mirror of
https://github.com/Tencent/BrowserSkill.git
synced 2026-09-10 13:01:23 +08:00
feat: support Agent Window sizing (session start --width/--height, window resize) (#52)
- `bsk session start --width <px> --height <px>` opens the Agent Window with the given outer size (chrome.windows.create) - new `bsk window resize --session <id> --width <px> --height <px>` resizes the Agent Window at runtime (chrome.windows.update) - both validated on CLI, daemon and extension sides (100..=7680 CSS px) - protocol: new tool.window_resize method (BrowserMutation) and optional width/height on tool.session_start params; schemas regenerated
This commit is contained in:
@@ -39,6 +39,13 @@ describe("SessionManager", () => {
|
||||
expect(ctx.refStore.isEmpty()).toBe(true);
|
||||
expect(ctx.borrowedTabs.size).toBe(0);
|
||||
});
|
||||
it("forwards an optional window size when starting a session", async () => {
|
||||
const aw = fakeAgentWindow();
|
||||
const sm = new SessionManager({ agentWindow: aw });
|
||||
const ctx = await sm.start("aa11", { width: 1280, height: 800 });
|
||||
expect(aw.createMock).toHaveBeenCalledWith("about:blank", { width: 1280, height: 800 });
|
||||
expect(ctx.agentWindowId).toBe(100);
|
||||
});
|
||||
|
||||
it("indexes the session by sessionId and agent window id", async () => {
|
||||
const aw = fakeAgentWindow();
|
||||
|
||||
@@ -8,7 +8,7 @@
|
||||
*/
|
||||
|
||||
export interface AgentWindowApi {
|
||||
create(url: string): Promise<number>;
|
||||
create(url: string, size?: { width: number; height: number }): Promise<number>;
|
||||
remove(windowId: number): Promise<void>;
|
||||
/**
|
||||
* Guarantee the Agent Window has an active, CDP-navigable tab.
|
||||
@@ -22,11 +22,12 @@ export interface AgentWindowApi {
|
||||
export const AGENT_WINDOW_HOME = "about:blank";
|
||||
|
||||
export const chromeAgentWindowApi: AgentWindowApi = {
|
||||
async create(url: string): Promise<number> {
|
||||
async create(url: string, size?: { width: number; height: number }): Promise<number> {
|
||||
const win = await chrome.windows.create({
|
||||
type: "normal",
|
||||
focused: true,
|
||||
url,
|
||||
...(size ? { width: size.width, height: size.height } : {}),
|
||||
});
|
||||
if (typeof win?.id !== "number") {
|
||||
throw new Error("[bh] chrome.windows.create returned no window id");
|
||||
|
||||
@@ -128,11 +128,18 @@ export class SessionManager {
|
||||
* Returns the created window id so callers can echo it back to the
|
||||
* daemon in the `tool.session_start` reply.
|
||||
*/
|
||||
async start(sessionId: string): Promise<SessionContext> {
|
||||
async start(
|
||||
sessionId: string,
|
||||
size?: { width: number; height: number },
|
||||
): Promise<SessionContext> {
|
||||
if (this.sessions.has(sessionId)) {
|
||||
throw new Error(`[bh] session ${sessionId} already exists`);
|
||||
}
|
||||
const windowId = await this.agentWindow.create(AGENT_WINDOW_HOME);
|
||||
// Only pass `size` when given so the no-size call shape (and its
|
||||
// chrome.windows.create payload) stays exactly as before.
