@pedropaulovc/playwright-core 1.59.0-next

package/lib/server/agent/context.js

@@ -0,0 +1,150 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var context_exports = {};
+__export(context_exports, {
+  Context: () => Context
+});
+module.exports = __toCommonJS(context_exports);
+var import_browserContext = require("../browserContext");
+var import_actionRunner = require("./actionRunner");
+var import_codegen = require("./codegen");
+var import_stringUtils = require("../../utils/isomorphic/stringUtils");
+class Context {
+  constructor(page, agentParams, events) {
+    this._actions = [];
+    this._history = [];
+    this.page = page;
+    this.agentParams = agentParams;
+    this.sdkLanguage = page.browserContext._browser.sdkLanguage();
+    this.events = events;
+    this._budget = { tokens: agentParams.maxTokens };
+  }
+  async runActionAndWait(progress, action) {
+    return await this.runActionsAndWait(progress, [action]);
+  }
+  async runActionsAndWait(progress, action, options) {
+    const error = await this.waitForCompletion(progress, async () => {
+      for (const a of action) {
+        await (0, import_actionRunner.runAction)(progress, "generate", this.page, a, this.agentParams?.secrets ?? []);
+        const code = await (0, import_codegen.generateCode)(this.sdkLanguage, a);
+        this._actions.push({ ...a, code });
+      }
+      return void 0;
+    }, options).catch((error2) => error2);
+    return await this.snapshotResult(progress, error);
+  }
+  async runActionNoWait(progress, action) {
+    return await this.runActionsAndWait(progress, [action], { noWait: true });
+  }
+  actions() {
+    return this._actions.slice();
+  }
+  history() {
+    return this._history;
+  }
+  pushHistory(item) {
+    this._history.push(item);
+    this._actions = [];
+  }
+  consumeTokens(tokens) {
+    if (this._budget.tokens === void 0)
+      return;
+    this._budget.tokens = Math.max(0, this._budget.tokens - tokens);
+  }
+  maxTokensRemaining() {
+    return this._budget.tokens;
+  }
+  async waitForCompletion(progress, callback, options) {
+    if (options?.noWait)
+      return await callback();
+    const requests = [];
+    const requestListener = (request) => requests.push(request);
+    const disposeListeners = () => {
+      this.page.browserContext.off(import_browserContext.BrowserContext.Events.Request, requestListener);
+    };
+    this.page.browserContext.on(import_browserContext.BrowserContext.Events.Request, requestListener);
+    let result;
+    try {
+      result = await callback();
+      await progress.wait(500);
+    } finally {
+      disposeListeners();
+    }
+    const requestedNavigation = requests.some((request) => request.isNavigationRequest());
+    if (requestedNavigation) {
+      await this.page.mainFrame().waitForLoadState(progress, "load");
+      return result;
+    }
+    const promises = [];
+    for (const request of requests) {
+      if (["document", "stylesheet", "script", "xhr", "fetch"].includes(request.resourceType()))
+        promises.push(request.response().then((r) => r?.finished()));
+      else
+        promises.push(request.response());
+    }
+    await progress.race([...promises, progress.wait(5e3)]);
+    if (!promises.length)
+      await progress.wait(500);
+    return result;
+  }
+  async takeSnapshot(progress) {
+    const { full } = await this.page.snapshotForAI(progress, { doNotRenderActive: this.agentParams.doNotRenderActive });
+    return full;
+  }
+  async snapshotResult(progress, error) {
+    const snapshot = this._redactText(await this.takeSnapshot(progress));
+    const text = [];
+    if (error)
+      text.push(`# Error
+${(0, import_stringUtils.stripAnsiEscapes)(error.message)}`);
+    else
+      text.push(`# Success`);
+    text.push(`# Page snapshot
+${snapshot}`);
+    return {
+      isError: !!error,
+      content: [{ type: "text", text: text.join("\n\n") }]
+    };
+  }
+  async refSelectors(progress, params) {
+    return Promise.all(params.map(async (param) => {
+      try {
+        const { resolvedSelector } = await this.page.mainFrame().resolveSelector(progress, `aria-ref=${param.ref}`);
+        return resolvedSelector;
+      } catch (e) {
+        throw new Error(`Ref ${param.ref} not found in the current page snapshot. Try capturing new snapshot.`);
+      }
+    }));
+  }
+  _redactText(text) {
+    const secrets = this.agentParams?.secrets;
+    if (!secrets)
+      return text;
+    const redactText = (text2) => {
+      for (const { name, value } of secrets)
+        text2 = text2.replaceAll(value, `<secret>${name}</secret>`);
+      return text2;
+    };
+    return redactText(text);
+  }
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  Context
+});

package/lib/server/agent/expectTools.js

@@ -0,0 +1,156 @@
+"use strict";
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var expectTools_exports = {};
+__export(expectTools_exports, {
+  default: () => expectTools_default
+});
+module.exports = __toCommonJS(expectTools_exports);
+var import_mcpBundle = require("../../mcpBundle");
+var import_locatorUtils = require("../../utils/isomorphic/locatorUtils");
+var import_yaml = require("../../utils/isomorphic/yaml");
+var import_tool = require("./tool");
+const expectVisible = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_visible",
+    title: "Expect element visible",
+    description: "Expect element is visible on the page",
+    inputSchema: import_mcpBundle.z.object({
