fleet-python 0.2.1__py3-none-any.whl → 0.2.3__py3-none-any.whl

fleet/instance/client.py

@@ -0,0 +1,295 @@
+"""Fleet SDK Base Environment Classes."""
+
+from typing import Any, Callable, Dict, List, Optional, Tuple
+import asyncio
+import httpx
+import inspect
+import time
+import logging
+from urllib.parse import urlparse
+
+from ..resources.sqlite import AsyncSQLiteResource
+from ..resources.browser import AsyncBrowserResource
+from ..resources.base import Resource
+
+from ..verifiers import DatabaseSnapshot
+
+from ..exceptions import FleetEnvironmentError, FleetAPIError
+
+from .base import SyncWrapper, AsyncWrapper
+from .models import (
+    ResetRequest,
+    ResetResponse,
+    Resource as ResourceModel,
+    ResourceType,
+    HealthResponse,
+    ExecuteFunctionRequest,
+    ExecuteFunctionResponse,
+)
+
+
+logger = logging.getLogger(__name__)
+
+
+RESOURCE_TYPES = {
+    ResourceType.db: AsyncSQLiteResource,
+    ResourceType.cdp: AsyncBrowserResource,
+}
+
+ValidatorType = Callable[
+    [DatabaseSnapshot, DatabaseSnapshot, Optional[str]],
+    int,
+]
+
+
+class InstanceClient:
+    def __init__(
+        self,
+        url: str,
+        httpx_client: Optional[httpx.Client] = None,
+    ):
+        self.base_url = url
+        self.client = SyncWrapper(
+            url=self.base_url, httpx_client=httpx_client or httpx.Client()
+        )
+        raise NotImplementedError("SyncManager is not implemented")
+
+    def reset(self) -> ResetResponse:
+        response = self.client.request("POST", "/reset")
+        return ResetResponse(**response.json())
+
+
+class AsyncInstanceClient:
+    def __init__(
+        self,
+        url: str,
+        httpx_client: Optional[httpx.AsyncClient] = None,
+    ):
+        self.base_url = url
+        self.client = AsyncWrapper(
+            url=self.base_url,
+            httpx_client=httpx_client or httpx.AsyncClient(timeout=60.0),
+        )
+        self._resources: Optional[List[ResourceModel]] = None
+        self._resources_state: Dict[str, Dict[str, Resource]] = {
+            resource_type.value: {} for resource_type in ResourceType
+        }
+
+    async def load(self) -> None:
+        await self._load_resources()
+
+    async def reset(
+        self, reset_request: Optional[ResetRequest] = None
+    ) -> ResetResponse:
+        response = await self.client.request(
+            "POST", "/reset", json=reset_request.model_dump() if reset_request else None
+        )
+        return ResetResponse(**response.json())
+
+    def state(self, uri: str) -> Resource:
+        url = urlparse(uri)
+        return self._resources_state[url.scheme][url.netloc]
+
+    def db(self, name: str) -> AsyncSQLiteResource:
+        """
+        Returns an AsyncSQLiteResource object for the given SQLite database name.
+
+        Args:
+            name: The name of the SQLite database to return
+
+        Returns:
+            An AsyncSQLiteResource object for the given SQLite database name
+        """
+        return AsyncSQLiteResource(
+            self._resources_state[ResourceType.db.value][name], self.client
+        )
+
+    def browser(self, name: str) -> AsyncBrowserResource:
+        return AsyncBrowserResource(
+            self._resources_state[ResourceType.cdp.value][name], self.client
+        )
+
+    async def resources(self) -> List[Resource]:
+        await self._load_resources()
+        return [
+            resource
+            for resources_by_name in self._resources_state.values()
+            for resource in resources_by_name.values()
+        ]
+
+    async def verify(self, validator: ValidatorType) -> ExecuteFunctionResponse:
+        function_code = inspect.getsource(validator)
+        function_name = validator.__name__
+        return await self.verify_raw(function_code, function_name)
+
+    async def verify_raw(
+        self, function_code: str, function_name: str
+    ) -> ExecuteFunctionResponse:
+        response = await self.client.request(
+            "POST",
+            "/execute_verifier_function",
+            json=ExecuteFunctionRequest(
+                function_code=function_code,
+                function_name=function_name,
+            ).model_dump(),
+        )
+        return ExecuteFunctionResponse(**response.json())
+
+    async def _load_resources(self) -> None:
+        if self._resources is None:
+            response = await self.client.request("GET", "/resources")
+            if response.status_code != 200:
+                self._resources = []
+                return
+
+            # Handle both old and new response formats
+            response_data = response.json()
+            if isinstance(response_data, dict) and "resources" in response_data:
+                # Old format: {"resources": [...]}
+                resources_list = response_data["resources"]
+            else:
+                # New format: [...]
+                resources_list = response_data
+
+            self._resources = [ResourceModel(**resource) for resource in resources_list]
+            for resource in self._resources:
+                if resource.type not in self._resources_state:
