wcgw 0.0.5__tar.gz → 0.0.8__tar.gz

wcgw-0.0.8/.python-version

@@ -0,0 +1 @@
+3.12

{wcgw-0.0.5 → wcgw-0.0.8}/PKG-INFO

@@ -1,10 +1,11 @@
 Metadata-Version: 2.3
 Name: wcgw
-Version: 0.0.5
+Version: 0.0.8
 Summary: What could go wrong giving full shell access to chatgpt?
 Project-URL: Homepage, https://github.com/rusiaaman/wcgw
 Author-email: Aman Rusia <gapypi@arcfu.com>
-Requires-Python: >=3.7
+Requires-Python: <3.13,>=3.8
+Requires-Dist: anthropic>=0.36.2
 Requires-Dist: fastapi>=0.115.0
 Requires-Dist: mypy>=1.11.2
 Requires-Dist: openai>=1.46.0
@@ -22,10 +23,18 @@ Requires-Dist: uvicorn>=0.31.0
 Requires-Dist: websockets>=13.1
 Description-Content-Type: text/markdown
 
-# Giving full shell access to Chatgpt web (What could go wrong?)
-Steps: 
-1. First run the client in any directory of choice
-2. Use this custom gpt `https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access` to let it interact with your shell.
+# Shell access to chatgpt.com
+
+### 🚀 Highlights
+- ⚡ **Full Shell Access**: No restrictions, complete control.
+- ⚡ **Create, Execute, Iterate**: Seamless workflow for development and execution.
+- ⚡ **Interactive Command Handling**: Supports interactive commands with ease.
+
+
+###  🪜 Steps: 
+1. Run the [cli client](https://github.com/rusiaaman/wcgw?tab=readme-ov-file#client) in any directory of choice.
+2. Share the generated id with the GPT: `https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access`
+3. The custom GPT can now run any command on your cli
 
 ## Client
 
@@ -49,7 +58,7 @@ https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access
 
 Add user id the client generated to the first message along with the instructions.
 
-# How does it work?
+# How it works
 Your commands are relayed through a server I've hosted at https://wcgw.arcfu.com. The code for that is at `src/relay/serve.py`. 
 
 The user id that you share with chatgpt is added in the request it sents to the relay server which holds a websocket with the terminal client.

{wcgw-0.0.5 → wcgw-0.0.8}/README.md

@@ -1,7 +1,15 @@
-# Giving full shell access to Chatgpt web (What could go wrong?)
-Steps: 
-1. First run the client in any directory of choice
-2. Use this custom gpt `https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access` to let it interact with your shell.
+# Shell access to chatgpt.com
+
+### 🚀 Highlights
+- ⚡ **Full Shell Access**: No restrictions, complete control.
+- ⚡ **Create, Execute, Iterate**: Seamless workflow for development and execution.
+- ⚡ **Interactive Command Handling**: Supports interactive commands with ease.
+
+
+###  🪜 Steps: 
+1. Run the [cli client](https://github.com/rusiaaman/wcgw?tab=readme-ov-file#client) in any directory of choice.
+2. Share the generated id with the GPT: `https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access`
+3. The custom GPT can now run any command on your cli
 
 ## Client
 
@@ -25,7 +33,7 @@ https://chatgpt.com/g/g-Us0AAXkRh-wcgw-giving-shell-access
 
 Add user id the client generated to the first message along with the instructions.
 
-# How does it work?
+# How it works
 Your commands are relayed through a server I've hosted at https://wcgw.arcfu.com. The code for that is at `src/relay/serve.py`. 
 
 The user id that you share with chatgpt is added in the request it sents to the relay server which holds a websocket with the terminal client.

