browser-pilot 0.0.7 → 0.0.9

package/dist/index.mjs

@@ -1,23 +1,22 @@
 import {
-  Tracer,
-  devices,
-  disableTracing,
-  enableTracing,
-  getTracer
-} from "./chunk-ZIQA4JOT.mjs";
-import {
+  AudioInput,
+  AudioOutput,
   Browser,
-  ElementNotFoundError,
-  NavigationError,
   Page,
   RequestInterceptor,
-  TimeoutError,
+  bufferToBase64,
+  calculateRMS,
   connect,
+  generateSilence,
+  generateTone,
+  grantAudioPermissions,
+  parseWavHeader,
+  pcmToWav,
   waitForAnyElement,
   waitForElement,
   waitForNavigation,
   waitForNetworkIdle
-} from "./chunk-PCNEJAJ7.mjs";
+} from "./chunk-7OSR2CAE.mjs";
 import {
   CDPError,
   createCDPClient
@@ -32,9 +31,346 @@ import {
 } from "./chunk-R3PS4PCM.mjs";
 import {
   BatchExecutor,
-  addBatchToPage
-} from "./chunk-6RB3GKQP.mjs";
+  ElementNotFoundError,
+  NavigationError,
+  TimeoutError,
+  addBatchToPage,
+  validateSteps
+} from "./chunk-KKW2SZLV.mjs";
+
+// src/audio/flags.ts
+function getAudioChromeFlags(options) {
+  const flags = [
+    "--use-fake-device-for-media-stream",
+    "--use-fake-ui-for-media-stream",
+    "--autoplay-policy=no-user-gesture-required"
+  ];
+  if (options?.inputWavPath) {
+    let path = options.inputWavPath;
+    if (options.noLoop) {
+      path += "%noloop";
+    }
+    flags.push(`--use-file-for-fake-audio-capture=${path}`);
+  }
+  return flags;
+}
+
+// src/audio/transcribe.ts
+async function transcribe(audio, options) {
+  const apiKey = options?.apiKey ?? getEnvVar("OPENAI_API_KEY");
+  if (!apiKey) {
+    throw new Error(
+      "OpenAI API key required for transcription. Set OPENAI_API_KEY environment variable or pass apiKey option."
+    );
+  }
+  if (audio.left.length === 0) {
+    return { text: "", audioDurationMs: 0, apiDurationMs: 0 };
+  }
+  const model = options?.model ?? "whisper-1";
+  const responseFormat = options?.responseFormat ?? "text";
+  const wavBuffer = pcmToWav({
+    left: audio.left,
+    right: audio.right.length > 0 ? audio.right : void 0,
+    sampleRate: audio.sampleRate
+  });
+  const boundary = `----bpAudio${Date.now()}`;
+  const parts = [];
+  appendFormField(parts, boundary, "file", new Uint8Array(wavBuffer), "audio.wav", "audio/wav");
+  appendFormTextField(parts, boundary, "model", model);
+  appendFormTextField(parts, boundary, "response_format", responseFormat);
+  if (options?.language) {
+    appendFormTextField(parts, boundary, "language", options.language);
+  }
+  if (options?.prompt) {
+    appendFormTextField(parts, boundary, "prompt", options.prompt);
+  }
+  const closing = new TextEncoder().encode(`\r
+--${boundary}--\r
+`);
+  parts.push(closing);
+  const totalLength = parts.reduce((sum, p) => sum + p.length, 0);
+  const body = new Uint8Array(totalLength);
+  let offset = 0;
+  for (const part of parts) {
+    body.set(part, offset);
+    offset += part.length;
+  }
+  const start = Date.now();
+  const response = await fetch("https://api.openai.com/v1/audio/transcriptions", {
+    method: "POST",
+    headers: {
+      Authorization: `Bearer ${apiKey}`,
+      "Content-Type": `multipart/form-data; boundary=${boundary}`
+    },
+    body
+  });
+  const apiDurationMs = Date.now() - start;
+  if (!response.ok) {
+    const errorBody = await response.text().catch(() => "");
+    throw new Error(`Whisper API error (${response.status}): ${errorBody}`);
+  }
+  let text;
+  if (responseFormat === "text") {
+    text = (await response.text()).trim();
+  } else {
+    const json = await response.json();
+    text = json.text ?? "";
+  }
+  return {
+    text,
+    audioDurationMs: audio.durationMs,
+    apiDurationMs
+  };
+}
+function isTranscriptionAvailable() {
+  return !!getEnvVar("OPENAI_API_KEY");
+}
+function getEnvVar(name) {
+  if (typeof globalThis.process !== "undefined" && globalThis.process.env) {
+    return globalThis.process.env[name];
+  }
+  return void 0;
+}
+function appendFormTextField(parts, boundary, name, value) {
+  const text = `\r
+--${boundary}\r
+Content-Disposition: form-data; name="${name}"\r
+\r
+${value}`;
+  parts.push(new TextEncoder().encode(text));
+}
+function appendFormField(parts, boundary, name, data, filename, contentType) {
+  const header = `\r
+--${boundary}\r
+Content-Disposition: form-data; name="${name}"; filename="${filename}"\r
+Content-Type: ${contentType}\r
+\r
+`;
+  parts.push(new TextEncoder().encode(header));
+  parts.push(data);
+}
+
+// src/emulation/devices.ts
+var devices = {
+  "iPhone 14": {
+    name: "iPhone 14",
+    viewport: {
+      width: 390,
+      height: 844,
+      deviceScaleFactor: 3,
+      isMobile: true,
+      hasTouch: true
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (iPhone; CPU iPhone OS 16_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 Mobile/15E148 Safari/604.1",
+      platform: "iPhone",
+      userAgentMetadata: {
+        mobile: true,
+        platform: "iOS",
+        platformVersion: "16.0"
+      }
+    }
+  },
+  "iPhone 14 Pro Max": {
+    name: "iPhone 14 Pro Max",
+    viewport: {
+      width: 430,
+      height: 932,
+      deviceScaleFactor: 3,
+      isMobile: true,
+      hasTouch: true
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (iPhone; CPU iPhone OS 16_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 Mobile/15E148 Safari/604.1",
