Modified to comply with generator update, and so that no changes to t…

…he base interpreter class or anything in the core folder was needed. Update README from base/main merge rebased branch to main. (#2) * fix: stop overwriting boolean config values Without the default set to None, any boolean CLI flag that isn't passed reverts to its default state even if it is configured in the config.yaml file. * The Generator Update (English docs) * Improved --conversations, --config --------- quality of life and error messages errors and stuff again re-add readline method because doc formatting removed it somehow fix readline method of wrapper added file upload and download functionality finalized upload and download commands. tested stuff visual Improved --conversations, --config The Generator Update (English docs) fix: stop overwriting boolean config values Without the default set to None, any boolean CLI flag that isn't passed reverts to its default state even if it is configured in the config.yaml file. Update WINDOWS.md Warns the user to re-launch cmd windows after installing llama locally Fix ARM64 llama-cpp-python Install on Apple Silicon This commit updates the `MACOS.md` documentation to include detailed steps for correctly installing `llama-cpp-python` with ARM64 architecture support on Apple Silicon-based macOS systems. The update provides: - A prerequisite check for Xcode Command Line Tools. - Step-by-step installation instructions for `llama-cpp-python` with ARM64 and Metal support. - A verification step to confirm the correct installation of `llama-cpp-python` for ARM64 architecture. - An additional step for installing server components for `llama-cpp-python`. This commit resolves the issue described in `ARM64 Installation Issue with llama-cpp-python on Apple Silicon Macs for interpreter --local #503`. Broken empty message response fix crash on unknwon command on call to display help message removed unnecessary spaces Update get_relevant_procedures.py Fixed a typo in the instructions to the model The Generator Update The Generator Update The Generator Update - Azure fix The Generator Update - Azure function calling The Generator Update - Azure fix Better debugging Better debugging Proper TokenTrimming for new models Generator Update Fixes (Updated Version) Generator Update Quick Fixes Added example JARVIS Colab Notebook Added example JARVIS Colab Notebook Skip wrap_in_trap on Windows fix: allow args to have choices and defaults This allows non-boolean args to define possible options and default values, which were ignored previously. feat: add semgrep code scanning via --safe flag This reintroduces the --safe functionality from #24. --safe has 3 possible values auto, ask, and off Code scanning is opt-in. fix: default to 'off' for scan_code attribute fix: toggle code_scan based on auto_run setting; update --scan docs revert: undo default and choices change to cli.py This is being removed from this PR in favor of a standalone fix in #511 feat: cleanup code scanning and convert to safe mode docs: fix naming of safe_mode flag in README fix: pass debug_mode flag into file cleanup for code scan fix: remove extra tempfile import from scan_code util Fixed first message inturruption error Holding `--safe` docs for pip release fix: stop overwriting safe_mode config.yaml setting with default in args Fixed `%load` magic command But I think we should deprecate it in favor of `--conversations`. Generalized API key error message Better model validation, better config debugging Better config debugging Better config debugging Better config debugging Better --config Cleaned up initial message Generator Update Quick Fixes II Force then squashing (#3)
OpenInterpreter · Sep 29, 2023 · 6ae3f20 · 6ae3f20
1 parent 98407b9
commit 6ae3f20
Show file tree

Hide file tree

Showing 39 changed files with 1,959 additions and 213 deletions.
diff --git a/.vscode/settings.json b/.vscode/settings.json
@@ -0,0 +1,3 @@
+{
+    "python.analysis.typeCheckingMode": "basic"
+}
diff --git a/README.md b/README.md
@@ -51,6 +51,10 @@ https://github.com/KillianLucas/open-interpreter/assets/63927363/37152071-680d-4
 
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1WKmRXZgsErej2xUriKzxrEAXdxMSgWbb?usp=sharing)
 
+#### Along with an example implementation of a voice interface (inspired by _Her_):
+
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1NojYGHDgxH6Y1G1oxThEBBb2AtyODBIK)
+
 ## Quick Start
 
 ```shell
@@ -93,6 +97,15 @@ This combines the power of GPT-4's Code Interpreter with the flexibility of your
 
 ## Commands
 
+**Update:** The Generator Update (0.1.5) introduced streaming:
+
+```python
+message = "What operating system are we on?"
+
+for chunk in interpreter.chat(message, display=False, stream=True):
+  print(chunk)
+```
+
 ### Interactive Chat
 
 To start an interactive chat in your terminal, either run `interpreter` from the command line:
@@ -107,6 +120,15 @@ Or `interpreter.chat()` from a .py file:
 interpreter.chat()
 ```
 