+                    self._resources_state[resource.type.value] = {}
+                self._resources_state[resource.type.value][resource.name] = (
+                    RESOURCE_TYPES[resource.type](resource, self.client)
+                )
+
+    async def step(self, action: Dict[str, Any]) -> Tuple[Dict[str, Any], float, bool]:
+        """Execute one step in the environment."""
+        if not self._instance_id:
+            raise FleetEnvironmentError(
+                "Environment not initialized. Call reset() first."
+            )
+
+        try:
+            # Increment step count
+            self._increment_step()
+
+            # Execute action through instance manager API
+            # This is a placeholder - actual implementation depends on the manager API spec
+            state, reward, done = await self._execute_action(action)
+
+            return state, reward, done
+
+        except Exception as e:
+            raise FleetEnvironmentError(f"Failed to execute step: {e}")
+
+    async def close(self) -> None:
+        """Close the environment and clean up resources."""
+        try:
+            # Delete instance if it exists
+            if self._instance_id:
+                try:
+                    await self._client.delete_instance(self._instance_id)
+                    logger.info(f"Deleted instance: {self._instance_id}")
+                except FleetAPIError as e:
+                    logger.warning(f"Failed to delete instance: {e}")
+                finally:
+                    self._instance_id = None
+                    self._instance_response = None
+
+            # Close manager client
+            if self._manager_client:
+                await self._manager_client.close()
+                self._manager_client = None
+
+            # Close API client
+            await self._client.close()
+
+        except Exception as e:
+            logger.error(f"Error closing environment: {e}")
+
+    async def manager_health_check(self) -> Optional[HealthResponse]:
+        response = await self.client.request("GET", "/health")
+        return HealthResponse(**response.json())
+
+    async def _wait_for_instance_ready(self, timeout: float = 300.0) -> None:
+        """Wait for instance to be ready.
+
+        Args:
+            timeout: Maximum time to wait in seconds
+        """
+        start_time = time.time()
+
+        while time.time() - start_time < timeout:
+            try:
+                instance = await self._client.get_instance(self._instance_id)
+                self._instance_response = instance
+
+                if instance.status == "running":
+                    logger.info(f"Instance {self._instance_id} is ready")
+                    return
+
+                elif instance.status == "error":
+                    raise FleetEnvironmentError(
+                        f"Instance {self._instance_id} failed to start"
+                    )
+
+                # Wait before checking again
+                await asyncio.sleep(5)
+
+            except FleetAPIError as e:
+                if time.time() - start_time >= timeout:
+                    raise FleetEnvironmentError(
+                        f"Timeout waiting for instance to be ready: {e}"
+                    )
+                await asyncio.sleep(5)
+
+        raise FleetEnvironmentError(
+            f"Timeout waiting for instance {self._instance_id} to be ready"
+        )
+
+    async def _execute_action(
+        self, action: Dict[str, Any]
+    ) -> Tuple[Dict[str, Any], float, bool]:
+        """Execute an action through the instance manager API.
+
+        This is a placeholder implementation that should be extended based on
+        the actual manager API specification.
+
+        Args:
+            action: The action to execute as a dictionary
+
+        Returns:
+            Tuple of (state, reward, done)
+        """
+        # Ensure manager client is available
+        await self._ensure_manager_client()
+
+        # TODO: In the future, this would use the manager API to execute actions
+        # For example: await self._manager_client.log_action(action)
+        # For now, return placeholder values
+
+        # Create a placeholder state
+        state = self._create_state_from_action(action)
+
+        # Create a placeholder reward
+        reward = 0.0
+
+        # Determine if episode is done (placeholder logic)
+        done = self._step_count >= 100  # Example: done after 100 steps
+
+        return state, reward, done
+
+    def _create_state_from_action(self, action: Dict[str, Any]) -> Dict[str, Any]:
+        """Create state based on executed action."""
+        return {
+            "instance_id": self._instance_id,
+            "step": self._step_count,
+            "last_action": action,
+            "timestamp": time.time(),
+            "status": "running",
+        }
+
+    async def __aenter__(self):
+        """Async context manager entry."""
+        return self
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        """Async context manager exit."""
+        await self.close()

