🌿 Fern Regeneration -- March 4, 2025

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "scrapybara"
-version = "2.2.9"
+version = "2.3.0"
 description = ""
 readme = "README.md"
 authors = []
@@ -32,9 +32,7 @@ Repository = 'https://github.com/scrapybara/scrapybara-python'
 
 [tool.poetry.dependencies]
 python = "^3.8"
-anthropic = "^0.47.2"
 httpx = ">=0.21.2"
-playwright = "^1.48.0"
 pydantic = ">= 1.9.2"
 pydantic-core = "^2.18.2"
 typing_extensions = ">= 4.0.0"

reference.md

@@ -336,13 +336,16 @@ client.instance.get_stream_url(
 
 ```python
 from scrapybara import Scrapybara
+from scrapybara.instance import Request_MoveMouse
 
 client = Scrapybara(
     api_key="YOUR_API_KEY",
 )
 client.instance.computer(
     instance_id="instance_id",
-    action="key",
+    request=Request_MoveMouse(
+        coordinates=[1],
+    ),
 )
 
 ```
@@ -367,23 +370,7 @@ client.instance.computer(
 <dl>
 <dd>
 
-**action:** `Action` 
-
-</dd>
-</dl>
-
-<dl>
-<dd>
-
-**coordinate:** `typing.Optional[typing.Sequence[int]]` 
-
-</dd>
-</dl>
-
-<dl>
-<dd>
-
-**text:** `typing.Optional[str]` 
+**request:** `Request` 
 
 </dd>
 </dl>

requirements.txt

@@ -0,0 +1,4 @@
+httpx>=0.21.2
+pydantic>= 1.9.2
+pydantic-core==^2.18.2
+typing_extensions>= 4.0.0

src/scrapybara/__init__.py

@@ -2,72 +2,121 @@
 
 from .types import (
     AuthStateResponse,
+    BashResponse,
     BrowserAuthenticateResponse,
     BrowserGetCdpUrlResponse,
+    Button,
     CellType,
+    ClickMouseAction,
+    ClickMouseActionClickType,
+    ComputerResponse,
     DeploymentConfigInstanceType,
+    DragMouseAction,
+    EditResponse,
     EnvGetResponse,
     EnvResponse,
     ExecuteCellRequest,
     FileDownloadResponse,
     FileReadResponse,
+    GetCursorPositionAction,
     GetInstanceResponse,
     GetInstanceResponseInstanceType,
     HttpValidationError,
     InstanceGetStreamUrlResponse,
     InstanceScreenshotResponse,
     KernelInfo,
     ModifyBrowserAuthResponse,
+    MoveMouseAction,
     Notebook,
     NotebookCell,
+    PressKeyAction,
     SaveBrowserAuthResponse,
+    ScrollAction,
     StartBrowserResponse,
     Status,
     StopBrowserResponse,
     StopInstanceResponse,
+    TakeScreenshotAction,
+    TypeTextAction,
     ValidationError,
     ValidationErrorLocItem,
+    WaitAction,
 )
 from .errors import UnprocessableEntityError
 from . import browser, code, env, file, instance, notebook
 from .client import AsyncScrapybara, Scrapybara
 from .environment import ScrapybaraEnvironment
-from .instance import Action, Command
+from .instance import (
+    Command,
+    Request,
+    Request_ClickMouse,
+    Request_DragMouse,
+    Request_GetCursorPosition,
+    Request_MoveMouse,
+    Request_PressKey,
+    Request_Scroll,
+    Request_TakeScreenshot,
+    Request_TypeText,
+    Request_Wait,
+)
 from .version import __version__
 