+**You can also stream each chunk:**
+
+```python
+message = "What operating system are we on?"
+
+for chunk in interpreter.chat(message, display=False, stream=True):
+  print(chunk)
+```
+
 ### Programmatic Chat
 
 For more precise control, you can pass messages directly to `.chat(message)`:
@@ -131,13 +153,13 @@ interpreter.reset()
 
 ### Save and Restore Chats
 
-`interpreter.chat()` returns a List of messages when return_messages=True, which can be used to resume a conversation with `interpreter.load(messages)`:
+`interpreter.chat()` returns a List of messages, which can be used to resume a conversation with `interpreter.messages = messages`:
 
 ```python
-messages = interpreter.chat("My name is Killian.", return_messages=True) # Save messages to 'messages'
+messages = interpreter.chat("My name is Killian.") # Save messages to 'messages'
 interpreter.reset() # Reset interpreter ("Killian" will be forgotten)
 
-interpreter.load(messages) # Resume chat from 'messages' ("Killian" will be remembered)
+interpreter.messages = messages # Resume chat from 'messages' ("Killian" will be remembered)
 ```
 
 ### Customize System Message
@@ -151,20 +173,26 @@ Run shell commands with -y so the user doesn't have to confirm them.
 print(interpreter.system_message)
 ```
 
-### Change the Model
+### Change your Language Model
 
-For `gpt-3.5-turbo`, use fast mode:
+Open Interpreter uses [LiteLLM](https://docs.litellm.ai/docs/providers/) to connect to language models.
+
+You can change the model by setting the model parameter:
 
 ```shell
-interpreter --fast
+interpreter --model gpt-3.5-turbo
+interpreter --model claude-2
+interpreter --model command-nightly
 ```
 
-In Python, you will need to set the model manually:
+In Python, set the model on the object:
 
 ```python
 interpreter.model = "gpt-3.5-turbo"
 ```
 
+[Find the appropriate "model" string for your language model here.](https://docs.litellm.ai/docs/providers/)
+
 ### Running Open Interpreter locally
 
 ⓘ **Issues running locally?** Read our new [GPU setup guide](./docs/GPU.md) and [Windows setup guide](./docs/WINDOWS.md).
@@ -175,10 +203,10 @@ You can run `interpreter` in local mode from the command line to use `Code Llama
 interpreter --local
 ```
 
-Or run any Hugging Face model **locally** by using its repo ID (e.g. "tiiuae/falcon-180B"):
+Or run any Hugging Face model **locally** by running `--local` in conjunction with a repo ID (e.g. "tiiuae/falcon-180B"):
 
 ```shell
-interpreter --model tiiuae/falcon-180B
+interpreter --local --model tiiuae/falcon-180B
 ```
 
 #### Local model params
@@ -191,25 +219,6 @@ Smaller context windows will use less RAM, so we recommend trying a shorter wind
 interpreter --max_tokens 2000 --context_window 16000
 ```
 
-### Azure Support
-
-To connect to an Azure deployment, the `--use-azure` flag will walk you through setting this up:
-
-```shell
-interpreter --use-azure
-```
-
-In Python, set the following variables:
-
-```
-interpreter.use_azure = True
-interpreter.api_key = "your_openai_api_key"
-interpreter.azure_api_base = "your_azure_api_base"
-interpreter.azure_api_version = "your_azure_api_version"
-interpreter.azure_deployment_name = "your_azure_deployment_name"
-interpreter.azure_api_type = "azure"
-```
-
 ### Debug mode
 
 To help contributors inspect Open Interpreter, `--debug` mode is highly verbose.
@@ -239,24 +248,18 @@ provided, it defaults to 'messages.json'.
  is provided, it defaults to 'messages.json'.  
  • `%help`: Show the help message.
 
-Feel free to try out these commands and let us know your feedback!
+### Configuration
 
-### Configuration with .env
+Open Interpreter allows you to set default behaviors using a `config.yaml` file.
 
-Open Interpreter allows you to set default behaviors using a .env file. This provides a flexible way to configure the interpreter without changing command-line arguments every time.
+This provides a flexible way to configure the interpreter without changing command-line arguments every time.
 
-Here's a sample .env configuration:
+Run the following command to open the configuration file:
 