fleet/{env → instance}/models.py

@@ -126,3 +126,16 @@ class Resource(BaseModel):
     type: ResourceType
     mode: ResourceMode
     label: Optional[str] = Field(None, title="Label")
+
+
+class ExecuteFunctionRequest(BaseModel):
+    function_code: str
+    function_name: str
+    text_solution: Optional[str] = None
+
+
+class ExecuteFunctionResponse(BaseModel):
+    success: bool
+    result: Optional[Any] = None
+    error: Optional[str] = None
+    message: str

fleet/playwright.py

@@ -0,0 +1,291 @@
+import base64
+from typing import List, Dict, Any
+from playwright.async_api import async_playwright, Browser, Page
+from .client import AsyncEnvironment
+
+
+# Key mapping for computer use actions
+CUA_KEY_TO_PLAYWRIGHT_KEY = {
+    "/": "Divide",
+    "\\": "Backslash",
+    "alt": "Alt",
+    "arrowdown": "ArrowDown",
+    "arrowleft": "ArrowLeft",
+    "arrowright": "ArrowRight",
+    "arrowup": "ArrowUp",
+    "backspace": "Backspace",
+    "capslock": "CapsLock",
+    "cmd": "Meta",
+    "ctrl": "Control",
+    "delete": "Delete",
+    "end": "End",
+    "enter": "Enter",
+    "esc": "Escape",
+    "home": "Home",
+    "insert": "Insert",
+    "option": "Alt",
+    "pagedown": "PageDown",
+    "pageup": "PageUp",
+    "shift": "Shift",
+    "space": " ",
+    "super": "Meta",
+    "tab": "Tab",
+    "win": "Meta",
+}
+
+
+class FleetPlaywrightWrapper:
+    """
+    A wrapper that adds Playwright browser automation to Fleet environment instances.
+
+    This class handles:
+    - Browser connection via CDP
+    - Computer actions (click, scroll, type, etc.)
+    - Screenshot capture
+    - Integration with OpenAI computer use API
+
+    Usage:
+        instance = await fleet.env.make(env_key="hubspot", version="v1.2.7")
+        browser = FleetPlaywrightWrapper(instance)
+        await browser.start()
+
+        # Use browser methods
+        screenshot = await browser.screenshot()
+        tools = [browser.openai_cua_tool]
+
+        # Clean up when done
+        await browser.close()
+    """
+
+    def get_environment(self):
+        return "browser"
+
+    def get_dimensions(self):
+        return (1920, 1080)
+
+    def __init__(
+        self,
+        env: AsyncEnvironment,
+        display_width: int = 1920,
+        display_height: int = 1080,
+    ):
+        """
+        Initialize the Fleet Playwright wrapper.
+
+        Args:
+            env: Fleet environment instance
+            display_width: Browser viewport width
+            display_height: Browser viewport height
+        """
+        self.env = env
+        self.display_width = display_width
+        self.display_height = display_height
+
+        self._playwright = None
+        self._browser: Browser | None = None
+        self._page: Page | None = None
+        self._started = False
+
+    async def start(self):
+        """Start the browser and establish connection."""
+        if self._started:
+            return
+
+        # Start Playwright
+        self._playwright = await async_playwright().start()
+
+        # Start browser on the Fleet instance
+        print("Starting browser...")
+        await self.env.browser().start()
+        cdp = await self.env.browser().describe()
+
+        # Connect to browser
+        self._browser = await self._playwright.chromium.connect_over_cdp(
+            cdp.cdp_browser_url
+        )
+        self._page = self._browser.contexts[0].pages[0]
+        await self._page.set_viewport_size(
+            {"width": self.display_width, "height": self.display_height}
+        )
+
+        self._started = True
+        print(f"Track agent: {cdp.cdp_devtools_url}")
+
+    async def close(self):
+        """Close the browser connection."""
+        if self._playwright:
+            await self._playwright.stop()
+            self._playwright = None
+            self._browser = None
+            self._page = None
+            self._started = False
+
+    def _ensure_started(self):
+        """Ensure browser is started before operations."""
+        if not self._started:
+            raise RuntimeError("Browser not started. Call await browser.start() first.")
+
+    @property
+    def openai_cua_tool(self) -> Dict[str, Any]:
+        """
+        Tool definition for OpenAI computer use API.
+
+        Returns:
+            Tool definition dict for use with OpenAI responses API
+        """
+        return {
+            "type": "computer_use_preview",
+            "display_width": self.display_width,
+            "display_height": self.display_height,
+            "environment": "browser",
+        }
+
+    async def screenshot(self) -> str:
+        """
+        Take a screenshot and return base64 encoded string.
+
+        Returns:
+            Base64 encoded PNG screenshot
+        """
+        self._ensure_started()
+
+        png_bytes = await self._page.screenshot(full_page=False)
