Merge pull request #149 from rgbkrk/update

Update

chatlab/chat.py

@@ -390,16 +390,14 @@ def register(
         self,
         function: None = None,
         parameter_schema: Optional[Union[Type["BaseModel"], dict]] = None,
-    ) -> Callable:
-        ...
+    ) -> Callable: ...
 
     @overload
     def register(
         self,
         function: Callable,
         parameter_schema: Optional[Union[Type["BaseModel"], dict]] = None,
-    ) -> FunctionDefinition:
-        ...
+    ) -> FunctionDefinition: ...
 
     def register(
         self,

chatlab/decorators.py

@@ -29,17 +29,19 @@
 
 """
 
-
 from typing import Callable, Optional
 
 from pydantic import BaseModel
 
+
 class ChatlabMetadata(BaseModel):
     """ChatLab metadata for a function."""
+
     expose_exception_to_llm: bool = True
     render: Optional[Callable] = None
     bubble_exceptions: bool = False
 
+
 def bubble_exceptions(func):
     if not hasattr(func, "chatlab_metadata"):
         func.chatlab_metadata = ChatlabMetadata()
@@ -51,6 +53,7 @@ def bubble_exceptions(func):
     func.chatlab_metadata.bubble_exceptions = True
     return func
 
+
 def expose_exception_to_llm(func):
     """Expose exceptions from calling the function to the LLM.
 
@@ -107,6 +110,7 @@ def store_knowledge_graph(kg: KnowledgeGraph, comment: str = "Knowledge Graph"):
 chat.register(store_knowledge_graph)
 '''
 
+
 def incremental_display(render_func: Callable):
     def decorator(func):
         if not hasattr(func, "chatlab_metadata"):
@@ -118,5 +122,5 @@ def decorator(func):
 
         func.chatlab_metadata.render = render_func
         return func
-    return decorator
 
+    return decorator

chatlab/messaging.py

@@ -105,8 +105,7 @@ def function_result(name: str, content: str) -> ChatCompletionMessageParam:
 
 
 class HasGetToolArgumentsParameter(Protocol):
-    def get_tool_arguments_parameter(self) -> ChatCompletionMessageToolCallParam:
-        ...
+    def get_tool_arguments_parameter(self) -> ChatCompletionMessageToolCallParam: ...
 
 
 def assistant_tool_calls(tool_calls: Iterable[HasGetToolArgumentsParameter]) -> ChatCompletionMessageParam:

chatlab/models.py

@@ -5,6 +5,7 @@
 
 class ChatModel(Enum):
     """Models available for use with chatlab."""
+
     GPT_4_TURBO_PREVIEW = "gpt-4-turbo-preview"
     GPT_4_0125_PREVIEW = "gpt-4-0125-preview"
     GPT_4_1106_PREVIEW = "gpt-4-1106-preview"

chatlab/registry.py

@@ -288,16 +288,14 @@ def register(
         self,
         function: None = None,
         parameter_schema: Optional[Union[Type["BaseModel"], dict]] = None,
-    ) -> Callable:
-        ...
+    ) -> Callable: ...
 
     @overload
     def register(
         self,
         function: Callable,
         parameter_schema: Optional[Union[Type["BaseModel"], dict]] = None,
-    ) -> FunctionDefinition:
-        ...
+    ) -> FunctionDefinition: ...
 
     def register(
         self,
@@ -438,7 +436,13 @@ def api_manifest(self, function_call_option: FunctionCall = "auto") -> APIManife
 
     @property
     def tools(self) -> Iterable[ChatCompletionToolParam]:
-        return [{"type": "function", "function": adapt_function_definition(f)} for f in self.__schemas.values()]
+        return [
+            ChatCompletionToolParam(
+                type="function",
+                function=adapt_function_definition(f),  # type: ignore
+            )
+            for f in self.__schemas.values()
+        ]
 
     async def call(self, name: str, arguments: Optional[str] = None) -> Any:
         """Call a function by name with the given parameters."""

