mercury-agent 0.4.7 → 0.4.9

package/src/storage/pi-auth.ts

@@ -1,95 +1,95 @@
-import fs from "node:fs";
-import path from "node:path";
-import {
-  getOAuthApiKey,
-  type OAuthCredentials,
-  type OAuthProviderId,
-} from "@earendil-works/pi-ai/oauth";
-import { logger } from "../logger.js";
-
-type AuthEntry =
-  | ({ type: "oauth" } & OAuthCredentials)
-  | { type: "api_key"; key: string }
-  | Record<string, unknown>;
-
-type AuthFile = Record<string, AuthEntry>;
-
-function readAuthFile(authPath: string): AuthFile {
-  if (!fs.existsSync(authPath)) return {};
-  try {
-    return JSON.parse(fs.readFileSync(authPath, "utf8")) as AuthFile;
-  } catch (err) {
-    logger.warn(
-      `pi-auth: auth file at ${authPath} is malformed, ignoring`,
-      err instanceof Error ? err : undefined,
-    );
-    return {};
-  }
-}
-
-function writeAuthFile(authPath: string, auth: AuthFile): void {
-  fs.mkdirSync(path.dirname(authPath), { recursive: true });
-  fs.writeFileSync(authPath, JSON.stringify(auth, null, 2), "utf8");
-  fs.chmodSync(authPath, 0o600);
-}
-
-export async function getApiKeyFromPiAuthFile(options: {
-  provider: string;
-  authPath: string;
-}): Promise<string | undefined> {
-  if (
-    process.env.MERCURY_ANTHROPIC_API_KEY ||
-    process.env.MERCURY_ANTHROPIC_OAUTH_TOKEN
-  ) {
-    return undefined;
-  }
-
-  if (options.provider !== "anthropic") {
-    return undefined;
-  }
-
-  const authPath = options.authPath;
-  const auth = readAuthFile(authPath);
-
-  const entry = auth.anthropic;
-  if (!entry || typeof entry !== "object" || entry.type !== "oauth") {
-    return undefined;
-  }
-
-  const access = typeof entry.access === "string" ? entry.access : undefined;
-  const refresh = typeof entry.refresh === "string" ? entry.refresh : undefined;
-  const expires = typeof entry.expires === "number" ? entry.expires : undefined;
-  if (!access || !refresh || typeof expires !== "number") return undefined;
-
-  try {
-    const result = await getOAuthApiKey("anthropic" satisfies OAuthProviderId, {
-      anthropic: {
-        access,
-        refresh,
-        expires,
-      },
-    });
-
-    if (!result) return undefined;
-
-    const nextAuth = {
-      ...auth,
-      anthropic: {
-        type: "oauth" as const,
-        ...result.newCredentials,
-      },
-    };
-
-    writeAuthFile(authPath, nextAuth);
-    logger.debug("Loaded anthropic oauth token from pi auth.json", {
-      authPath,
-    });
-    return result.apiKey;
-  } catch (error) {
-    logger.warn(
-      "Failed to load anthropic oauth token from pi auth.json",
-      error instanceof Error ? error : undefined,
-    );
-    return undefined;
-  }
-}
+import fs from "node:fs";
+import path from "node:path";
+import {
+  getOAuthApiKey,
+  type OAuthCredentials,
+  type OAuthProviderId,
+} from "@earendil-works/pi-ai/oauth";
+import { logger } from "../logger.js";
+
+type AuthEntry =
+  | ({ type: "oauth" } & OAuthCredentials)
+  | { type: "api_key"; key: string }
+  | Record<string, unknown>;
+
+type AuthFile = Record<string, AuthEntry>;
+
+function readAuthFile(authPath: string): AuthFile {
+  if (!fs.existsSync(authPath)) return {};
+  try {
+    return JSON.parse(fs.readFileSync(authPath, "utf8")) as AuthFile;
+  } catch (err) {
+    logger.warn(
+      `pi-auth: auth file at ${authPath} is malformed, ignoring`,
+      err instanceof Error ? err : undefined,
+    );
+    return {};
+  }
+}
+
+function writeAuthFile(authPath: string, auth: AuthFile): void {
+  fs.mkdirSync(path.dirname(authPath), { recursive: true });
+  fs.writeFileSync(authPath, JSON.stringify(auth, null, 2), "utf8");