|
||||
const windowId = size
|
||||
? await this.agentWindow.create(AGENT_WINDOW_HOME, size)
|
||||
: await this.agentWindow.create(AGENT_WINDOW_HOME);
|
||||
await this.agentWindow.ensureActiveTab(windowId, AGENT_WINDOW_HOME);
|
||||
const ctx: SessionContext = {
|
||||
sessionId,
|
||||
|
||||
+138
@@ -0,0 +1,138 @@
|
||||
import { describe, expect, it, vi } from "vitest";
|
||||
import { SessionManager } from "@/session-manager/manager";
|
||||
import { handleSessionStart } from "../session";
|
||||
import { handleWindowResize, type WindowResizeApi } from "../window";
|
||||
|
||||
function fakeAgentWindow(ids: number[]) {
|
||||
let i = 0;
|
||||
const create = vi.fn(async () => {
|
||||
const id = ids[i++];
|
||||
if (id === undefined) throw new Error("ran out of fake ids");
|
||||
return id;
|
||||
});
|
||||
const remove = vi.fn(async () => {});
|
||||
const ensureActiveTab = vi.fn(async () => {});
|
||||
return { create, remove, ensureActiveTab };
|
||||
}
|
||||
|
||||
async function makeManager(): Promise<SessionManager> {
|
||||
const sm = new SessionManager({ agentWindow: fakeAgentWindow([100]) });
|
||||
await sm.start("aa11");
|
||||
return sm;
|
||||
}
|
||||
|
||||
function fakeUpdateApi(opts?: { throw?: boolean }) {
|
||||
const calls: Array<{ windowId: number; width: number; height: number }> = [];
|
||||
const api: WindowResizeApi = {
|
||||
update: vi.fn(async (windowId: number, updateInfo: { width: number; height: number }) => {
|
||||
if (opts?.throw) throw new Error("simulated chrome failure");
|
||||
calls.push({ windowId, width: updateInfo.width, height: updateInfo.height });
|
||||
return undefined;
|
||||
}),
|
||||
};
|
||||
return { api, calls };
|
||||
}
|
||||
|
||||
describe("handleWindowResize", () => {
|
||||
it("resizes the session's Agent Window", async () => {
|
||||
const sm = await makeManager();
|
||||
const { api, calls } = fakeUpdateApi();
|
||||
const result = await handleWindowResize(
|
||||
sm,
|
||||
{ session_id: "aa11", width: 1280, height: 800 },
|
||||
api,
|
||||
);
|
||||
expect(result).toEqual({ window_id: 100, width: 1280, height: 800 });
|
||||
expect(calls).toEqual([{ windowId: 100, width: 1280, height: 800 }]);
|
||||
});
|
||||
|
||||
it("rejects an unknown session", async () => {
|
||||
const sm = await makeManager();
|
||||
const { api } = fakeUpdateApi();
|
||||
const result = await handleWindowResize(
|
||||
sm,
|
||||
{ session_id: "zz99", width: 1280, height: 800 },
|
||||
api,
|
||||
);
|
||||
expect(result).toMatchObject({ code: "not_found" });
|
||||
});
|
||||
|
||||
it("rejects missing session_id", async () => {
|
||||
const sm = await makeManager();
|
||||
const { api } = fakeUpdateApi();
|
||||
const result = await handleWindowResize(sm, { session_id: "", width: 1280, height: 800 }, api);
|
||||
expect(result).toMatchObject({ code: "invalid_params" });
|
||||
});
|
||||
|
||||
it("rejects missing dimensions", async () => {
|
||||
const sm = await makeManager();
|
||||
const { api, calls } = fakeUpdateApi();
|
||||
const result = await handleWindowResize(
|
||||
sm,
|
||||
{ session_id: "aa11" } as unknown as { session_id: string; width: number; height: number },
|
||||
api,
|
||||
);
|
||||
expect(result).toMatchObject({ code: "invalid_params" });
|
||||
expect(calls).toEqual([]);
|
||||
});
|
||||
|
||||
it.each([
|
||||
[99, 800],
|
||||
[100, 7681],
|
||||
[1280.5, 800],
|
||||
[Number.NaN, 800],
|
||||
])("rejects out-of-range or non-integer dimensions (%s, %s)", async (width, height) => {
|
||||
const sm = await makeManager();
|
||||
const { api, calls } = fakeUpdateApi();
|
||||
const result = await handleWindowResize(sm, { session_id: "aa11", width, height }, api);
|
||||
expect(result).toMatchObject({ code: "invalid_params" });
|
||||
expect(calls).toEqual([]);
|
||||
});
|
||||
|
||||
