Skip to content

AI-native knowledge base & Evernote/Obsidian alternative with instant sync & native MCP. Free on Cloudflare or self-hosted via Docker.

revert: withdraw unreleased video transcription feature

tianma-if2e15f325c5c2
diff --git a/apps/api/src/ai-routes.test.mjs b/apps/api/src/ai-routes.test.mjs
index 6cb17bc9..af72b279 100644
--- a/apps/api/src/ai-routes.test.mjs
+++ b/apps/api/src/ai-routes.test.mjs
@@ -742,370 +742,3 @@ describe("video outline", () => {
expect(await invalid.json()).toMatchObject({ error: { code: "video_outline_invalid" } });
});
});
-
-const transcriptionToken = "asr-token-should-not-leak-9f3c";
-const otherTranscriptionToken = "speech-test-token-b";
-
-const transcriptionRequest = (app, databaseEnvironment, path, method, body) => app.request(
- path,
- {
- method,
- headers: { "content-type": "application/json" },
- ...(body === undefined ? {} : { body: JSON.stringify(body) }),
- },
- databaseEnvironment,
-);
-
-const createTranscriptionProvider = (app, databaseEnvironment, body) => transcriptionRequest(
- app,
- databaseEnvironment,
- "/api/v1/ai/transcription-providers",
- "POST",
- body,
-);
-
-describe("speech transcription settings", () => {
- test("stores several encrypted services and never returns their tokens", async () => {
- const app = createApp();
- const { sqlite, environment: databaseEnvironment } = createDatabaseEnvironment();
- try {
- const empty = await app.request("/api/v1/ai/transcription-settings", {}, databaseEnvironment);
- expect(empty.status).toBe(200);
- expect(await empty.json()).toEqual({
- providers: [],
- defaultModelId: null,
- enabled: false,
- encryptionConfigured: true,
- readOnly: false,
- });
-
- const saved = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: " Groq ",
- baseUrl: "https://api.groq.com/openai/v1/",
- apiKey: ` ${transcriptionToken} `,
- initialModelId: " whisper-large-v3-turbo ",
- });
- expect(saved.status).toBe(201);
- const first = await saved.json();
- expect(first.providers).toHaveLength(1);
- expect(first.providers[0]).toMatchObject({
- provider: "openai-compatible",
- displayName: "Groq",
- baseUrl: "https://api.groq.com/openai/v1",
- isEnabled: true,
- hasApiKey: true,
- });
- expect(first.providers[0].models).toEqual([
- expect.objectContaining({
- modelId: "whisper-large-v3-turbo",
- displayName: "whisper-large-v3-turbo",
- }),
- ]);
- expect(first.defaultModelId).toBe(first.providers[0].models[0].id);
- expect(first.enabled).toBe(true);
- expect(JSON.stringify(first)).not.toContain(transcriptionToken);
- const chatSettings = await app.request("/api/v1/ai/settings", {}, databaseEnvironment);
- expect((await chatSettings.json()).providers).toEqual([]);
- expect(JSON.stringify(await app.request("/api/v1/ai/settings", {}, databaseEnvironment).then((response) => response.json()))).not.toContain(transcriptionToken);
-
- const stored = sqlite.query(
- "SELECT base_url, api_key_encrypted FROM ai_transcription_providers WHERE id = ?",
- ).get(first.providers[0].id);
- expect(stored.base_url).toBe("https://api.groq.com/openai/v1");
- expect(stored.api_key_encrypted.startsWith("v1.")).toBe(true);
- expect(stored.api_key_encrypted).not.toContain(transcriptionToken);
- const audits = sqlite.query(
- "SELECT metadata_json FROM audit_events WHERE action = ?",
- ).all("workspace.ai_transcription_provider.create");
- expect(audits).toHaveLength(1);
- expect(audits[0].metadata_json).not.toContain(transcriptionToken);
- expect(JSON.parse(audits[0].metadata_json)).toEqual({
- isEnabled: true,
- provider: "openai-compatible",
- modelId: "whisper-large-v3-turbo",
- });
-
- const second = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Local",
- baseUrl: "http://127.0.0.1:8080/v1",
- apiKey: otherTranscriptionToken,
- initialModelId: "local-whisper",
- });
- expect(second.status).toBe(201);
- const both = await second.json();
- expect(both.providers.map((provider) => provider.displayName).sort()).toEqual(["Groq", "Local"]);
- expect(both.defaultModelId).toBe(first.defaultModelId);
- expect(both.enabled).toBe(true);
- expect(JSON.stringify(both)).not.toContain(transcriptionToken);
- expect(JSON.stringify(both)).not.toContain(otherTranscriptionToken);
-
- const kept = await transcriptionRequest(
- app,
- databaseEnvironment,
- `/api/v1/ai/transcription-providers/${first.providers[0].id}`,