+  fs.chmodSync(authPath, 0o600);
+}
+
+export async function getApiKeyFromPiAuthFile(options: {
+  provider: string;
+  authPath: string;
+}): Promise<string | undefined> {
+  if (
+    process.env.MERCURY_ANTHROPIC_API_KEY ||
+    process.env.MERCURY_ANTHROPIC_OAUTH_TOKEN
+  ) {
+    return undefined;
+  }
+
+  if (options.provider !== "anthropic") {
+    return undefined;
+  }
+
+  const authPath = options.authPath;
+  const auth = readAuthFile(authPath);
+
+  const entry = auth.anthropic;
+  if (!entry || typeof entry !== "object" || entry.type !== "oauth") {
+    return undefined;
+  }
+
+  const access = typeof entry.access === "string" ? entry.access : undefined;
+  const refresh = typeof entry.refresh === "string" ? entry.refresh : undefined;
+  const expires = typeof entry.expires === "number" ? entry.expires : undefined;
+  if (!access || !refresh || typeof expires !== "number") return undefined;
+
+  try {
+    const result = await getOAuthApiKey("anthropic" satisfies OAuthProviderId, {
+      anthropic: {
+        access,
+        refresh,
+        expires,
+      },
+    });
+
+    if (!result) return undefined;
+
+    const nextAuth = {
+      ...auth,
+      anthropic: {
+        type: "oauth" as const,
+        ...result.newCredentials,
+      },
+    };
+
+    writeAuthFile(authPath, nextAuth);
+    logger.debug("Loaded anthropic oauth token from pi auth.json", {
+      authPath,
+    });
+    return result.apiKey;
+  } catch (error) {
+    logger.warn(
+      "Failed to load anthropic oauth token from pi auth.json",
+      error instanceof Error ? error : undefined,
+    );
+    return undefined;
+  }
+}

package/src/tts/azure.ts

@@ -1,52 +1,52 @@
-import type { TtsLanguage } from "./language.js";
-
-const AZURE_VOICES: Record<TtsLanguage, string> = {
-  "he-IL": "he-IL-HilaNeural",
-  "en-US": "en-US-JennyNeural",
-};
-
-/** MP3 128 kbps mono 16 kHz — widely compatible for chat attachments. */
-const OUTPUT_FORMAT = "audio-16khz-128kbitrate-mono-mp3";
-
-function escapeXml(text: string): string {
-  return text
-    .replace(/&/g, "&amp;")
-    .replace(/</g, "&lt;")
-    .replace(/>/g, "&gt;")
-    .replace(/"/g, "&quot;")
-    .replace(/'/g, "&apos;");
-}
-
-export async function synthesizeAzure(opts: {
-  key: string;
-  region: string;
-  text: string;
-  language: TtsLanguage;
-}): Promise<Buffer> {
-  const voiceName = AZURE_VOICES[opts.language];
-  const ssml = `<?xml version="1.0" encoding="UTF-8"?>
-<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="${opts.language}">
-  <voice name="${voiceName}">${escapeXml(opts.text)}</voice>
-</speak>`;
-
-  const url = `https://${opts.region.trim()}.tts.speech.microsoft.com/cognitiveservices/v1`;
-  const res = await fetch(url, {
-    method: "POST",
-    headers: {
-      "Content-Type": "application/ssml+xml",
-      "X-Microsoft-OutputFormat": OUTPUT_FORMAT,
-      "Ocp-Apim-Subscription-Key": opts.key.trim(),
-      "User-Agent": "mercury-agent-tts",
-    },
-    body: ssml,
-  });
-
-  if (!res.ok) {
-    const errText = await res.text().catch(() => "");
-    throw new Error(
-      `Azure TTS HTTP ${res.status}: ${errText.slice(0, 500) || res.statusText}`,
-    );
-  }
-
-  return Buffer.from(await res.arrayBuffer());
-}
+import type { TtsLanguage } from "./language.js";
+
+const AZURE_VOICES: Record<TtsLanguage, string> = {
+  "he-IL": "he-IL-HilaNeural",
+  "en-US": "en-US-JennyNeural",
+};
+
+/** MP3 128 kbps mono 16 kHz — widely compatible for chat attachments. */
+const OUTPUT_FORMAT = "audio-16khz-128kbitrate-mono-mp3";
+
+function escapeXml(text: string): string {
+  return text