{wcgw-0.0.5 → wcgw-0.0.8}/pyproject.toml

@@ -1,10 +1,10 @@
 [project]
 authors = [{ name = "Aman Rusia", email = "gapypi@arcfu.com" }]
 name = "wcgw"
-version = "0.0.5"
+version = "0.0.8"
 description = "What could go wrong giving full shell access to chatgpt?"
 readme = "README.md"
-requires-python = ">=3.7"
+requires-python = ">=3.8, <3.13"
 dependencies = [
     "openai>=1.46.0",
     "mypy>=1.11.2",
@@ -21,6 +21,7 @@ dependencies = [
     "fastapi>=0.115.0",
     "uvicorn>=0.31.0",
     "websockets>=13.1",
+    "anthropic>=0.36.2",
 ]
 
 [project.urls]
@@ -37,8 +38,8 @@ wcgw = "wcgw:listen"
 
 [tool.uv]
 dev-dependencies = [
-    "ipython>=8.27.0",
     "mypy>=1.11.2",
     "types-toml>=0.10.8.20240310",
     "autoflake",
+    "ipython>=8.12.3",
 ]

{wcgw-0.0.5 → wcgw-0.0.8}/src/relay/serve.py

@@ -1,7 +1,8 @@
 import asyncio
+import base64
 import threading
 import time
-from typing import Callable, Coroutine, Literal, Optional, Sequence
+from typing import Any, Callable, Coroutine, DefaultDict, Literal, Optional, Sequence
 from uuid import UUID
 import fastapi
 from fastapi import WebSocket, WebSocketDisconnect
@@ -36,6 +37,30 @@ clients: dict[UUID, Callable[[Mdata], Coroutine[None, None, None]]] = {}
 websockets: dict[UUID, WebSocket] = {}
 gpts: dict[UUID, Callable[[str], None]] = {}
 
+images: DefaultDict[UUID, dict[str, dict[str, Any]]] = DefaultDict(dict)
+
+
+@app.websocket("/register_serve_image/{uuid}")
+async def register_serve_image(websocket: WebSocket, uuid: UUID) -> None:
+    raise Exception("Disabled")
+    await websocket.accept()
+    received_data = await websocket.receive_json()
+    name = received_data["name"]
+    image_b64 = received_data["image_b64"]
+    image_bytes = base64.b64decode(image_b64)
+    images[uuid][name] = {
+        "content": image_bytes,
+        "media_type": received_data["media_type"],
+    }
+
+
+@app.get("/get_image/{uuid}/{name}")
+async def get_image(uuid: UUID, name: str) -> fastapi.responses.Response:
+    return fastapi.responses.Response(
+        content=images[uuid][name]["content"],
+        media_type=images[uuid][name]["media_type"],
+    )
+
 
 @app.websocket("/register/{uuid}")
 async def register_websocket(websocket: WebSocket, uuid: UUID) -> None:
@@ -61,9 +86,34 @@ async def register_websocket(websocket: WebSocket, uuid: UUID) -> None:
         print(f"Client {uuid} disconnected")
 
 
-@app.post("/action")
-async def chatgpt_server(json_data: Mdata) -> str:
-    user_id = json_data.user_id
+@app.post("/write_file")
+async def write_file(write_file_data: Writefile, user_id: UUID) -> str:
+    if user_id not in clients:
+        raise fastapi.HTTPException(
+            status_code=404, detail="User with the provided id not found"
+        )
+
+    results: Optional[str] = None
+
+    def put_results(result: str) -> None:
+        nonlocal results
+        results = result
+
+    gpts[user_id] = put_results
+
+    await clients[user_id](Mdata(data=write_file_data, user_id=user_id))
+
+    start_time = time.time()
+    while time.time() - start_time < 30:
+        if results is not None:
+            return results
+        await asyncio.sleep(0.1)
+
+    raise fastapi.HTTPException(status_code=500, detail="Timeout error")
+
+
+@app.post("/execute_bash")
+async def execute_bash(excute_bash_data: ExecuteBash, user_id: UUID) -> str:
     if user_id not in clients:
         raise fastapi.HTTPException(
             status_code=404, detail="User with the provided id not found"
@@ -77,7 +127,7 @@ async def chatgpt_server(json_data: Mdata) -> str:
 
     gpts[user_id] = put_results
 
-    await clients[user_id](json_data)
+    await clients[user_id](Mdata(data=excute_bash_data, user_id=user_id))
 
     start_time = time.time()
     while time.time() - start_time < 30:

