python-codex 0.0.1__py3-none-any.whl → 0.1.0__py3-none-any.whl

pycodex/__init__.py

@@ -1,3 +1,140 @@
-"""Placeholder package for the initial pycodex release."""
+from .agent import AgentLoop
+from .context import ContextConfig, ContextManager
+from .model import (
+    ModelClient,
+    NOOP_MODEL_STREAM_EVENT_HANDLER,
+    ResponsesApiError,
+    ResponsesModelClient,
+    ResponsesProviderConfig,
+)
+from .protocol import (
+    AgentEvent,
+    AssistantMessage,
+    ContextMessage,
+    ModelResponse,
+    ModelStreamEvent,
+    Prompt,
+    ReasoningItem,
+    Submission,
+    ToolCall,
+    ToolResult,
+    ToolSpec,
+    TurnResult,
+    UserMessage,
+)
+from .runtime import AgentRuntime
+from .runtime_services import (
+    PlanStore,
+    RequestPermissionsManager,
+    RequestUserInputManager,
+    SubAgentManager,
+    get_runtime_environment,
+)
+from .tools import (
+    ApplyPatchTool,
+    BaseTool,
+    CloseAgentTool,
+    CodeModeManager,
+    ExecTool,
+    ExecCommandTool,
+    GrepFilesTool,
+    ListDirTool,
+    ReadFileTool,
+    Registry,
+    RequestPermissionsTool,
+    RequestUserInputTool,
+    ResumeAgentTool,
+    SendInputTool,
+    ShellCommandTool,
+    ShellTool,
+    SpawnAgentTool,
+    ToolContext,
+    ToolRegistry,
+    UnifiedExecManager,
+    UpdatePlanTool,
+    ViewImageTool,
+    WaitAgentTool,
+    WaitTool,
+    WebSearchTool,
+    WriteStdinTool,
+)
 
-__version__ = "0.0.1"
+def debug(stop: bool = False):
+
+    import socket
+
+    import debugpy
+
+    if debugpy.is_client_connected():
+        return
+    try:
+        host = socket.gethostbyname(socket.getfqdn(socket.gethostname()))
+        port = 5000
+
+        debugpy.listen((host, port), in_process_debug_adapter=False)
+        print("Waiting for debugger...\a")
+        print(f"ip: {host} port:{port}", flush=True)
+        debugpy.wait_for_client()
+        print("Connected.")
+        if stop:
+            debugpy.breakpoint()
+    except Exception as e:
+        import traceback
+
+        print("\n".join(traceback.format_exception(e)))
+
+__all__ = [
+    "AgentEvent",
+    "AgentLoop",
+    "AgentRuntime",
+    "ApplyPatchTool",
+    "AssistantMessage",
+    "BaseTool",
+    "CloseAgentTool",
+    "CodeModeManager",
+    "ContextConfig",
+    "ContextManager",
+    "ContextMessage",
+    "ExecTool",
+    "ExecCommandTool",
+    "GrepFilesTool",
+    "ListDirTool",
+    "ModelClient",
+    "NOOP_MODEL_STREAM_EVENT_HANDLER",
+    "Registry",
+    "ReadFileTool",
+    "RequestPermissionsTool",
+    "RequestUserInputTool",
+    "ModelResponse",
+    "ModelStreamEvent",
+    "PlanStore",
+    "Prompt",
+    "ReasoningItem",
+    "RequestPermissionsManager",
+    "RequestUserInputManager",
+    "ResumeAgentTool",
+    "ResponsesApiError",
+    "ResponsesModelClient",
+    "ResponsesProviderConfig",
+    "SendInputTool",
+    "ShellCommandTool",
+    "ShellTool",
+    "SpawnAgentTool",
+    "Submission",
+    "SubAgentManager",
+    "ToolCall",
+    "ToolContext",
+    "ToolRegistry",
+    "UnifiedExecManager",
+    "UpdatePlanTool",
+    "ToolResult",
+    "ToolSpec",
+    "TurnResult",
+    "UserMessage",
+    "ViewImageTool",
+    "WaitAgentTool",
+    "WaitTool",
+    "WebSearchTool",
+    "WriteStdinTool",
+    "get_runtime_environment",
+]