 ```
-INTERPRETER_CLI_AUTO_RUN=False
-INTERPRETER_CLI_FAST_MODE=False
-INTERPRETER_CLI_LOCAL_RUN=False
-INTERPRETER_CLI_DEBUG=False
-INTERPRETER_CLI_USE_AZURE=False
+interpreter --config
 ```
 
-You can modify these values in the .env file to change the default behavior of the Open Interpreter.
-
 ## Safety Notice
 
 Since generated code is executed in your local environment, it can interact with your files and system settings, potentially leading to unexpected outcomes like data loss or security risks.

diff --git a/del.py b/del.py
@@ -0,0 +1,20 @@
+from textual.app import App
+from textual_terminal import Terminal
+
+class TerminalApp(App):
+
+    async def on_mount(self) -> None:
+        # Create a layout with two terminals
+        await self.layout.dock(Terminal(command="htop", id="terminal_htop"), edge="top", size=10)
+        await self.layout.dock(Terminal(command="bash", id="terminal_bash"), edge="bottom")
+
+    async def on_ready(self) -> None:
+        # Start the commands in each terminal
+        terminal_htop: Terminal = await self.get_widget("terminal_htop")
+        await terminal_htop.start()
+
+        terminal_bash: Terminal = await self.get_widget("terminal_bash")
+        await terminal_bash.start()
+
+app = TerminalApp()
+app.run()
diff --git a/docs/MACOS.md b/docs/MACOS.md
@@ -4,42 +4,80 @@ When running Open Interpreter on macOS with Code-Llama (either because you did
 not enter an OpenAI API key or you ran `interpreter --local`) you may want to
 make sure it works correctly by following the instructions below.
 
-Tested on **MacOS Ventura 13.5** with **M2 Pro Chip**.
+Tested on **MacOS Ventura 13.5** with **M2 Pro Chip** and **MacOS Ventura 13.5.1** with **M1 Max**.
 
 I use conda as a virtual environment but you can choose whatever you want. If you go with conda you will find the Apple M1 version of miniconda here: [Link](https://docs.conda.io/projects/miniconda/en/latest/)
 
-```
+```bash
 conda create -n openinterpreter python=3.11.4
 ```
 
 **Activate your environment:**
 
-```
+```bash
 conda activate openinterpreter
 ```
 
 **Install open-interpreter:**
 
-```
+```bash
 pip install open-interpreter
 ```
 