- "PUT",
- {
- displayName: "Groq",
- baseUrl: "https://api.groq.com/openai/v1",
- isEnabled: false,
- },
- );
- expect(kept.status).toBe(200);
- const disabled = await kept.json();
- expect(disabled.providers.find((provider) => provider.id === first.providers[0].id)).toMatchObject({
- isEnabled: false,
- hasApiKey: true,
- });
- expect(disabled.defaultModelId).toBe(first.defaultModelId);
- expect(disabled.enabled).toBe(false);
- expect(sqlite.query(
- "SELECT api_key_encrypted FROM ai_transcription_providers WHERE id = ?",
- ).get(first.providers[0].id).api_key_encrypted).toBe(stored.api_key_encrypted);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM ai_provider_configs").get().count).toBe(0);
- } finally {
- sqlite.close();
- }
- });
-
- test("rejects a service without a token, with a blank model, or with an embedded credential", async () => {
- const app = createApp();
- const { sqlite, environment: databaseEnvironment } = createDatabaseEnvironment();
- try {
- const missingToken = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Example",
- baseUrl: "https://api.example.com/v1",
- initialModelId: "whisper-1",
- });
- expect(missingToken.status).toBe(400);
- expect(await missingToken.json()).toMatchObject({
- error: { code: "ai_transcription_request_invalid" },
- });
- const blankModel = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Example",
- baseUrl: "https://api.example.com/v1",
- apiKey: transcriptionToken,
- initialModelId: " ",
- });
- expect(blankModel.status).toBe(400);
- expect(JSON.stringify(await blankModel.json())).not.toContain(transcriptionToken);
-
- const embedded = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Example",
- baseUrl: "https://user:[email protected]/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(embedded.status).toBe(400);
- expect(await embedded.json()).toMatchObject({
- error: { code: "ai_transcription_request_invalid" },
- });
- const ftp = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Example",
- baseUrl: "ftp://files.example.com/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(ftp.status).toBe(400);
- expect(JSON.stringify(await ftp.json())).not.toContain(transcriptionToken);
- const otherStandard = await createTranscriptionProvider(app, databaseEnvironment, {
- provider: "anthropic",
- displayName: "Example",
- baseUrl: "https://api.example.com/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(otherStandard.status).toBe(400);
- const otherStandardBody = await otherStandard.json();
- expect(otherStandardBody).toMatchObject({
- error: { code: "ai_transcription_request_invalid" },
- });
- expect(JSON.stringify(otherStandardBody)).not.toContain(transcriptionToken);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM ai_transcription_providers").get().count).toBe(0);
- } finally {
- sqlite.close();
- }
- });
-
- test("keeps a saved token when the encryption key is later unavailable, and still refuses a replacement", async () => {
- const app = createApp();
- const { sqlite, environment: databaseEnvironment } = createDatabaseEnvironment();
- const withoutKey = { ...databaseEnvironment, EDGE_EVER_AUTH_PASSWORD: undefined };
- try {
- const created = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "OpenAI",
- baseUrl: "https://api.openai.com/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(created.status).toBe(201);
- const providerId = (await created.json()).providers[0].id;
- const stored = sqlite.query(
- "SELECT api_key_encrypted FROM ai_transcription_providers WHERE id = ?",
- ).get(providerId).api_key_encrypted;
-
- const renamed = await transcriptionRequest(app, withoutKey, `/api/v1/ai/transcription-providers/${providerId}`, "PUT", {
- displayName: "OpenAI renamed",
- baseUrl: "https://api.openai.com/v1",
- isEnabled: true,
- });
- expect(renamed.status).toBe(200);
- expect(await renamed.json()).toMatchObject({
- encryptionConfigured: false,
- enabled: false,
- providers: [expect.objectContaining({ displayName: "OpenAI renamed", hasApiKey: true })],
- });
- expect(sqlite.query(
- "SELECT api_key_encrypted FROM ai_transcription_providers WHERE id = ?",
- ).get(providerId).api_key_encrypted).toBe(stored);
-
- const replaced = await transcriptionRequest(app, withoutKey, `/api/v1/ai/transcription-providers/${providerId}`, "PUT", {