+      platform: "iPhone"
+    }
+  },
+  "Pixel 7": {
+    name: "Pixel 7",
+    viewport: {
+      width: 412,
+      height: 915,
+      deviceScaleFactor: 2.625,
+      isMobile: true,
+      hasTouch: true
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (Linux; Android 13; Pixel 7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Mobile Safari/537.36",
+      platform: "Linux armv8l",
+      userAgentMetadata: {
+        mobile: true,
+        platform: "Android",
+        platformVersion: "13",
+        model: "Pixel 7"
+      }
+    }
+  },
+  "iPad Pro 11": {
+    name: "iPad Pro 11",
+    viewport: {
+      width: 834,
+      height: 1194,
+      deviceScaleFactor: 2,
+      isMobile: true,
+      hasTouch: true
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (iPad; CPU OS 16_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 Mobile/15E148 Safari/604.1",
+      platform: "iPad"
+    }
+  },
+  "Desktop Chrome": {
+    name: "Desktop Chrome",
+    viewport: {
+      width: 1920,
+      height: 1080,
+      deviceScaleFactor: 1,
+      isMobile: false,
+      hasTouch: false
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
+      platform: "Win32",
+      userAgentMetadata: {
+        brands: [
+          { brand: "Not_A Brand", version: "8" },
+          { brand: "Chromium", version: "120" },
+          { brand: "Google Chrome", version: "120" }
+        ],
+        platform: "Windows",
+        platformVersion: "10.0.0",
+        architecture: "x86",
+        bitness: "64",
+        mobile: false
+      }
+    }
+  },
+  "Desktop Firefox": {
+    name: "Desktop Firefox",
+    viewport: {
+      width: 1920,
+      height: 1080,
+      deviceScaleFactor: 1,
+      isMobile: false,
+      hasTouch: false
+    },
+    userAgent: {
+      userAgent: "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:121.0) Gecko/20100101 Firefox/121.0",
+      platform: "Win32"
+    }
+  }
+};
+
+// src/trace/tracer.ts
+var LEVEL_ORDER = {
+  debug: 0,
+  info: 1,
+  warn: 2,
+  error: 3
+};
+var Tracer = class _Tracer {
+  options;
+  constructor(options = {}) {
+    this.options = {
+      enabled: options.enabled ?? false,
+      output: options.output ?? "console",
+      callback: options.callback,
+      level: options.level ?? "info",
+      includeTimings: options.includeTimings ?? true
+    };
+  }
+  /**
+   * Emit a trace event
+   */
+  emit(event) {
+    if (!this.options.enabled) return;
+    if (LEVEL_ORDER[event.level] < LEVEL_ORDER[this.options.level]) {
+      return;
+    }
+    const fullEvent = {
+      ...event,
+      timestamp: (/* @__PURE__ */ new Date()).toISOString()
+    };
+    switch (this.options.output) {
+      case "console":
+        this.logToConsole(fullEvent);
+        break;
+      case "callback":
+        this.options.callback?.(fullEvent);
+        break;
+      case "silent":
+        break;
+    }
+  }
+  /**
+   * Log event to console
+   */
+  logToConsole(event) {
+    const { level, category, action, selectorUsed, success, durationMs, error } = event;
+    const icon = success === true ? "\u2713" : success === false ? "\u2717" : "\u25CB";
+    const timing = this.options.includeTimings && durationMs !== void 0 ? ` (${durationMs}ms)` : "";
+    const selector = selectorUsed ? ` ${selectorUsed}` : "";
+    const errorStr = error ? ` - ${error}` : "";
+    const message = `[${level.toUpperCase()}] [${category}] ${icon} ${action}${selector}${timing}${errorStr}`;
+    switch (level) {
+      case "debug":
+        console.debug(message);
+        break;
+      case "info":
+        console.info(message);
+        break;
+      case "warn":
+        console.warn(message);
+        break;
+      case "error":
+        console.error(message);
+        break;
+    }
+  }
+  /**
+   * Create a child tracer with modified options
+   */
+  child(options) {
+    return new _Tracer({ ...this.options, ...options });
+  }
+  /**
+   * Enable tracing
+   */
+  enable() {
+    this.options.enabled = true;
+  }
+  /**
+   * Disable tracing
+   */
+  disable() {
+    this.options.enabled = false;
+  }
+  /**
+   * Check if tracing is enabled
+   */
+  get isEnabled() {
+    return this.options.enabled;
+  }
+};
+var globalTracer = null;
+function getTracer() {
+  if (!globalTracer) {
+    globalTracer = new Tracer({ enabled: false });
+  }
+  return globalTracer;
+}
+function enableTracing(options = {}) {
+  globalTracer = new Tracer({ ...options, enabled: true });
+  return globalTracer;
+}
+function disableTracing() {
+  if (globalTracer) {
+    globalTracer.disable();
+  }
+}
 export {
+  AudioInput,
+  AudioOutput,
   BatchExecutor,
   Browser,
   BrowserBaseProvider,
@@ -48,6 +384,8 @@ export {
   TimeoutError,
   Tracer,
   addBatchToPage,
+  bufferToBase64,
+  calculateRMS,
   connect,
   createCDPClient,
   createProvider,
@@ -55,8 +393,17 @@ export {
   disableTracing,
   discoverTargets,
   enableTracing,
+  generateSilence,
+  generateTone,
+  getAudioChromeFlags,
   getBrowserWebSocketUrl,
   getTracer,
+  grantAudioPermissions,
+  isTranscriptionAvailable,
+  parseWavHeader,
+  pcmToWav,
+  transcribe,
+  validateSteps,
   waitForAnyElement,
   waitForElement,
   waitForNavigation,