{wcgw-0.0.5 → wcgw-0.0.8}/src/wcgw/basic.py

@@ -1,4 +1,6 @@
+import base64
 import json
+import mimetypes
 from pathlib import Path
 import sys
 import traceback
@@ -8,17 +10,21 @@ from openai import OpenAI
 from openai.types.chat import (
     ChatCompletionMessageParam,
     ChatCompletionAssistantMessageParam,
+    ChatCompletionUserMessageParam,
+    ChatCompletionContentPartParam,
     ChatCompletionMessage,
     ParsedChatCompletionMessage,
 )
 import rich
+import petname
 from typer import Typer
 import uuid
 
-from .common import Models, discard_input
-from .common import CostData, History
+from wcgw.common import Config, text_from_editor
+
+from .common import Models
 from .openai_utils import get_input_cost, get_output_cost
-from .tools import ExecuteBash, GetShellOutputLastCommand
+from .tools import ExecuteBash, ReadImage, ImageData
 
 from .tools import (
     BASH_CLF_OUTPUT,
@@ -34,40 +40,14 @@ from .tools import (
 import tiktoken
 
 from urllib import parse
-import subprocess
 import os
-import tempfile
 
 import toml
-from pydantic import BaseModel
 
 
 from dotenv import load_dotenv
 
-
-class Config(BaseModel):
-    model: Models
-    secondary_model: Models
-    cost_limit: float
-    cost_file: dict[Models, CostData]
-    cost_unit: str = "$"
-
-
-def text_from_editor(console: rich.console.Console) -> str:
-    # First consume all the input till now
-    discard_input()
-    console.print("\n---------------------------------------\n# User message")
-    data = input()
-    if data:
-        return data
-    editor = os.environ.get("EDITOR", "vim")
-    with tempfile.NamedTemporaryFile(suffix=".tmp") as tf:
-        subprocess.run([editor, tf.name], check=True)
-        with open(tf.name, "r") as f:
-            data = f.read()
-            console.print(data)
-            return data
-
+History = list[ChatCompletionMessageParam]
 
 def save_history(history: History, session_id: str) -> None:
     myid = str(history[1]["content"]).replace("/", "_").replace(" ", "_").lower()[:60]
@@ -80,6 +60,38 @@ def save_history(history: History, session_id: str) -> None:
         json.dump(history, f, indent=3)
 
 
+def parse_user_message_special(msg: str) -> ChatCompletionUserMessageParam:
+    # Search for lines starting with `%` and treat them as special commands
+    parts: list[ChatCompletionContentPartParam] = []
+    for line in msg.split("\n"):
+        if line.startswith("%"):
+            args = line[1:].strip().split(" ")
+            command = args[0]
+            assert command == 'image'
+            image_path = args[1]
+            with open(image_path, 'rb') as f:
+                image_bytes = f.read()
+                image_b64 = base64.b64encode(image_bytes).decode("utf-8")
+                image_type = mimetypes.guess_type(image_path)[0]
+                dataurl=f'data:{image_type};base64,{image_b64}'
+            parts.append({
+                'type': 'image_url',
+                'image_url': {
+                    'url': dataurl,
+                    'detail': 'auto'
+                }
+            })
+        else:
+            if len(parts) > 0 and parts[-1]['type'] == 'text':
+                parts[-1]['text'] += '\n' + line
+            else:
+                parts.append({'type': 'text', 'text': line})
+    return {
+        'role': 'user',
+        'content': parts
+    }
+
+
 app = Typer(pretty_exceptions_show_locals=False)
 
 
@@ -94,6 +106,7 @@ def loop(
     session_id = str(uuid.uuid4())[:6]
 
     history: History = []
+    waiting_for_assistant = False
     if resume:
         if resume == "latest":
             resume_path = sorted(Path(".wcgw").iterdir(), key=os.path.getmtime)[-1]
@@ -108,6 +121,7 @@ def loop(
         if history[1]["role"] != "user":
             raise ValueError("Invalid history file, second message should be user")
         first_message = ""
+        waiting_for_assistant = history[-1]['role'] != 'assistant'
 