- displayName: "OpenAI renamed",
- baseUrl: "https://api.openai.com/v1",
- isEnabled: true,
- apiKey: "replacement-token",
- });
- expect(replaced.status).toBe(400);
- const replacedBody = await replaced.json();
- expect(replacedBody).toMatchObject({ error: { code: "ai_encryption_key_missing" } });
- expect(JSON.stringify(replacedBody)).not.toContain("replacement-token");
- expect(sqlite.query(
- "SELECT api_key_encrypted FROM ai_transcription_providers WHERE id = ?",
- ).get(providerId).api_key_encrypted).toBe(stored);
- } finally {
- sqlite.close();
- }
- });
-
- test("refuses demo mode and API tokens", async () => {
- const { sqlite, environment: databaseEnvironment } = createDatabaseEnvironment();
- try {
- const demo = await createTranscriptionProvider(createApp({ demoMode: true }), databaseEnvironment, {
- displayName: "Example",
- baseUrl: "https://api.example.com/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(demo.status).toBe(403);
- const demoRead = await createApp({ demoMode: true }).request(
- "/api/v1/ai/transcription-settings",
- {},
- databaseEnvironment,
- );
- expect(demoRead.status).toBe(200);
- expect(await demoRead.json()).toMatchObject({ readOnly: true, providers: [], enabled: false });
-
- const tokenApp = createApp({ currentAuth: { ...auth, kind: "agent", actorType: "agent", scopes: ["ai:generate"] } });
- const deniedRead = await tokenApp.request("/api/v1/ai/transcription-settings", {}, environment);
- const deniedWrite = await createTranscriptionProvider(tokenApp, environment, {
- displayName: "Example",
- baseUrl: "https://api.example.com/v1",
- apiKey: transcriptionToken,
- initialModelId: "whisper-1",
- });
- expect(deniedRead.status).toBe(403);
- expect(deniedWrite.status).toBe(403);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM ai_transcription_providers").get().count).toBe(0);
- expect(JSON.stringify(await deniedWrite.json())).not.toContain(transcriptionToken);
- } finally {
- sqlite.close();
- }
- });
-
- test("prepares only the selected enabled service for an interactive session", async () => {
- const app = createApp();
- const { sqlite, environment: databaseEnvironment } = createDatabaseEnvironment();
- const prepare = () => app.request(
- "/api/v1/ai/transcription-settings/prepare",
- { method: "POST" },
- databaseEnvironment,
- );
- try {
- const disabled = await prepare();
- expect(disabled.status).toBe(200);
- expect(await disabled.json()).toEqual({ enabled: false });
-
- const firstResponse = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Groq",
- baseUrl: "https://api.groq.com/openai/v1/",
- apiKey: transcriptionToken,
- initialModelId: "whisper-large-v3-turbo",
- });
- const first = await firstResponse.json();
- const secondResponse = await createTranscriptionProvider(app, databaseEnvironment, {
- displayName: "Local",
- baseUrl: "http://127.0.0.1:8080/v1",
- apiKey: otherTranscriptionToken,
- initialModelId: "local-whisper",
- });
- const second = await secondResponse.json();
- const localModelId = second.providers.find((provider) => provider.displayName === "Local").models[0].id;
- const prepared = await prepare();
- expect(prepared.status).toBe(200);
- expect(await prepared.json()).toEqual({
- enabled: true,
- provider: "openai-compatible",
- baseUrl: "https://api.groq.com/openai/v1",
- modelId: "whisper-large-v3-turbo",
- apiKey: transcriptionToken,
- });
- const listed = await app.request("/api/v1/ai/transcription-settings", {}, databaseEnvironment);
- const listedBody = JSON.stringify(await listed.json());
- expect(listedBody).not.toContain(transcriptionToken);
- expect(listedBody).not.toContain(otherTranscriptionToken);
-
- const switched = await transcriptionRequest(
- app,
- databaseEnvironment,
- "/api/v1/ai/transcription-default-model",
- "PUT",
- { modelConfigId: localModelId },
- );
- expect(switched.status).toBe(200);
- expect((await switched.json()).defaultModelId).toBe(localModelId);
- expect(await (await prepare()).json()).toEqual({
- enabled: true,
- provider: "openai-compatible",
- baseUrl: "http://127.0.0.1:8080/v1",
- modelId: "local-whisper",
- apiKey: otherTranscriptionToken,
- });
-
- const cleared = await transcriptionRequest(
- app,
- databaseEnvironment,
- "/api/v1/ai/transcription-default-model",
- "PUT",
- { modelConfigId: null },
- );
- expect(cleared.status).toBe(200);