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;")
+    .replace(/"/g, "&quot;")
+    .replace(/'/g, "&apos;");
+}
+
+export async function synthesizeAzure(opts: {
+  key: string;
+  region: string;
+  text: string;
+  language: TtsLanguage;
+}): Promise<Buffer> {
+  const voiceName = AZURE_VOICES[opts.language];
+  const ssml = `<?xml version="1.0" encoding="UTF-8"?>
+<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="${opts.language}">
+  <voice name="${voiceName}">${escapeXml(opts.text)}</voice>
+</speak>`;
+
+  const url = `https://${opts.region.trim()}.tts.speech.microsoft.com/cognitiveservices/v1`;
+  const res = await fetch(url, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/ssml+xml",
+      "X-Microsoft-OutputFormat": OUTPUT_FORMAT,
+      "Ocp-Apim-Subscription-Key": opts.key.trim(),
+      "User-Agent": "mercury-agent-tts",
+    },
+    body: ssml,
+  });
+
+  if (!res.ok) {
+    const errText = await res.text().catch(() => "");
+    throw new Error(
+      `Azure TTS HTTP ${res.status}: ${errText.slice(0, 500) || res.statusText}`,
+    );
+  }
+
+  return Buffer.from(await res.arrayBuffer());
+}

package/src/tts/synthesize.ts

@@ -1,133 +1,133 @@
-import { existsSync } from "node:fs";
-import { resolveProjectPath } from "../config.js";
-import { synthesizeAzure } from "./azure.js";
-import { synthesizeGoogle } from "./google.js";
-import {
-  resolveTtsLanguageFromText,
-  type TtsLanguageInput,
-} from "./language.js";
-
-export class TtsConfigError extends Error {
-  readonly code: "not_configured";
-
-  constructor(message: string) {
-    super(message);
-    this.name = "TtsConfigError";
-    this.code = "not_configured";
-  }
-}
-
-/** Subset of AppConfig used by TTS (extensions import via `mercury-agent/tts`). */
-export interface MercuryTtsConfig {
-  ttsProvider: "google" | "azure" | "auto";
-  azureSpeechKey?: string;
-  azureSpeechRegion?: string;
-  googleApplicationCredentials?: string;
-  ttsMaxChars: number;
-}
-
-export interface TtsSynthesizeOptions {
-  text: string;
-  language?: TtsLanguageInput;
-  providerOverride?: "google" | "azure" | "auto";
-}
-
-export interface TtsSynthesizeResult {
-  buffer: Buffer;
-  mimeType: string;
-  filename: string;
-}
-
-function hasGoogleCredentials(config: MercuryTtsConfig): boolean {
-  const p = config.googleApplicationCredentials?.trim();
-  if (!p) return false;
-  try {
-    return existsSync(resolveProjectPath(p));
-  } catch {
-    return false;
-  }
-}
-
-function hasAzure(config: MercuryTtsConfig): boolean {
-  return Boolean(
-    config.azureSpeechKey?.trim() && config.azureSpeechRegion?.trim(),
-  );
-}
-
-function resolveEffectiveBackend(
-  config: MercuryTtsConfig,
-  override?: "google" | "azure" | "auto",
-): "google" | "azure" {
-  const mode = override ?? config.ttsProvider;
-
-  if (mode === "google") {
-    if (!hasGoogleCredentials(config)) {
-      throw new TtsConfigError(
-        "Google TTS selected but MERCURY_GOOGLE_APPLICATION_CREDENTIALS / GOOGLE_APPLICATION_CREDENTIALS file is missing or not found",
-      );
-    }
-    return "google";
-  }
-  if (mode === "azure") {
-    if (!hasAzure(config)) {
-      throw new TtsConfigError(
-        "Azure TTS selected but MERCURY_AZURE_SPEECH_KEY or MERCURY_AZURE_SPEECH_REGION is missing",
-      );
-    }
-    return "azure";
-  }
-
-  if (hasGoogleCredentials(config)) return "google";
-  if (hasAzure(config)) return "azure";
-
-  throw new TtsConfigError(
-    "No TTS provider configured: set Google service account path or Azure key+region",
-  );
-}
-
-/**
- * Synthesize speech using host credentials. Intended for `/api/tts` and voice-synth extension.