     my_dir = os.path.dirname(__file__)
     config_file = os.path.join(my_dir, "..", "..", "config.toml")
@@ -134,19 +148,18 @@ Execute a bash script. Stateful (beware with subsequent calls).
 Execute commands using `execute_command` attribute.
 Do not use interactive commands like nano. Prefer writing simpler commands.
 Last line will always be `(exit <int code>)` except if
-the last line is `(waiting for input)` which will be the case if you've run any interactive command (which you shouldn't run) by mistake. You can then send input using `send_ascii` attributes.
+the last line is `(pending)` if the program is still running or waiting for user inputs. You can then send input using `send_ascii` attributes. You get status by sending `send_ascii: [10]`.
 Optionally the last line is `(won't exit)` in which case you need to kill the process if you want to run a new command.
 Optionally `exit shell has restarted` is the output, in which case environment resets, you can run fresh commands.
 The first line might be `(...truncated)` if the output is too long.""",
         ),
-        openai.pydantic_function_tool(
-            GetShellOutputLastCommand,
-            description="Get output of the last command run in the shell. Use this in case you want to know status of a running program.",
-        ),
         openai.pydantic_function_tool(
             Writefile,
             description="Write content to a file. Provide file path and content. Use this instead of ExecuteBash for writing files.",
         ),
+        openai.pydantic_function_tool(
+            ReadImage, description="Read an image from the shell."
+        ),
     ]
     uname_sysname = os.uname().sysname
     uname_machine = os.uname().machine
@@ -165,12 +178,11 @@ System information:
     - Machine: {uname_machine}
 """
 
-    has_tool_output = False
     if not history:
         history = [{"role": "system", "content": system}]
     else:
         if history[-1]["role"] == "tool":
-            has_tool_output = True
+            waiting_for_assistant = True
 
     client = OpenAI()
 
@@ -189,16 +201,16 @@ System information:
             )
             break
 
-        if not has_tool_output:
+        if not waiting_for_assistant:
             if first_message:
                 msg = first_message
                 first_message = ""
             else:
                 msg = text_from_editor(user_console)
 
-            history.append({"role": "user", "content": msg})
+            history.append(parse_user_message_special(msg))
         else:
-            has_tool_output = False
+            waiting_for_assistant = False
 
         cost_, input_toks_ = get_input_cost(
             config.cost_file[config.model], enc, history
@@ -223,6 +235,7 @@ System information:
         _histories: History = []
         item: ChatCompletionMessageParam
         full_response: str = ""
+        image_histories: History = []
         try:
             for chunk in stream:
                 if chunk.choices[0].finish_reason == "tool_calls":
@@ -236,7 +249,7 @@ System information:
                                 "type": "function",
                                 "function": {
                                     "arguments": tool_args,
-                                    "name": "execute_bash",
+                                    "name": type(which_tool(tool_args)).__name__,
                                 },
                             }
                             for tool_call_id, toolcallargs in tool_call_args_by_id.items()
@@ -252,7 +265,7 @@ System information:
                     )
                     system_console.print(f"\nTotal cost: {config.cost_unit}{cost:.3f}")
                     output_toks += output_toks_
-
+                    
                     _histories.append(item)
                     for tool_call_id, toolcallargs in tool_call_args_by_id.items():
                         for toolindex, tool_args in toolcallargs.items():
@@ -284,13 +297,50 @@ System information:
                                     f"\nTotal cost: {config.cost_unit}{cost:.3f}"
                                 )
                                 return output_or_done.task_output, cost
+                                
                             output = output_or_done
 