pycodex/agent.py

@@ -0,0 +1,290 @@
+from __future__ import annotations
+
+import asyncio
+import json
+from collections.abc import Callable
+
+from .context import ContextManager
+from .model import ModelClient
+from .protocol import (
+    AgentEvent,
+    AssistantMessage,
+    ConversationItem,
+    ModelStreamEvent,
+    ReasoningItem,
+    ToolCall,
+    ToolResult,
+    TurnResult,
+    UserMessage,
+)
+from .tools import ToolContext, ToolRegistry
+from .utils import uuid7_string
+
+
+EventHandler = Callable[[AgentEvent], None]
+NOOP_EVENT_HANDLER: EventHandler = lambda _event: None
+
+
+class TurnInterrupted(RuntimeError):
+    pass
+
+
+class AgentLoop:
+    """Minimal Python port of Codex's turn loop.
+
+    The core idea mirrors the Rust implementation:
+    build a prompt from history, ask the model for output items, run any tool
+    calls, append tool results to history, and keep going until the model emits
+    a pure assistant response.
+    """
+
+    def __init__(
+        self,
+        model_client: ModelClient,
+        tool_registry: ToolRegistry,
+        context_manager: ContextManager | None = None,
+        parallel_tool_calls: bool = True,
+        event_handler: EventHandler = NOOP_EVENT_HANDLER,
+        initial_history: tuple[ConversationItem, ...] = (),
+    ) -> None:
+        self._model_client = model_client
+        self._tool_registry = tool_registry
+        self._context_manager = context_manager or ContextManager()
+        self._parallel_tool_calls = parallel_tool_calls
+        self._event_handler = event_handler
+        self._history: list[ConversationItem] = list(initial_history)
+        self.interrupt_asap = False
+
+    @property
+    def history(self) -> tuple[ConversationItem, ...]:
+        return tuple(self._history)
+
+    def set_event_handler(
+        self, event_handler: EventHandler = NOOP_EVENT_HANDLER
+    ) -> None:
+        self._event_handler = event_handler
+
+    def _raise_if_interrupt_requested(
+        self,
+        turn_id: str,
+        iteration: int,
+        output_text: str | None = None,
+    ) -> None:
+        if self.interrupt_asap:
+            self.interrupt_asap = False
+            payload: dict[str, object] = {"iteration": iteration}
+            if output_text is not None:
+                payload["output_text"] = output_text
+            self._emit("turn_interrupted", turn_id, **payload)
+            raise TurnInterrupted("turn interrupted")
+
+    async def run_turn(
+        self, texts: list[str], turn_id: str | None = None
+    ) -> TurnResult:
+        turn_id = turn_id or uuid7_string()
+        self.interrupt_asap = False
+        for text in texts:
+            self._history.append(UserMessage(text=text))
+
+        self._emit(
+            "turn_started",
+            turn_id,
+            user_text="\n".join(texts),
+            user_texts=list(texts),
+        )
+
+        last_assistant_message: str | None = None
+        final_response_items: tuple[
+            AssistantMessage | ToolCall | ReasoningItem, ...
+        ] = ()
+
+        iteration = 0
+        try:
+            while True:
+                self._raise_if_interrupt_requested(
+                    turn_id,
+                    iteration,
+                    output_text=last_assistant_message,
+                )
+                iteration += 1
+                prompt = self._context_manager.build_prompt(
+                    self._history,
+                    self._tool_registry.model_visible_specs(),
+                    self._parallel_tool_calls,
+                    turn_id=turn_id,
+                )
+                self._emit(
+                    "model_called",
+                    turn_id,
+                    iteration=iteration,
+                    history_size=len(prompt.input),
+                    tool_count=len(prompt.tools),
+                )
+                response = await self._model_client.complete(
+                    prompt,
+                    lambda event: self._handle_model_stream_event(turn_id, event),
+                )
+                final_response_items = tuple(response.items)
+                self._emit(
+                    "model_completed",
+                    turn_id,
+                    iteration=iteration,
+                    item_count=len(response.items),
+                )
+
+                tool_calls: list[ToolCall] = []
+                for item in response.items:
+                    self._history.append(item)
+                    if isinstance(item, AssistantMessage):
+                        last_assistant_message = item.text
+                    elif isinstance(item, ToolCall):
+                        tool_calls.append(item)
+
+                if not tool_calls:
+                    self._raise_if_interrupt_requested(
+                        turn_id,
+                        iteration,
+                        output_text=last_assistant_message,
+                    )
+                    self._emit(
+                        "turn_completed",
+                        turn_id,
+                        iteration=iteration,
+                        output_text=last_assistant_message,