 __all__ = [
-    "Action",
     "AsyncScrapybara",
     "AuthStateResponse",
+    "BashResponse",
     "BrowserAuthenticateResponse",
     "BrowserGetCdpUrlResponse",
+    "Button",
     "CellType",
+    "ClickMouseAction",
+    "ClickMouseActionClickType",
     "Command",
+    "ComputerResponse",
     "DeploymentConfigInstanceType",
+    "DragMouseAction",
+    "EditResponse",
     "EnvGetResponse",
     "EnvResponse",
     "ExecuteCellRequest",
     "FileDownloadResponse",
     "FileReadResponse",
+    "GetCursorPositionAction",
     "GetInstanceResponse",
     "GetInstanceResponseInstanceType",
     "HttpValidationError",
     "InstanceGetStreamUrlResponse",
     "InstanceScreenshotResponse",
     "KernelInfo",
     "ModifyBrowserAuthResponse",
+    "MoveMouseAction",
     "Notebook",
     "NotebookCell",
+    "PressKeyAction",
+    "Request",
+    "Request_ClickMouse",
+    "Request_DragMouse",
+    "Request_GetCursorPosition",
+    "Request_MoveMouse",
+    "Request_PressKey",
+    "Request_Scroll",
+    "Request_TakeScreenshot",
+    "Request_TypeText",
+    "Request_Wait",
     "SaveBrowserAuthResponse",
     "Scrapybara",
     "ScrapybaraEnvironment",
+    "ScrollAction",
     "StartBrowserResponse",
     "Status",
     "StopBrowserResponse",
     "StopInstanceResponse",
+    "TakeScreenshotAction",
+    "TypeTextAction",
     "UnprocessableEntityError",
     "ValidationError",
     "ValidationErrorLocItem",
+    "WaitAction",
     "__version__",
     "browser",
     "code",

src/scrapybara/anthropic/__init__.py

@@ -1,18 +1,10 @@
-from typing import Literal, Optional, TypedDict, Any, Dict
-from anthropic.types.beta import (
-    BetaToolComputerUse20241022Param,
-    BetaToolTextEditor20241022Param,
-    BetaToolBash20241022Param,
-)
-import asyncio
+from typing import Literal, Optional
+
 from pydantic import Field
 
-from ..client import BaseInstance, UbuntuInstance
 from ..types.act import Model
-from .base import BaseAnthropicTool, CLIResult, ToolError, ToolResult
 
 
-# New: universal act API
 class Anthropic(Model):
     """Model adapter for Anthropic.
 