package/dist/providers.d.cts

@@ -1,5 +1,5 @@
-import { P as Provider, a as CreateSessionOptions, b as ProviderSession, C as ConnectOptions } from './types-D_uDqh0Z.cjs';
-export { c as ProxyConfig } from './types-D_uDqh0Z.cjs';
+import { P as Provider, C as CreateSessionOptions, a as ProviderSession, b as ConnectOptions } from './types--wXNHUwt.cjs';
+export { c as ProxyConfig } from './types--wXNHUwt.cjs';
 
 /**
  * BrowserBase provider implementation

package/dist/providers.d.ts

@@ -1,5 +1,5 @@
-import { P as Provider, a as CreateSessionOptions, b as ProviderSession, C as ConnectOptions } from './types-D_uDqh0Z.js';
-export { c as ProxyConfig } from './types-D_uDqh0Z.js';
+import { P as Provider, C as CreateSessionOptions, a as ProviderSession, b as ConnectOptions } from './types--wXNHUwt.js';
+export { c as ProxyConfig } from './types--wXNHUwt.js';
 
 /**
  * BrowserBase provider implementation

package/dist/{types-D_uDqh0Z.d.cts → types--wXNHUwt.d.cts}

@@ -53,4 +53,4 @@ interface ConnectOptions {
     timeout?: number;
 }
 
-export type { ConnectOptions as C, Provider as P, CreateSessionOptions as a, ProviderSession as b, ProxyConfig as c };
+export type { CreateSessionOptions as C, Provider as P, ProviderSession as a, ConnectOptions as b, ProxyConfig as c };

package/dist/{types-D_uDqh0Z.d.ts → types--wXNHUwt.d.ts}

@@ -53,4 +53,4 @@ interface ConnectOptions {
     timeout?: number;
 }
 
-export type { ConnectOptions as C, Provider as P, CreateSessionOptions as a, ProviderSession as b, ProxyConfig as c };
+export type { CreateSessionOptions as C, Provider as P, ProviderSession as a, ConnectOptions as b, ProxyConfig as c };