+                    )
+                    return TurnResult(
+                        turn_id=turn_id,
+                        output_text=last_assistant_message,
+                        iterations=iteration,
+                        response_items=final_response_items,
+                        history=tuple(self._history),
+                    )
+
+                tool_results = await self._execute_tool_batch(turn_id, tool_calls)
+                self._history.extend(tool_results)
+                self._history.extend(self._build_follow_up_messages(tool_results))
+                self._raise_if_interrupt_requested(
+                    turn_id,
+                    iteration,
+                    output_text=last_assistant_message,
+                )
+        except TurnInterrupted:
+            raise
+        except Exception as exc:
+            self._emit(
+                "turn_failed",
+                turn_id,
+                iteration=iteration,
+                error=str(exc),
+                error_type=type(exc).__name__,
+            )
+            raise
+
+    async def _execute_tool_batch(
+        self,
+        turn_id: str,
+        tool_calls: list[ToolCall],
+    ) -> list[ToolResult]:
+        results: list[ToolResult] = []
+        parallel_batch: list[ToolCall] = []
+
+        for call in tool_calls:
+            can_run_parallel = (
+                self._parallel_tool_calls
+                and self._tool_registry.supports_parallel(call.name)
+            )
+            if can_run_parallel:
+                parallel_batch.append(call)
+                continue
+
+            if parallel_batch:
+                prior_results = tuple(results)
+                results.extend(
+                    await asyncio.gather(
+                        *(
+                            self._run_single_tool(turn_id, batched_call, prior_results)
+                            for batched_call in parallel_batch
+                        )
+                    )
+                )
+            parallel_batch = []
+            results.append(await self._run_single_tool(turn_id, call, tuple(results)))
+
+        if parallel_batch:
+            prior_results = tuple(results)
+            results.extend(
+                await asyncio.gather(
+                    *(
+                        self._run_single_tool(turn_id, batched_call, prior_results)
+                        for batched_call in parallel_batch
+                    )
+                )
+            )
+        return results
+
+    async def _run_single_tool(
+        self,
+        turn_id: str,
+        call: ToolCall,
+        prior_results: tuple[ToolResult, ...] = (),
+    ) -> ToolResult:
+        self._emit("tool_started", turn_id, tool_name=call.name, call_id=call.call_id)
+        result = await self._tool_registry.execute(
+            call,
+            ToolContext(
+                turn_id=turn_id,
+                history=tuple(self._history) + prior_results,
+                collaboration_mode=self._context_manager.collaboration_mode,
+            ),
+        )
+        payload: dict[str, object] = {
+            "tool_name": call.name,
+            "call_id": call.call_id,
+            "is_error": result.is_error,
+            "call": call,
+            "result": result,
+        }
+        self._emit("tool_completed", turn_id, **payload)
+        return result
+
+    def _emit(self, kind: str, turn_id: str, **payload: object) -> None:
+        self._event_handler(
+            AgentEvent(kind=kind, turn_id=turn_id, payload=dict(payload))
+        )
+
+    def _handle_model_stream_event(self, turn_id: str, event: ModelStreamEvent) -> None:
+        if event.kind == "assistant_delta":
+            self._emit("assistant_delta", turn_id, **event.payload)
+        elif event.kind == "tool_call":
+            self._emit("tool_called", turn_id, **event.payload)
+
+    def _build_follow_up_messages(
+        self,
+        tool_results: list[ToolResult],
+    ) -> list[UserMessage]:
+        follow_ups: list[UserMessage] = []
+        for result in tool_results:
+            statuses = None
+            if (
+                result.name == "wait_agent"
+                and not result.is_error
+                and isinstance(result.output, dict)
+            ):
+                statuses = result.output.get("status")
+            if isinstance(statuses, dict):
+                for agent_id, status in statuses.items():
+                    if isinstance(agent_id, str) and isinstance(status, dict):
+                        payload = {
+                            "agent_id": agent_id,
+                            "status": status,
+                        }
+                        follow_ups.append(
+                            UserMessage(
+                                text=(
+                                    "<subagent_notification>\n"
+                                    f"{json.dumps(payload, ensure_ascii=False, separators=(',', ':'))}\n"
+                                    "</subagent_notification>"
+                                )
+                            )
+                        )
+        return follow_ups