Merge pull request #1839 from CapSoftware/ai-generation-language-preferences

feat: add AI generation language preferences

Author: richiemcilroy

apps/web/__tests__/unit/generate-ai-title.test.ts

@@ -31,7 +31,10 @@ vi.mock("workflow", () => ({
 
 vi.mock("server-only", () => ({}));
 
-import { shouldReplaceVideoTitle } from "@/workflows/generate-ai";
+import {
+	getAiLanguageInstruction,
+	shouldReplaceVideoTitle,
+} from "@/workflows/generate-ai";
 
 describe("shouldReplaceVideoTitle", () => {
 	it("replaces default Cap titles", () => {
@@ -84,3 +87,15 @@ describe("shouldReplaceVideoTitle", () => {
 		).toBe(false);
 	});
 });
+
+describe("getAiLanguageInstruction", () => {
+	it("uses transcript language when auto-detect is selected", () => {
+		expect(getAiLanguageInstruction("auto")).toContain(
+			"same language as the transcript",
+		);
+	});
+
+	it("uses the selected language name", () => {
+		expect(getAiLanguageInstruction("es")).toContain("Spanish");
+	});
+});

apps/web/__tests__/unit/transcribe-language.test.ts

@@ -0,0 +1,88 @@
+import { describe, expect, it, vi } from "vitest";
+
+vi.mock("@cap/database", () => ({
+	db: vi.fn(),
+}));
+
+vi.mock("@cap/env", () => ({
+	serverEnv: vi.fn(() => ({})),
+}));
+
+vi.mock("@cap/utils", () => ({
+	userIsPro: vi.fn(),
+}));
+
+vi.mock("@cap/web-backend", () => ({
+	Storage: {},
+}));
+
+vi.mock("@deepgram/sdk", () => ({
+	createClient: vi.fn(),
+}));
+
+vi.mock("@/lib/audio-enhance", () => ({
+	ENHANCED_AUDIO_CONTENT_TYPE: "audio/mpeg",
+	ENHANCED_AUDIO_EXTENSION: "mp3",
+	enhanceAudioFromUrl: vi.fn(),
+}));
+
+vi.mock("@/lib/audio-extract", () => ({
+	checkHasAudioTrack: vi.fn(),
+	extractAudioFromUrl: vi.fn(),
+}));
+
+vi.mock("@/lib/generate-ai", () => ({
+	startAiGeneration: vi.fn(),
+}));
+
+vi.mock("@/lib/media-client", () => ({
+	checkHasAudioTrackViaMediaServer: vi.fn(),
+	extractAudioViaMediaServer: vi.fn(),
+	isMediaServerConfigured: vi.fn(),
+	probeVideoViaMediaServer: vi.fn(),
+}));
+
+vi.mock("@/lib/server", () => ({
+	runPromise: vi.fn(),
+}));
+
+vi.mock("@/lib/transcribe-utils", () => ({
+	formatToWebVTT: vi.fn(),
+}));
+
+vi.mock("@/lib/video-storage", () => ({
+	decodeStorageVideo: vi.fn(),
+}));
+
+vi.mock("workflow", () => ({
+	FatalError: class FatalError extends Error {},
+}));
+
+import {
+	AI_GENERATION_LANGUAGES,
+	isAiGenerationLanguage,
+	parseAiGenerationLanguage,
+} from "@cap/web-domain";
+import { getDeepgramTranscriptionOptions } from "@/workflows/transcribe";
+
+describe("AI generation language support", () => {
+	it("does not expose unsupported transcription languages", () => {
+		expect(AI_GENERATION_LANGUAGES).not.toHaveProperty("pa");
+		expect(isAiGenerationLanguage("pa")).toBe(false);
+		expect(parseAiGenerationLanguage("pa")).toBe("auto");
+	});
+
+	it("constrains Deepgram auto-detection to detectable languages", () => {
+		expect(getDeepgramTranscriptionOptions("auto")).toMatchObject({
+			model: "nova-3",
+			detect_language: expect.arrayContaining(["en", "es", "zh"]),
+		});
+	});
+
+	it("passes explicit languages to Deepgram", () => {
+		expect(getDeepgramTranscriptionOptions("zh")).toMatchObject({
+			model: "nova-3",
+			language: "zh",
+		});
+	});
+});