chatlab/tools/__init__.py

@@ -8,4 +8,3 @@
     "run_python",
     "shell_functions",
 ]
-

chatlab/tools/_mediatypes.py

@@ -1,4 +1,5 @@
 """Media types for rich output for LLMs and in-notebook."""
+
 import json
 from typing import Optional
 

chatlab/tools/colors.py

@@ -1,4 +1,5 @@
 """Let models pick and show color palettes to you."""
+
 import hashlib
 from typing import List, Optional
 from pydantic import BaseModel, validator, Field

chatlab/tools/files.py

@@ -7,6 +7,7 @@
 
 You've been warned. Have fun and be safe!
 """
+
 import asyncio
 import os
 

chatlab/tools/python.py

@@ -1,4 +1,5 @@
 """The in-IPython python code runner for ChatLab."""
+
 from traceback import TracebackException
 from typing import Optional
 

chatlab/tools/shell.py

@@ -1,4 +1,5 @@
 """Shell commands for ChatLab."""
+
 import asyncio
 import subprocess
 

chatlab/views/__init__.py

@@ -1,9 +1,6 @@
 """Views for ChatLab."""
+
 from .assistant import AssistantMessageView
 from .tools import ToolArguments, ToolCalled
 
-__all__ = [
-    "AssistantMessageView",
-    "ToolArguments",
-    "ToolCalled"
-]
+__all__ = ["AssistantMessageView", "ToolArguments", "ToolCalled"]

chatlab/views/tools.py

@@ -15,8 +15,7 @@
 from IPython.display import display
 from IPython.core.getipython import get_ipython
 
-from instructor.dsl.partialjson import JSONParser
-
+from jiter import from_json
 
 
 class ToolArguments(AutoUpdate):
@@ -75,11 +74,11 @@ def update(self) -> None:
 
     def render(self):
         if self.custom_render is not None:
-            # We use the same definition as was in the original function
             try:
-                parser = JSONParser()
-                possible_args = parser.parse(self.arguments)
-
+                possible_args = from_json(self.arguments.encode("utf-8"), partial_mode="trailing-strings")
+            except Exception:
+                return None
+            try:
                 Model = extract_model_from_function(self.name, self.custom_render)
                 # model = Model.model_validate(possible_args)
                 model = Model(**possible_args)
@@ -110,13 +109,17 @@ def append_arguments(self, arguments: str):
     def apply_result(self, result: str):
         """Replaces the existing display with a new one that shows the result of the tool being called."""
         tc = ToolCalled(
-            id=self.id, name=self.name, arguments=self.arguments, result=result, display_id=self.display_id,
-            custom_render=self.custom_render
+            id=self.id,
+            name=self.name,
+            arguments=self.arguments,
+            result=result,
+            display_id=self.display_id,
+            custom_render=self.custom_render,
         )
         tc.update()
         return tc
 
-    async def call(self, function_registry: FunctionRegistry) -> 'ToolCalled':
+    async def call(self, function_registry: FunctionRegistry) -> "ToolCalled":
         """Call the function and return a stack of messages for LLM and human consumption."""
         function_name = self.name
         function_args = self.arguments
@@ -185,9 +188,11 @@ def render(self):
         if self.custom_render is not None:
             # We use the same definition as was in the original function
             try:
-                parser = JSONParser()
-                possible_args = parser.parse(self.arguments)
+                possible_args = from_json(self.arguments.encode("utf-8"), partial_mode="trailing-strings")
+            except Exception:
+                return None
 
+            try:
                 Model = extract_model_from_function(self.name, self.custom_render)
                 # model = Model.model_validate(possible_args)
                 model = Model(**possible_args)

notebooks/basics.ipynb

@@ -102,7 +102,7 @@
    "outputs": [],
    "source": [
     "from datetime import datetime\n",
-    "from pytz import timezone, all_timezones, utc\n",
+    "from pytz import timezone, all_timezones\n",
     "from typing import Optional\n",
     "from pydantic import BaseModel\n",
     "\n",