@@ -37,179 +29,3 @@ def __init__(
         api_key: Optional[str] = None,
     ) -> None:
         super().__init__(provider="anthropic", name=name, api_key=api_key)
-
-
-# Legacy: Anthropic SDK-compatible tools
-class ComputerToolOptions(TypedDict):
-    display_height_px: int
-    display_width_px: int
-    display_number: Optional[int]
-
-
-class ComputerTool(BaseAnthropicTool):
-    """A computer interaction tool that allows the agent to control mouse and keyboard.
-
-    Available for Ubuntu, Browser, and Windows instances."""
-
-    api_type: Literal["computer_20241022"] = "computer_20241022"
-    name: Literal["computer"] = "computer"
-    width: int = 1024
-    height: int = 768
-    display_num: Optional[int] = 1
-    _instance: BaseInstance
-
-    def __init__(self, instance: BaseInstance):
-        self._instance = instance
-        super().__init__()
-
-    @property
-    def options(self) -> ComputerToolOptions:
-        return {
-            "display_width_px": self.width,
-            "display_height_px": self.height,
-            "display_number": self.display_num,
-        }
-
-    def to_params(self) -> BetaToolComputerUse20241022Param:
-        return {
-            "name": self.name,
-            "type": self.api_type,
-            "display_width_px": self.width,
-            "display_height_px": self.height,
-            "display_number": self.display_num,
-        }
-
-    async def __call__(self, **kwargs: Any) -> ToolResult:
-        action = kwargs.pop("action")
-        coordinate = kwargs.pop("coordinate", None)
-        text = kwargs.pop("text", None)
-
-        try:
-            loop = asyncio.get_event_loop()
-            result = await loop.run_in_executor(
-                None,
-                lambda: self._instance.computer(
-                    action=action,
-                    coordinate=tuple(coordinate) if coordinate else None,
-                    text=text,
-                ),
-            )
-            return CLIResult(
-                output=result.get("output") if result else "",
-                error=result.get("error") if result else None,
-                base64_image=result.get("base64_image") if result else None,
-                system=result.get("system") if result else None,
-            )
-        except Exception as e:
-            raise ToolError(str(e)) from None
-
-
-class EditTool(BaseAnthropicTool):
-    """A filesystem editor tool that allows the agent to view, create, and edit files.
-
-    Available for Ubuntu instances."""
-
-    api_type: Literal["text_editor_20241022"] = "text_editor_20241022"
-    name: Literal["str_replace_editor"] = "str_replace_editor"
-    _instance: UbuntuInstance
-
-    def __init__(self, instance: UbuntuInstance):
-        self._instance = instance
-        super().__init__()
-
-    def to_params(self) -> BetaToolTextEditor20241022Param:
-        return {
-            "name": self.name,
-            "type": self.api_type,
-        }
-
-    async def __call__(self, **kwargs: Any) -> ToolResult:
-        command = kwargs.pop("command")
-        path = kwargs.pop("path")
-        file_text = kwargs.pop("file_text", None)
-        view_range = kwargs.pop("view_range", None)
-        old_str = kwargs.pop("old_str", None)
-        new_str = kwargs.pop("new_str", None)
-        insert_line = kwargs.pop("insert_line", None)
-        try:
-            loop = asyncio.get_event_loop()
-            result = await loop.run_in_executor(
-                None,
-                lambda: self._instance.edit(
-                    command=command,
-                    path=path,
-                    file_text=file_text,
-                    view_range=view_range,
-                    old_str=old_str,
-                    new_str=new_str,
-                    insert_line=insert_line,
-                ),
-            )
-            return CLIResult(
-                output=result.get("output") if result else "",
-                error=result.get("error") if result else None,
-                base64_image=result.get("base64_image") if result else None,
-                system=result.get("system") if result else None,
-            )
-        except Exception as e:
-            raise ToolError(str(e)) from None
-
-
-class BashTool(BaseAnthropicTool):
-    """A shell execution tool that allows the agent to run bash commands.
-
-    Available for Ubuntu instances."""
-
-    api_type: Literal["bash_20241022"] = "bash_20241022"
-    name: Literal["bash"] = "bash"
-    _instance: UbuntuInstance
-
-    def __init__(self, instance: UbuntuInstance):
-        self._instance = instance
-        super().__init__()
-
-    def to_params(self) -> BetaToolBash20241022Param:
-        return {
-            "name": self.name,
-            "type": self.api_type,
-        }
-
-    async def __call__(self, **kwargs: Any) -> ToolResult:
-        command = kwargs.pop("command")
-        restart = kwargs.pop("restart", False)
-        try:
-            loop = asyncio.get_event_loop()
-            result = await loop.run_in_executor(
-                None,
-                lambda: self._instance.bash(command=command, restart=restart),
-            )
-            return CLIResult(
-                output=result.get("output") if result else "",
-                error=result.get("error") if result else None,
-                base64_image=result.get("base64_image") if result else None,
-                system=result.get("system") if result else None,
-            )
-        except Exception as e:
-            raise ToolError(str(e)) from None
-
-
-class ToolCollection:
-    """A collection of anthropic-defined tools."""
-
-    def __init__(self, *tools):
-        self.tools = tools
-        self.tool_map = {tool.to_params()["name"]: tool for tool in tools}
-
-    def to_params(self) -> list:
-        return [tool.to_params() for tool in self.tools]
-
-    async def run(self, *, name: str, tool_input: Dict[str, Any]) -> ToolResult:
-        tool = self.tool_map.get(name)
-        if not tool:
-            return ToolResult(error=f"Tool {name} not found")
-        try:
-            r = await tool(**tool_input)
-            return r if r else ToolResult()
-        except Exception as e:
-            print(f"Error running tool {name}: {e}")
-            return ToolResult(error=str(e))