- */
-export async function synthesizeSpeech(
-  config: MercuryTtsConfig,
-  opts: TtsSynthesizeOptions,
-): Promise<TtsSynthesizeResult> {
-  const trimmed = opts.text.trim();
-  if (!trimmed) {
-    throw new Error("TTS text is empty");
-  }
-
-  const max = Math.min(10_000, Math.max(500, config.ttsMaxChars));
-  if (trimmed.length > max) {
-    throw new Error(`TTS text exceeds max length (${max} characters)`);
-  }
-
-  const language = resolveTtsLanguageFromText(trimmed, opts.language);
-  const backend = resolveEffectiveBackend(config, opts.providerOverride);
-
-  let buffer: Buffer;
-  if (backend === "azure") {
-    buffer = await synthesizeAzure({
-      key: config.azureSpeechKey as string,
-      region: config.azureSpeechRegion as string,
-      text: trimmed,
-      language,
-    });
-  } else {
-    const credPath = resolveProjectPath(
-      config.googleApplicationCredentials?.trim() ?? "",
-    );
-    buffer = await synthesizeGoogle({
-      credentialsPath: credPath,
-      text: trimmed,
-      language,
-    });
-  }
-
-  return {
-    buffer,
-    mimeType: "audio/mpeg",
-    filename:
-      language === "he-IL" ? "mercury-tts-he.mp3" : "mercury-tts-en.mp3",
-  };
-}
+import { existsSync } from "node:fs";
+import { resolveProjectPath } from "../config.js";
+import { synthesizeAzure } from "./azure.js";
+import { synthesizeGoogle } from "./google.js";
+import {
+  resolveTtsLanguageFromText,
+  type TtsLanguageInput,
+} from "./language.js";
+
+export class TtsConfigError extends Error {
+  readonly code: "not_configured";
+
+  constructor(message: string) {
+    super(message);
+    this.name = "TtsConfigError";
+    this.code = "not_configured";
+  }
+}
+
+/** Subset of AppConfig used by TTS (extensions import via `mercury-agent/tts`). */
+export interface MercuryTtsConfig {
+  ttsProvider: "google" | "azure" | "auto";
+  azureSpeechKey?: string;
+  azureSpeechRegion?: string;
+  googleApplicationCredentials?: string;
+  ttsMaxChars: number;
+}
+
+export interface TtsSynthesizeOptions {
+  text: string;
+  language?: TtsLanguageInput;
+  providerOverride?: "google" | "azure" | "auto";
+}
+
+export interface TtsSynthesizeResult {
+  buffer: Buffer;
+  mimeType: string;
+  filename: string;
+}
+
+function hasGoogleCredentials(config: MercuryTtsConfig): boolean {
+  const p = config.googleApplicationCredentials?.trim();
+  if (!p) return false;
+  try {
+    return existsSync(resolveProjectPath(p));
+  } catch {
+    return false;
+  }
+}
+
+function hasAzure(config: MercuryTtsConfig): boolean {
+  return Boolean(
+    config.azureSpeechKey?.trim() && config.azureSpeechRegion?.trim(),
+  );
+}
+
+function resolveEffectiveBackend(
+  config: MercuryTtsConfig,
+  override?: "google" | "azure" | "auto",
+): "google" | "azure" {
+  const mode = override ?? config.ttsProvider;
+
+  if (mode === "google") {
+    if (!hasGoogleCredentials(config)) {
+      throw new TtsConfigError(
+        "Google TTS selected but MERCURY_GOOGLE_APPLICATION_CREDENTIALS / GOOGLE_APPLICATION_CREDENTIALS file is missing or not found",
+      );
+    }
+    return "google";
+  }
+  if (mode === "azure") {
+    if (!hasAzure(config)) {
+      throw new TtsConfigError(
+        "Azure TTS selected but MERCURY_AZURE_SPEECH_KEY or MERCURY_AZURE_SPEECH_REGION is missing",
+      );
+    }
+    return "azure";
+  }
+
+  if (hasGoogleCredentials(config)) return "google";
+  if (hasAzure(config)) return "azure";
+
+  throw new TtsConfigError(
+    "No TTS provider configured: set Google service account path or Azure key+region",
+  );
+}
+
+/**
+ * Synthesize speech using host credentials. Intended for `/api/tts` and voice-synth extension.
+ */
+export async function synthesizeSpeech(
+  config: MercuryTtsConfig,
+  opts: TtsSynthesizeOptions,
+): Promise<TtsSynthesizeResult> {
+  const trimmed = opts.text.trim();
+  if (!trimmed) {
+    throw new Error("TTS text is empty");
+  }
+
+  const max = Math.min(10_000, Math.max(500, config.ttsMaxChars));
+  if (trimmed.length > max) {
+    throw new Error(`TTS text exceeds max length (${max} characters)`);
+  }
+
+  const language = resolveTtsLanguageFromText(trimmed, opts.language);
+  const backend = resolveEffectiveBackend(config, opts.providerOverride);
+
+  let buffer: Buffer;
+  if (backend === "azure") {
+    buffer = await synthesizeAzure({
+      key: config.azureSpeechKey as string,
+      region: config.azureSpeechRegion as string,
+      text: trimmed,
+      language,
+    });
+  } else {
+    const credPath = resolveProjectPath(
+      config.googleApplicationCredentials?.trim() ?? "",
+    );
+    buffer = await synthesizeGoogle({
+      credentialsPath: credPath,
+      text: trimmed,
+      language,
+    });
+  }
+
+  return {
+    buffer,
+    mimeType: "audio/mpeg",
+    filename:
+      language === "he-IL" ? "mercury-tts-he.mp3" : "mercury-tts-en.mp3",
+  };
+}