apps/web/actions/organization/settings.ts

@@ -4,6 +4,11 @@ import { db } from "@cap/database";
 import { getCurrentUser } from "@cap/database/auth/session";
 import { organizations } from "@cap/database/schema";
 import { userIsPro } from "@cap/utils";
+import {
+	AI_GENERATION_LANGUAGE_AUTO,
+	type AiGenerationLanguage,
+	isAiGenerationLanguage,
+} from "@cap/web-domain";
 import { eq } from "drizzle-orm";
 import { revalidatePath } from "next/cache";
 import { requireOrganizationSettingsManager } from "./authorization";
@@ -17,6 +22,7 @@ type OrganizationSettingsInput = {
 	disableComments?: boolean;
 	hideShareableLinkCapLogo?: boolean;
 	shareableLinkUseOrganizationIcon?: boolean;
+	aiGenerationLanguage?: AiGenerationLanguage;
 };
 
 const proOrganizationSettingKeys = [
@@ -25,8 +31,21 @@ const proOrganizationSettingKeys = [
 	"disableTranscript",
 	"hideShareableLinkCapLogo",
 	"shareableLinkUseOrganizationIcon",
+	"aiGenerationLanguage",
 ] as const satisfies readonly (keyof OrganizationSettingsInput)[];
 
+const defaultProOrganizationSettings = {
+	disableSummary: false,
+	disableChapters: false,
+	disableTranscript: false,
+	hideShareableLinkCapLogo: false,
+	shareableLinkUseOrganizationIcon: false,
+	aiGenerationLanguage: AI_GENERATION_LANGUAGE_AUTO,
+} as const satisfies Pick<
+	Required<OrganizationSettingsInput>,
+	(typeof proOrganizationSettingKeys)[number]
+>;
+
 const preserveProSettings = (
 	submittedSettings: OrganizationSettingsInput,
 	existingSettings: OrganizationSettingsInput | null | undefined,
@@ -35,7 +54,7 @@ const preserveProSettings = (
 	...Object.fromEntries(
 		proOrganizationSettingKeys.map((key) => [
 			key,
-			existingSettings?.[key] ?? false,
+			existingSettings?.[key] ?? defaultProOrganizationSettings[key],
 		]),
 	),
 });
@@ -53,6 +72,13 @@ export async function updateOrganizationSettings(
 		throw new Error("Settings are required");
 	}
 
+	if (
+		settings.aiGenerationLanguage !== undefined &&
+		!isAiGenerationLanguage(settings.aiGenerationLanguage)
+	) {
+		throw new Error("Unsupported AI generation language");
+	}
+
 	if (!user.activeOrganizationId) {
 		throw new Error("Organization not found");
 	}

apps/web/actions/videos/translation-languages.ts

@@ -1,29 +1,4 @@
-export const SUPPORTED_LANGUAGES = {
-	en: "English",
-	es: "Spanish",
-	fr: "French",
-	de: "German",
-	pt: "Portuguese",
-	it: "Italian",
-	nl: "Dutch",
-	pl: "Polish",
-	sk: "Slovak",
-	ru: "Russian",
-	tr: "Turkish",
-	ja: "Japanese",
-	ko: "Korean",
-	zh: "Chinese (Simplified)",
-	ar: "Arabic",
-	hi: "Hindi",
-	bn: "Bengali",
-	ta: "Tamil",
-	te: "Telugu",
-	mr: "Marathi",
-	gu: "Gujarati",
-	pa: "Punjabi",
-	ur: "Urdu",
-	fa: "Persian",
-	he: "Hebrew",
-} as const;
-
-export type LanguageCode = keyof typeof SUPPORTED_LANGUAGES;
+export {
+	type LanguageCode,
+	SUPPORTED_LANGUAGES,
+} from "@cap/web-domain";