+      role: import_mcpBundle.z.string().describe('ROLE of the element. Can be found in the snapshot like this: `- {ROLE} "Accessible Name":`'),
+      accessibleName: import_mcpBundle.z.string().describe('ACCESSIBLE_NAME of the element. Can be found in the snapshot like this: `- role "{ACCESSIBLE_NAME}"`'),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    return await context.runActionAndWait(progress, {
+      method: "expectVisible",
+      selector: (0, import_locatorUtils.getByRoleSelector)(params.role, { name: params.accessibleName }),
+      isNot: params.isNot
+    });
+  }
+});
+const expectVisibleText = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_visible_text",
+    title: "Expect text visible",
+    description: `Expect text is visible on the page. Prefer ${expectVisible.schema.name} if possible.`,
+    inputSchema: import_mcpBundle.z.object({
+      text: import_mcpBundle.z.string().describe('TEXT to expect. Can be found in the snapshot like this: `- role "Accessible Name": {TEXT}` or like this: `- text: {TEXT}`'),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    return await context.runActionAndWait(progress, {
+      method: "expectVisible",
+      selector: (0, import_locatorUtils.getByTextSelector)(params.text),
+      isNot: params.isNot
+    });
+  }
+});
+const expectValue = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_value",
+    title: "Expect value",
+    description: "Expect element value",
+    inputSchema: import_mcpBundle.z.object({
+      type: import_mcpBundle.z.enum(["textbox", "checkbox", "radio", "combobox", "slider"]).describe("Type of the element"),
+      element: import_mcpBundle.z.string().describe("Human-readable element description"),
+      ref: import_mcpBundle.z.string().describe("Exact target element reference from the page snapshot"),
+      value: import_mcpBundle.z.string().describe('Value to expect. For checkbox, use "true" or "false".'),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    const [selector] = await context.refSelectors(progress, [{ ref: params.ref, element: params.element }]);
+    return await context.runActionAndWait(progress, {
+      method: "expectValue",
+      selector,
+      type: params.type,
+      value: params.value,
+      isNot: params.isNot
+    });
+  }
+});
+const expectList = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_list_visible",
+    title: "Expect list visible",
+    description: "Expect list is visible on the page, ensures items are present in the element in the exact order",
+    inputSchema: import_mcpBundle.z.object({
+      listRole: import_mcpBundle.z.string().describe("Aria role of the list element as in the snapshot"),
+      listAccessibleName: import_mcpBundle.z.string().optional().describe("Accessible name of the list element as in the snapshot"),
+      itemRole: import_mcpBundle.z.string().describe("Aria role of the list items as in the snapshot, should all be the same"),
+      items: import_mcpBundle.z.array(import_mcpBundle.z.string().describe("Text to look for in the list item, can be either from accessible name of self / nested text content")),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    const template = `- ${params.listRole}:
+${params.items.map((item) => `  - ${params.itemRole}: ${(0, import_yaml.yamlEscapeValueIfNeeded)(item)}`).join("\n")}`;
+    return await context.runActionAndWait(progress, {
+      method: "expectAria",
+      template
+    });
+  }
+});
+const expectURL = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_url",
+    title: "Expect URL",
+    description: "Expect the page URL to match the expected value. Either provide a url string or a regex pattern.",
+    inputSchema: import_mcpBundle.z.object({
+      url: import_mcpBundle.z.string().optional().describe("Expected URL string. Relative URLs are resolved against the baseURL."),
+      regex: import_mcpBundle.z.string().optional().describe("Regular expression pattern to match the URL against, e.g. /foo.*/i."),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    return await context.runActionAndWait(progress, {
+      method: "expectURL",
+      value: params.url,
+      regex: params.regex,
+      isNot: params.isNot
+    });
+  }
+});
+const expectTitle = (0, import_tool.defineTool)({
+  schema: {
+    name: "browser_expect_title",
+    title: "Expect title",
+    description: "Expect the page title to match the expected value.",
+    inputSchema: import_mcpBundle.z.object({
+      title: import_mcpBundle.z.string().describe("Expected page title."),
+      isNot: import_mcpBundle.z.boolean().optional().describe("Expect the opposite")
+    })
+  },
+  handle: async (progress, context, params) => {
+    return await context.runActionAndWait(progress, {
+      method: "expectTitle",
+      value: params.title,
+      isNot: params.isNot
+    });
+  }
+});
+var expectTools_default = [
+  expectVisible,
+  expectVisibleText,
+  expectValue,
+  expectList,
+  expectURL,
+  expectTitle
+];

package/lib/server/agent/pageAgent.js

@@ -0,0 +1,204 @@
+"use strict";
+var __create = Object.create;
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __getProtoOf = Object.getPrototypeOf;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
+  // If the importer is in node compatibility mode or this is not an ESM
+  // file that has been converted to a CommonJS file using a Babel-
+  // compatible transform (i.e. "__esModule" has not been set), then set
+  // "default" to the CommonJS "module.exports" for node compatibility.