 **Uninstall any previously installed llama-cpp-python packages:**
 
-```
+```bash
 pip uninstall llama-cpp-python -y
 ```
 
-**Install llama-cpp-python with Apple Silicon support:**
+## Install llama-cpp-python with Apple Silicon support
+
+### Prerequisites: Xcode Command Line Tools
+
+Before running the `CMAKE_ARGS` command to install `llama-cpp-python`, make sure you have Xcode Command Line Tools installed on your system. These tools include compilers and build systems essential for source code compilation.
 
-Part 1
+Before proceeding, make sure you have the Xcode Command Line Tools installed. You can check whether they are installed by running:
 
+```bash
+xcode-select -p
 ```
-CMAKE_ARGS="-DLLAMA_METAL=on" FORCE_CMAKE=1 pip install -U llama-cpp-python --no-cache-dir
+
+If this command returns a path, then the Xcode Command Line Tools are already installed. If not, you'll get an error message, and you can install them by running:
+
+```bash
+xcode-select --install
 ```
 
-Part 2
+Follow the on-screen instructions to complete the installation. Once installed, you can proceed with installing an Apple Silicon compatible `llama-cpp-python`.
+
+---
+### Step 1: Installing llama-cpp-python with ARM64 Architecture and Metal Support
 
+
+```bash
+CMAKE_ARGS="-DCMAKE_OSX_ARCHITECTURES=arm64 -DLLAMA_METAL=on" FORCE_CMAKE=1 pip install --upgrade --force-reinstall llama-cpp-python --no-cache-dir
+--no-cache-dir
 ```
+
+### Step 2: Verifying Installation of llama-cpp-python with ARM64 Support
+
+After completing the installation, you can verify that `llama-cpp-python` was correctly installed with ARM64 architecture support by running the following command:
+
+```bash
+lipo -info /path/to/libllama.dylib
+```
+
+Replace `/path/to/` with the actual path to the `libllama.dylib` file. You should see output similar to:
+
+```bash
+Non-fat file: /Users/[user]/miniconda3/envs/openinterpreter/lib/python3.11/site-packages/llama_cpp/libllama.dylib is architecture: arm64
+```
+
+If the architecture is indicated as `arm64`, then you've successfully installed the ARM64 version of `llama-cpp-python`.
+
+### Step 3: Installing Server Components for llama-cpp-python
+
+
+```bash
 pip install 'llama-cpp-python[server]'
 ```
diff --git a/docs/WINDOWS.md b/docs/WINDOWS.md
@@ -40,3 +40,7 @@ The resolve this issue, perform the following steps.
     ```
 
     Alternatively, if you want to include GPU suppport, follow the steps in [Local Language Models with GPU Support](./GPU.md)
+
+6.  Make sure you close and re-launch any cmd windows that were running interpreter
+
+
diff --git a/interpreter/cli/cli.py b/interpreter/cli/cli.py
@@ -2,6 +2,7 @@
 import subprocess
 import os
 import platform
+import pkg_resources
 import appdirs
 from ..utils.display_markdown_message import display_markdown_message
 from ..terminal_interface.conversation_navigator import conversation_navigator
@@ -72,6 +73,20 @@
         "nickname": "ak",
         "help_text": "optionally set the API key for your llm calls (this will override environment variables)",
         "type": str
+    },
+    {
+        "name": "use_containers",
+        "nickname": "uc",
+        "help_text": "optionally use a Docker Container for the interpreters code execution. this will seperate execution from your main computer. this also allows execution on a remote server via the 'DOCKER_HOST' environment variable and the dockerengine api.",
+        "type": bool
+    },
+    {
+
+        "name": "safe_mode",
+        "nickname": "safe",
+        "help_text": "optionally enable safety mechanisms like code scanning; valid options are off, ask, and auto",
+        "type": str,
+        "choices": ["off", "ask", "auto"]
     }
 ]
 
@@ -82,14 +97,15 @@ def cli(interpreter):
     # Add arguments
     for arg in arguments:
         if arg["type"] == bool:
-            parser.add_argument(f'-{arg["nickname"]}', f'--{arg["name"]}', dest=arg["name"], help=arg["help_text"], action='store_true')
+            parser.add_argument(f'-{arg["nickname"]}', f'--{arg["name"]}', dest=arg["name"], help=arg["help_text"], action='store_true', default=None)
         else:
             parser.add_argument(f'-{arg["nickname"]}', f'--{arg["name"]}', dest=arg["name"], help=arg["help_text"], type=arg["type"])
 
     # Add special arguments
     parser.add_argument('--config', dest='config', action='store_true', help='open config.yaml file in text editor')
     parser.add_argument('--conversations', dest='conversations', action='store_true', help='list conversations to resume')
     parser.add_argument('-f', '--fast', dest='fast', action='store_true', help='(depracated) runs `interpreter --model gpt-3.5-turbo`')
+    parser.add_argument('--version', dest='version', action='store_true', help="get Open Interpreter's version number")
 
     # TODO: Implement model explorer
     # parser.add_argument('--models', dest='models', action='store_true', help='list avaliable models')
@@ -99,7 +115,8 @@ def cli(interpreter):
     # This should be pushed into an open_config.py util
     # If --config is used, open the config.yaml file in the Open Interpreter folder of the user's config dir
     if args.config:
-        config_path = os.path.join(appdirs.user_config_dir(), 'Open Interpreter', 'config.yaml')
+        config_dir = appdirs.user_config_dir("Open Interpreter")
+        config_path = os.path.join(config_dir, 'config.yaml')
         print(f"Opening `{config_path}`...")
         # Use the default system editor to open the file
         if platform.system() == 'Windows':
@@ -111,6 +128,7 @@ def cli(interpreter):
             except FileNotFoundError:
                 # Fallback to using 'open' on macOS if 'xdg-open' is not available
                 subprocess.call(['open', config_path])
+        return
 
     # TODO Implement model explorer
     """
@@ -126,6 +144,10 @@ def cli(interpreter):
         if attr_value is not None and hasattr(interpreter, attr_name):
             setattr(interpreter, attr_name, attr_value)
 
+    # if safe_mode and auto_run are enabled, safe_mode disables auto_run
+    if interpreter.auto_run and not interpreter.safe_mode == "off":
+        setattr(interpreter, "auto_run", False)
+
     # Default to CodeLlama if --local is on but --model is unset
     if interpreter.local and args.model is None:
         # This will cause the terminal_interface to walk the user through setting up a local LLM
@@ -136,6 +158,11 @@ def cli(interpreter):
         conversation_navigator(interpreter)
         return
 
+    if args.version:
+        version = pkg_resources.get_distribution("open-interpreter").version
+        print(f"Open Interpreter {version}")
+        return
+
     # Depracated --fast
     if args.fast:
         # This will cause the terminal_interface to walk the user through setting up a local LLM