+        return base64.b64encode(png_bytes).decode("utf-8")
+
+    def get_current_url(self) -> str:
+        """Get the current page URL."""
+        self._ensure_started()
+        return self._page.url
+
+    async def execute_computer_action(self, action: Dict[str, Any]) -> Dict[str, Any]:
+        """
+        Execute a computer action and return the result for OpenAI API.
+
+        Args:
+            action: Computer action dict from OpenAI response
+
+        Returns:
+            Result dict for computer_call_output
+        """
+        self._ensure_started()
+
+        action_type = action["type"]
+        action_args = {k: v for k, v in action.items() if k != "type"}
+
+        print(f"Executing: {action_type}({action_args})")
+
+        # Execute the action
+        if hasattr(self, f"_{action_type}"):
+            method = getattr(self, f"_{action_type}")
+            await method(**action_args)
+        else:
+            raise ValueError(f"Unsupported action type: {action_type}")
+
+        # Take screenshot after action
+        screenshot_base64 = await self.screenshot()
+
+        return {
+            "type": "input_image",
+            "image_url": f"data:image/png;base64,{screenshot_base64}",
+            "current_url": self.get_current_url(),
+        }
+
+    # Computer action implementations
+    async def _click(self, x: int, y: int, button: str = "left") -> None:
+        """Click at coordinates."""
+        self._ensure_started()
+        await self._page.mouse.click(x, y, button=button)
+
+    async def _double_click(self, x: int, y: int) -> None:
+        """Double-click at coordinates."""
+        self._ensure_started()
+        await self._page.mouse.dblclick(x, y)
+
+    async def _scroll(self, x: int, y: int, scroll_x: int, scroll_y: int) -> None:
+        """Scroll from coordinates."""
+        self._ensure_started()
+        await self._page.mouse.move(x, y)
+        await self._page.evaluate(f"window.scrollBy({scroll_x}, {scroll_y})")
+
+    async def _type(self, text: str) -> None:
+        """Type text."""
+        self._ensure_started()
+        await self._page.keyboard.type(text)
+
+    async def _keypress(self, keys: List[str]) -> None:
+        """Press key combination."""
+        self._ensure_started()
+        mapped_keys = [CUA_KEY_TO_PLAYWRIGHT_KEY.get(key.lower(), key) for key in keys]
+        for key in mapped_keys:
+            await self._page.keyboard.down(key)
+        for key in reversed(mapped_keys):
+            await self._page.keyboard.up(key)
+
+    async def _move(self, x: int, y: int) -> None:
+        """Move mouse to coordinates."""
+        self._ensure_started()
+        await self._page.mouse.move(x, y)
+
+    async def _drag(self, path: List[Dict[str, int]]) -> None:
+        """Drag mouse along path."""
+        self._ensure_started()
+        if not path:
+            return
+        await self._page.mouse.move(path[0]["x"], path[0]["y"])
+        await self._page.mouse.down()
+        for point in path[1:]:
+            await self._page.mouse.move(point["x"], point["y"])
+        await self._page.mouse.up()
+
+    async def _wait(self, ms: int = 1000) -> None:
+        """Wait for specified milliseconds."""
+        import asyncio
+
+        await asyncio.sleep(ms / 1000)
+
+    # Browser-specific actions
+    async def _goto(self, url: str) -> None:
+        """Navigate to URL."""
+        self._ensure_started()
+        try:
+            await self._page.goto(url)
+        except Exception as e:
+            print(f"Error navigating to {url}: {e}")
+
+    async def _back(self) -> None:
+        """Go back in browser history."""
+        self._ensure_started()
+        await self._page.go_back()
+
+    async def _forward(self) -> None:
+        """Go forward in browser history."""
+        self._ensure_started()
+        await self._page.go_forward()
+
+    async def _refresh(self) -> None:
+        """Refresh the page."""
+        self._ensure_started()
+        await self._page.reload()
+
+    # ------------------------------------------------------------------
+    # Public aliases (no leading underscore) expected by the Agent &
+    # OpenAI computer-use API. They forward directly to the underscored
+    # implementations above so the external interface matches the older
+    # BasePlaywrightComputer class.
+    # ------------------------------------------------------------------
+
+    # Mouse / keyboard actions
+    click = _click
+    double_click = _double_click
+    scroll = _scroll
+    type = _type  # noqa: A003 – shadowing built-in for API compatibility
+    keypress = _keypress
+    move = _move
+    drag = _drag
+    wait = _wait
+
+    # Browser navigation actions
+    goto = _goto
+    back = _back
+    forward = _forward
+    refresh = _refresh