+  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
+  mod
+));
+var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
+var pageAgent_exports = {};
+__export(pageAgent_exports, {
+  pageAgentExpect: () => pageAgentExpect,
+  pageAgentExtract: () => pageAgentExtract,
+  pageAgentPerform: () => pageAgentPerform
+});
+module.exports = __toCommonJS(pageAgent_exports);
+var import_fs = __toESM(require("fs"));
+var import_path = __toESM(require("path"));
+var import_tool = require("./tool");
+var import_utilsBundle = require("../../utilsBundle");
+var import_mcpBundle = require("../../mcpBundle");
+var import_actionRunner = require("./actionRunner");
+var import_performTools = __toESM(require("./performTools"));
+var import_expectTools = __toESM(require("./expectTools"));
+var actions = __toESM(require("./actions"));
+async function pageAgentPerform(progress, context, userTask, callParams) {
+  const cacheKey = (callParams.cacheKey ?? userTask).trim();
+  if (await cachedPerform(progress, context, cacheKey))
+    return;
+  const task = `
+### Instructions
+- Perform the following task on the page.
+- Your reply should be a tool call that performs action the page".
+
+### Task
+${userTask}
+`;
+  progress.disableTimeout();
+  await runLoop(progress, context, import_performTools.default, task, void 0, callParams);
+  await updateCache(context, cacheKey);
+}
+async function pageAgentExpect(progress, context, expectation, callParams) {
+  const cacheKey = (callParams.cacheKey ?? expectation).trim();
+  if (await cachedPerform(progress, context, cacheKey))
+    return;
+  const task = `
+### Instructions
+- Call one of the "browser_expect_*" tools to verify / assert the condition.
+- You can call exactly one tool and it can't be report_results, must be one of the assertion tools.
+
+### Expectation
+${expectation}
+`;
+  progress.disableTimeout();
+  await runLoop(progress, context, import_expectTools.default, task, void 0, callParams);
+  await updateCache(context, cacheKey);
+}
+async function pageAgentExtract(progress, context, query, schema, callParams) {
+  const task = `
+### Instructions
+Extract the following information from the page. Do not perform any actions, just extract the information.
+
+### Query
+${query}`;
+  const { result } = await runLoop(progress, context, [], task, schema, callParams);
+  return result;
+}
+async function runLoop(progress, context, toolDefinitions, userTask, resultSchema, params) {
+  if (!context.agentParams.api || !context.agentParams.model)
+    throw new Error(`This action requires the API and API key to be set on the page agent. Did you mean to --run-agents=missing?`);
+  if (!context.agentParams.apiKey)
+    throw new Error(`This action requires API key to be set on the page agent.`);
+  if (context.agentParams.apiEndpoint && !URL.canParse(context.agentParams.apiEndpoint))
+    throw new Error(`Agent API endpoint "${context.agentParams.apiEndpoint}" is not a valid URL.`);
+  const snapshot = await context.takeSnapshot(progress);
+  const { tools, callTool, reportedResult, refusedToPerformReason } = (0, import_tool.toolsForLoop)(progress, context, toolDefinitions, { resultSchema, refuseToPerform: "allow" });
+  const secrets = Object.fromEntries((context.agentParams.secrets || [])?.map((s) => [s.name, s.value]));
+  const apiCacheTextBefore = context.agentParams.apiCacheFile ? await import_fs.default.promises.readFile(context.agentParams.apiCacheFile, "utf-8").catch(() => "{}") : "{}";
+  const apiCacheBefore = JSON.parse(apiCacheTextBefore || "{}");
+  const loop = new import_mcpBundle.Loop({
+    api: context.agentParams.api,
+    apiEndpoint: context.agentParams.apiEndpoint,
+    apiKey: context.agentParams.apiKey,
+    apiTimeout: context.agentParams.apiTimeout ?? 0,
+    model: context.agentParams.model,
+    maxTokens: params.maxTokens ?? context.maxTokensRemaining(),
+    maxToolCalls: params.maxActions ?? context.agentParams.maxActions ?? 10,