-                            item = {
-                                "role": "tool",
-                                "content": str(output),
-                                "tool_call_id": tool_call_id + str(toolindex),
-                            }
+                            if isinstance(output, ImageData):
+                                randomId = petname.Generate(2, "-")
+                                if not image_histories:
+                                    image_histories.extend([
+                                        {
+                                            'role': 'assistant',
+                                            'content': f'Share images with ids: {randomId}'
+
+                                        },
+                                        {
+                                            'role': 'user',
+                                            'content': [{
+                                                'type': 'image_url',
+                                                'image_url': {
+                                                    'url': output.dataurl,
+                                                    'detail': 'auto'
+                                                }
+                                            }]
+                                        }]
+                                    )
+                                else:
+                                    image_histories[0]['content'] += ', ' + randomId
+                                    image_histories[1]["content"].append({ # type: ignore
+                                        'type': 'image_url',
+                                        'image_url': {
+                                            'url': output.dataurl,
+                                            'detail': 'auto'
+                                        }
+                                    })
+
+                                item = {
+                                    "role": "tool",
+                                    "content": f'Ask user for image id: {randomId}',
+                                    "tool_call_id": tool_call_id + str(toolindex),
+                                }
+                            else:
+                                item = {
+                                    "role": "tool",
+                                    "content": str(output),
+                                    "tool_call_id": tool_call_id + str(toolindex),
+                                }
                             cost_, output_toks_ = get_output_cost(
                                 config.cost_file[config.model], enc, item
                             )
@@ -298,7 +348,7 @@ System information:
                             output_toks += output_toks_
 
                             _histories.append(item)
-                    has_tool_output = True
+                    waiting_for_assistant = True
                     break
                 elif chunk.choices[0].finish_reason:
                     assistant_console.print("")
@@ -327,11 +377,11 @@ System information:
                 assistant_console.print(chunk_str, end="")
                 full_response += chunk_str
         except KeyboardInterrupt:
-            has_tool_output = False
+            waiting_for_assistant = False
             input("Interrupted...enter to redo the current turn")
         else:
             history.extend(_histories)
-
+            history.extend(image_histories)
             save_history(history, session_id)
 
     return "Couldn't finish the task", cost

{wcgw-0.0.5 → wcgw-0.0.8}/src/wcgw/common.py

@@ -1,9 +1,13 @@
+import os
 import select
+import subprocess
 import sys
+import tempfile
 import termios
 import tty
 from typing import Literal
 from pydantic import BaseModel
+import rich
 
 
 class CostData(BaseModel):
@@ -11,14 +15,6 @@ class CostData(BaseModel):
     cost_per_1m_output_tokens: float
 
 
-from openai.types.chat import (
-    ChatCompletionMessageParam,
-    ChatCompletionAssistantMessageParam,
-    ChatCompletionMessage,
-    ParsedChatCompletionMessage,
-)
-
-History = list[ChatCompletionMessageParam]
 Models = Literal["gpt-4o-2024-08-06", "gpt-4o-mini"]
 
 
@@ -45,3 +41,27 @@ def discard_input() -> None:
     finally:
         # Restore old terminal settings
         termios.tcsetattr(fd, termios.TCSADRAIN, old_settings)
+
+
+class Config(BaseModel):
+    model: Models
+    secondary_model: Models
+    cost_limit: float
+    cost_file: dict[Models, CostData]
+    cost_unit: str = "$"
+
+
+def text_from_editor(console: rich.console.Console) -> str:
+    # First consume all the input till now
+    discard_input()
+    console.print("\n---------------------------------------\n# User message")
+    data = input()
+    if data:
+        return data
+    editor = os.environ.get("EDITOR", "vim")
+    with tempfile.NamedTemporaryFile(suffix=".tmp") as tf:
+        subprocess.run([editor, tf.name], check=True)
+        with open(tf.name, "r") as f:
+            data = f.read()
+            console.print(data)
+            return data

{wcgw-0.0.5 → wcgw-0.0.8}/src/wcgw/openai_utils.py

@@ -28,9 +28,19 @@ def get_input_cost(
     input_tokens = 0
     for msg in history:
         content = msg["content"]
-        if not isinstance(content, str):
+        refusal = msg.get("refusal")
+        if isinstance(content, list):
+            for part in content:
+                if 'text' in part:
+                    input_tokens += len(enc.encode(part['text']))
+        elif content is None:
+            if refusal is None:
+                raise ValueError("Expected content or refusal to be present")
+            input_tokens += len(enc.encode(str(refusal)))
+        elif not isinstance(content, str):
             raise ValueError(f"Expected content to be string, got {type(content)}")
-        input_tokens += len(enc.encode(content))
+        else:
+            input_tokens += len(enc.encode(content))
     cost = input_tokens * cost_map.cost_per_1m_input_tokens / 1_000_000
     return cost, input_tokens