key": cfg.elevenlabs_api_key
 }

-def say_text(text, voice_index=0):
+def eleven_labs_speech(text, voice_index=0):
    tts_url = "https://api.elevenlabs.io/v1/text-to-speech/{voice_id}".format(
        voice_id=voices[voice_index])
-
    formatted_message = {"text": text}
    response = requests.post(
        tts_url, headers=tts_headers, json=formatted_message)
@@ -24,8 +25,24 @@ def say_text(text, voice_index=0):
        with open("speech.mpeg", "wb") as f:
            f.write(response.content)
        playsound("speech.mpeg")
-        # Delete audio file
        os.remove("speech.mpeg")
+        return True
    else:
        print("Request failed with status code:", response.status_code)
        print("Response content:", response.content)
+        return False
+
+def gtts_speech(text):
+    tts = gtts.gTTS(text)
+    tts.save("speech.mp3")
+    playsound("speech.mp3")
+    os.remove("speech.mp3")
+
+def say_text(text, voice_index=0):
+    if not cfg.elevenlabs_api_key:
+        gtts_speech(text)
+    else:
+        success = eleven_labs_speech(text)
+        if not success:
+            gtts_speech(text)
+