+    maxToolCallRetries: params.maxActionRetries ?? context.agentParams.maxActionRetries ?? 3,
+    summarize: true,
+    debug: import_utilsBundle.debug,
+    callTool,
+    tools,
+    secrets,
+    cache: apiCacheBefore,
+    ...context.events
+  });
+  const task = [];
+  if (context.agentParams.systemPrompt) {
+    task.push("### System");
+    task.push(context.agentParams.systemPrompt);
+    task.push("");
+  }
+  task.push("### Task");
+  task.push(userTask);
+  if (context.history().length) {
+    task.push("### Context history");
+    task.push(context.history().map((h) => `- ${h.type}: ${h.description}`).join("\n"));
+    task.push("");
+  }
+  task.push("### Page snapshot");
+  task.push(snapshot);
+  task.push("");
+  const { error, usage } = await loop.run(task.join("\n"), { signal: progress.signal });
+  context.consumeTokens(usage.input + usage.output);
+  if (context.agentParams.apiCacheFile) {
+    const apiCacheAfter = { ...apiCacheBefore, ...loop.cache() };
+    const sortedCache = Object.fromEntries(Object.entries(apiCacheAfter).sort(([a], [b]) => a.localeCompare(b)));
+    const apiCacheTextAfter = JSON.stringify(sortedCache, void 0, 2);
+    if (apiCacheTextAfter !== apiCacheTextBefore) {
+      await import_fs.default.promises.mkdir(import_path.default.dirname(context.agentParams.apiCacheFile), { recursive: true });
+      await import_fs.default.promises.writeFile(context.agentParams.apiCacheFile, apiCacheTextAfter);
+    }
+  }
+  if (refusedToPerformReason())
+    throw new Error(`Agent refused to perform action: ${refusedToPerformReason()}`);
+  if (error)
+    throw new Error(`Agentic loop failed: ${error}`);
+  return { result: reportedResult ? reportedResult() : void 0 };
+}
+async function cachedPerform(progress, context, cacheKey) {
+  if (!context.agentParams?.cacheFile)
+    return;
+  const cache = await cachedActions(context.agentParams?.cacheFile);
+  const entry = cache.actions[cacheKey];
+  if (!entry)
+    return;
+  for (const action of entry.actions)
+    await (0, import_actionRunner.runAction)(progress, "run", context.page, action, context.agentParams.secrets ?? []);
+  return entry.actions;
+}
+async function updateCache(context, cacheKey) {
+  const cacheFile = context.agentParams?.cacheFile;
+  const cacheOutFile = context.agentParams?.cacheOutFile;
+  const cacheFileKey = cacheFile ?? cacheOutFile;
+  const cache = cacheFileKey ? await cachedActions(cacheFileKey) : { actions: {}, newActions: {} };
+  const newEntry = { actions: context.actions() };
+  cache.actions[cacheKey] = newEntry;
+  cache.newActions[cacheKey] = newEntry;
+  if (cacheOutFile) {
+    const entries = Object.entries(cache.newActions);
+    entries.sort((e1, e2) => e1[0].localeCompare(e2[0]));
+    await import_fs.default.promises.writeFile(cacheOutFile, JSON.stringify(Object.fromEntries(entries), void 0, 2));
+  } else if (cacheFile) {
+    const entries = Object.entries(cache.actions);
+    entries.sort((e1, e2) => e1[0].localeCompare(e2[0]));
+    await import_fs.default.promises.writeFile(cacheFile, JSON.stringify(Object.fromEntries(entries), void 0, 2));
+  }
+}
+const allCaches = /* @__PURE__ */ new Map();
+async function cachedActions(cacheFile) {
+  let cache = allCaches.get(cacheFile);
+  if (!cache) {
+    const content = await import_fs.default.promises.readFile(cacheFile, "utf-8").catch(() => "");
+    let json;
+    try {
+      json = JSON.parse(content.trim() || "{}");
+    } catch (error) {
+      throw new Error(`Failed to parse cache file ${cacheFile}:
+${error.message}`);
+    }
+    const parsed = actions.cachedActionsSchema.safeParse(json);
+    if (parsed.error)
+      throw new Error(`Failed to parse cache file ${cacheFile}:
+${import_mcpBundle.z.prettifyError(parsed.error)}`);
+    cache = { actions: parsed.data, newActions: {} };
+    allCaches.set(cacheFile, cache);
+  }
+  return cache;
+}
+// Annotate the CommonJS export names for ESM import in node:
+0 && (module.exports = {
+  pageAgentExpect,
+  pageAgentExtract,
+  pageAgentPerform
+});