it("maps chrome API failures to protocol_error", async () => {
|
||||
const sm = await makeManager();
|
||||
const { api } = fakeUpdateApi({ throw: true });
|
||||
const result = await handleWindowResize(
|
||||
sm,
|
||||
{ session_id: "aa11", width: 1280, height: 800 },
|
||||
api,
|
||||
);
|
||||
expect(result).toMatchObject({ code: "protocol_error", message: "simulated chrome failure" });
|
||||
});
|
||||
});
|
||||
|
||||
describe("handleSessionStart window size", () => {
|
||||
it("passes width/height through to Agent Window creation", async () => {
|
||||
const aw = fakeAgentWindow([100]);
|
||||
const sm = new SessionManager({ agentWindow: aw });
|
||||
const result = await handleSessionStart(sm, { session_id: "aa11", width: 1280, height: 800 });
|
||||
expect(result).toEqual({ agent_window_id: 100 });
|
||||
expect(aw.create).toHaveBeenCalledWith("about:blank", { width: 1280, height: 800 });
|
||||
});
|
||||
|
||||
it("creates the window without size when width/height are omitted", async () => {
|
||||
const aw = fakeAgentWindow([100]);
|
||||
const sm = new SessionManager({ agentWindow: aw });
|
||||
const result = await handleSessionStart(sm, { session_id: "aa11" });
|
||||
expect(result).toEqual({ agent_window_id: 100 });
|
||||
expect(aw.create).toHaveBeenCalledWith("about:blank");
|
||||
});
|
||||
|
||||
it("rejects a lone width without height", async () => {
|
||||
const sm = new SessionManager({ agentWindow: fakeAgentWindow([100]) });
|
||||
const result = await handleSessionStart(sm, { session_id: "aa11", width: 1280 });
|
||||
expect(result).toMatchObject({ code: "invalid_params" });
|
||||
expect(sm.has("aa11")).toBe(false);
|
||||
});
|
||||
|
||||
it.each([
|
||||
[99, 800],
|
||||
[1280, 7681],
|
||||
[1280.5, 800],
|
||||
])("rejects invalid dimensions (%s, %s)", async (width, height) => {
|
||||
const sm = new SessionManager({ agentWindow: fakeAgentWindow([100]) });
|
||||
const result = await handleSessionStart(sm, { session_id: "aa11", width, height });
|
||||
expect(result).toMatchObject({ code: "invalid_params" });
|
||||
expect(sm.has("aa11")).toBe(false);
|
||||
});
|
||||
});
|
||||
@@ -71,6 +71,7 @@ import {
|
||||
type TabSelectParams,
|
||||
} from "./tabs";
|
||||
import { handleWaitForNavigation } from "./waits";
|
||||
import { handleWindowResize, type WindowResizeParams } from "./window";
|
||||
|
||||
type DispatcherCdpRunner = CdpRunner &
|
||||
NetworkCdpRunner & {
|
||||
@@ -286,6 +287,8 @@ export class ToolDispatcher {
|
||||
});
|
||||
case "tool.tab_return":
|
||||
return handleTabReturn(this.sessions, req.params as TabReturnParams);
|
||||
case "tool.window_resize":
|
||||
return handleWindowResize(this.sessions, req.params as WindowResizeParams);
|
||||
case "tool.screenshot":
|
||||
return handleScreenshot(
|
||||
this.sessions,
|
||||
@@ -478,6 +481,7 @@ function sessionIdForBrowserControlMethod(req: RequestFrame): string | null {
|
||||
case "tool.tab_select":
|
||||
case "tool.tab_borrow":
|
||||
case "tool.tab_return":
|
||||
case "tool.window_resize":
|
||||
case "tool.navigate":
|
||||
case "tool.navigate_back":
|
||||
case "tool.navigate_forward":
|
||||
|
||||
@@ -1,11 +1,55 @@
|
||||
import type { SessionManager } from "@/session-manager/manager";
|
||||
import type { RpcError } from "@/transport/types";
|
||||
import { clearRecordingForSession } from "./record";
|
||||
import { isRpcError } from "./shared";
|
||||
import { returnBorrowedTab, type TabManagementDeps } from "./tabs";
|
||||
|
||||
/** Valid range for Agent Window dimensions in CSS pixels. */
|
||||
export const WINDOW_SIZE_MIN = 100;
|
||||
export const WINDOW_SIZE_MAX = 7680;
|
||||
|
||||
/**
|
||||
* Validate an optional width/height pair. Both must be given (or
|
||||
* neither), be integers, and fall within 100..=7680. Returns the size
|
||||
* to pass down, or an `invalid_params` RpcError.