fleet/resources/base.py

@@ -1,5 +1,5 @@
 from abc import ABC
-from ..env.models import Resource as ResourceModel, ResourceType, ResourceMode
+from ..instance.models import Resource as ResourceModel, ResourceType, ResourceMode
 
 
 class Resource(ABC):
@@ -8,7 +8,7 @@ class Resource(ABC):
 
     @property
     def uri(self) -> str:
-        return f"{self.resource.type}://{self.resource.name}"
+        return f"{self.resource.type.value}://{self.resource.name}"
 
     @property
     def name(self) -> str:
@@ -21,3 +21,6 @@ class Resource(ABC):
     @property
     def mode(self) -> ResourceMode:
         return self.resource.mode
+    
+    def __repr__(self) -> str:
+        return f"Resource(uri={self.uri}, mode={self.mode.value})"

fleet/resources/browser.py

@@ -1,6 +1,5 @@
 from typing import Optional
-
-from ..env.models import (
+from ..instance.models import (
     Resource as ResourceModel,
     CDPDescribeResponse,
     ChromeStartRequest,
@@ -11,7 +10,7 @@ from .base import Resource
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
-    from ..env.base import AsyncWrapper
+    from ..instance.base import AsyncWrapper
 
 
 class AsyncBrowserResource(Resource):
@@ -19,16 +18,24 @@ class AsyncBrowserResource(Resource):
         super().__init__(resource)
         self.client = client
 
-    async def start(
-        self, start_request: Optional[ChromeStartRequest] = None
-    ) -> ChromeStartResponse:
+    async def start(self, width: int = 1920, height: int = 1080) -> CDPDescribeResponse:
         response = await self.client.request(
             "POST",
             "/resources/cdp/start",
-            json=start_request.model_dump() if start_request else None,
+            json=ChromeStartRequest(resolution=f"{width},{height}").model_dump(),
         )
-        return ChromeStartResponse(**response.json())
+        ChromeStartResponse(**response.json())
+        return await self.describe()
 
     async def describe(self) -> CDPDescribeResponse:
         response = await self.client.request("GET", "/resources/cdp/describe")
+        if response.status_code != 200:
+            await self.start()
+            response = await self.client.request("GET", "/resources/cdp/describe")
         return CDPDescribeResponse(**response.json())
+
+    async def cdp_url(self) -> str:
+        return (await self.describe()).cdp_browser_url
+
+    async def devtools_url(self) -> str:
+        return (await self.describe()).cdp_devtools_url

fleet/resources/sqlite.py

@@ -1,12 +1,12 @@
 from typing import Any, List, Optional
-from ..env.models import Resource as ResourceModel
-from ..env.models import DescribeResponse, QueryRequest, QueryResponse
+from ..instance.models import Resource as ResourceModel
+from ..instance.models import DescribeResponse, QueryRequest, QueryResponse
 from .base import Resource
 
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
-    from ..env.base import AsyncWrapper
+    from ..instance.base import AsyncWrapper
 
 
 class AsyncSQLiteResource(Resource):