|
||||
*/
|
||||
export function validateWindowSize(
|
||||
width: unknown,
|
||||
height: unknown,
|
||||
): { width: number; height: number } | undefined | RpcError {
|
||||
if (width === undefined && height === undefined) return undefined;
|
||||
if (width === undefined || height === undefined) {
|
||||
return {
|
||||
code: "invalid_params",
|
||||
message: "width and height must be given together",
|
||||
};
|
||||
}
|
||||
for (const [name, value] of [
|
||||
["width", width],
|
||||
["height", height],
|
||||
] as const) {
|
||||
if (
|
||||
typeof value !== "number" ||
|
||||
!Number.isInteger(value) ||
|
||||
value < WINDOW_SIZE_MIN ||
|
||||
value > WINDOW_SIZE_MAX
|
||||
) {
|
||||
return {
|
||||
code: "invalid_params",
|
||||
message: `${name} must be an integer in 100..=7680`,
|
||||
};
|
||||
}
|
||||
}
|
||||
return { width: width as number, height: height as number };
|
||||
}
|
||||
|
||||
export interface SessionStartParams {
|
||||
session_id: string;
|
||||
browser_instance_id?: string;
|
||||
/** Optional Agent Window outer width in CSS pixels (100..=7680). */
|
||||
width?: number;
|
||||
/** Optional Agent Window outer height in CSS pixels (100..=7680). */
|
||||
height?: number;
|
||||
}
|
||||
|
||||
export interface SessionStartResult {
|
||||
@@ -52,8 +96,10 @@ export async function handleSessionStart(
|
||||
message: "session.start requires session_id",
|
||||
};
|
||||
}
|
||||
const sizeOrErr = validateWindowSize(params.width, params.height);
|
||||
if (isRpcError(sizeOrErr)) return sizeOrErr;
|
||||
try {
|
||||
const ctx = await manager.start(params.session_id);
|
||||
const ctx = await manager.start(params.session_id, sizeOrErr);
|
||||
return { agent_window_id: ctx.agentWindowId };
|
||||
} catch (err) {
|
||||
// chrome.windows.create / SessionManager failures are not CDP
|
||||
|
||||
Executable
+82
@@ -0,0 +1,82 @@
|
||||
// Window-tool handlers: `tool.window_resize` resizes the session's
|
||||
// Agent Window via `chrome.windows.update`.
|
||||
|
||||
import type { SessionManager } from "@/session-manager/manager";
|
||||
import type { RpcError } from "@/transport/types";
|
||||
import { validateWindowSize } from "./session";
|
||||
import { isRpcError, lookupSession } from "./shared";
|
||||
|
||||
/**
|
||||
* Mirror of bsk-protocol `WindowResizeParams` /
|
||||
* `WindowResizeResult` (see crates/bsk-protocol/src/tools/window.rs).
|
||||
*/
|
||||
export interface WindowResizeParams {
|
||||
session_id: string;
|
||||
width: number;
|
||||
height: number;
|
||||
}
|
||||
|
||||
export interface WindowResizeResult {
|
||||
window_id: number;
|
||||
width: number;
|
||||
height: number;
|
||||
}
|
||||
|
||||
/**
|
||||
* Subset of `chrome.windows` we depend on, injectable so unit tests
|
||||
* can fake the browser without monkey-patching the global `chrome`.
|
||||
*/
|
||||
export interface WindowResizeApi {
|
||||
update(windowId: number, updateInfo: { width: number; height: number }): Promise<unknown>;
|
||||
}
|
||||
|
||||
export const chromeWindowResizeApi: WindowResizeApi = {
|
||||
async update(windowId, updateInfo) {
|
||||
return chrome.windows.update(windowId, updateInfo);
|
||||
},
|
||||
};
|
||||
|
||||
/**
|
||||
* Handler for `tool.window_resize` (called by the daemon over WS).
|
||||
*
|
||||
* Resizes the session's Agent Window to the given outer dimensions in
|
||||
* CSS pixels. chrome API failures are surfaced as `protocol_error`
|
||||
* (§4.5 reserves `cdp_failed` for raw CDP errors).
|
||||
*/
|
||||
export async function handleWindowResize(
|
||||
manager: SessionManager,
|
||||
params: WindowResizeParams,
|
||||
api: WindowResizeApi = chromeWindowResizeApi,
|
||||
): Promise<WindowResizeResult | RpcError> {
|
||||
const ctxOrErr = lookupSession(manager, params, "window_resize");
|
||||
if (isRpcError(ctxOrErr)) return ctxOrErr;
|
||||
const ctx = ctxOrErr;
|
||||
|
||||
const sizeOrErr = validateWindowSize(params.width, params.height);
|
||||
if (isRpcError(sizeOrErr)) return sizeOrErr;
|
||||
if (!sizeOrErr) {
|
||||
// validateWindowSize only returns undefined when both are absent,
|
||||
// which is valid for session_start but not for an explicit resize.
|
||||
return {
|
||||
code: "invalid_params",
|
||||
message: "window_resize requires width and height",
|
||||
};
|
||||
}
|
||||
|
||||
try {
|
||||
await api.update(ctx.agentWindowId, {
|
||||
width: sizeOrErr.width,
|
||||
height: sizeOrErr.height,
|
||||
});
|
||||
} catch (err) {
|
||||
return {
|
||||
code: "protocol_error",
|
||||
message: err instanceof Error ? err.message : String(err),
|
||||
};
|
||||
}
|
||||
return {
|
||||
window_id: ctx.agentWindowId,
|
||||
width: sizeOrErr.width,
|
||||
height: sizeOrErr.height,
|
||||
};
|
||||
}
|
||||
Reference in New Issue
Block a user