- const turnedOff = await prepare();
- const turnedOffBody = await turnedOff.json();
- expect(turnedOffBody).toEqual({ enabled: false });
- expect(JSON.stringify(turnedOffBody)).not.toContain(transcriptionToken);
- expect(JSON.stringify(turnedOffBody)).not.toContain(otherTranscriptionToken);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM ai_transcription_providers").get().count).toBe(2);
-
- const demo = await createApp({ demoMode: true }).request(
- "/api/v1/ai/transcription-settings/prepare",
- { method: "POST" },
- databaseEnvironment,
- );
- expect(demo.status).toBe(403);
- expect(JSON.stringify(await demo.json())).not.toContain(transcriptionToken);
-
- const tokenApp = createApp({
- currentAuth: { ...auth, kind: "agent", actorType: "agent", scopes: ["ai:generate", "write:memos"] },
- });
- const denied = await tokenApp.request(
- "/api/v1/ai/transcription-settings/prepare",
- { method: "POST" },
- databaseEnvironment,
- );
- expect(denied.status).toBe(403);
- expect(JSON.stringify(await denied.json())).not.toContain(transcriptionToken);
- expect(first.providers[0].id).not.toBe(localModelId);
- } finally {
- sqlite.close();
- }
- });
-});
diff --git a/apps/api/src/ai-routes.ts b/apps/api/src/ai-routes.ts
index 5077068f..55c90eba 100644
--- a/apps/api/src/ai-routes.ts
+++ b/apps/api/src/ai-routes.ts
@@ -1,10 +1,6 @@
import {
AiDefaultModelUpdateSchema,
AiGenerateSchema,
- AiTranscriptionDefaultModelUpdateSchema,
- AiTranscriptionModelCreateSchema,
- AiTranscriptionProviderCreateSchema,
- AiTranscriptionProviderUpdateSchema,
InfographicAgentRequestSchema,
AiModelConfigCreateSchema,
AiProviderConfigCreateSchema,
@@ -44,12 +40,6 @@ import {
normalizeAiGenerationText,
normalizeAiBaseUrl,
prepareAiGeneration,
- prepareAiTranscriptionCredentials,
- aiTranscriptionDefaultStatement,
- getAiTranscriptionModel,
- getAiTranscriptionProvider,
- getAiTranscriptionSettings,
- getDefaultAiTranscriptionModelId,
resolvePrimaryAiCredentialEncryptionKey,
streamAiGeneration,
testAiModel,
@@ -119,28 +109,6 @@ const readSettings = (context: AppContext, dependencies: AiRouteDependencies) =>
context.env,
);
-const readTranscriptionSettings = (context: AppContext, dependencies: AiRouteDependencies) =>
- getAiTranscriptionSettings(
- context.env.storage.db,
- getWorkspaceId(context),
- encryptionConfigured(context),
- dependencies.isDemoMode(context.env),
- context.env,
- );
-
-const invalidTranscriptionSettings = (
- result: { success: boolean },
- context: Context,
-) => {
- if (result.success) return;
- return apiError(
- context,
- "ai_transcription_request_invalid",
- "The speech transcription settings are invalid.",
- 400,
- );
-};
-
const denyMutation = (context: AppContext, dependencies: AiRouteDependencies) => {
const denied = requireUser(context);
if (denied) return denied;
@@ -246,296 +214,6 @@ export const registerAiRoutes = (app: Hono<AppEnv>, dependencies: AiRouteDepende
return context.json(await readSettings(context, dependencies));
});
- app.get("/api/v1/ai/transcription-settings", async (context) => {
- const denied = requireUser(context);
- if (denied) return denied;
- return context.json(await readTranscriptionSettings(context, dependencies));
- });
-
- app.post(
- "/api/v1/ai/transcription-providers",
- zValidator("json", AiTranscriptionProviderCreateSchema, invalidTranscriptionSettings),
- async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- try {
- const input = context.req.valid("json");
- const workspaceId = getWorkspaceId(context);
- const providerId = createId("atp");
- const modelConfigId = input.initialModelId ? createId("atm") : null;
- const now = isoNow();
- const statements = [
- context.env.storage.db.prepare(
- `INSERT INTO ai_transcription_providers (
- id, workspace_id, provider, display_name, base_url, api_key_encrypted,
- is_enabled, created_at, updated_at
- ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,
- ).bind(
- providerId,
- workspaceId,
- input.provider,
- input.displayName,
- normalizeAiBaseUrl(input.baseUrl),
- await encryptSecret(input.apiKey, requireEncryptionKey(context)),
- input.isEnabled ? 1 : 0,
- now,
- now,
- ),
- ];
- if (modelConfigId && input.initialModelId) {
- statements.push(context.env.storage.db.prepare(
- `INSERT INTO ai_transcription_models (
- id, provider_id, model_id, display_name, created_at, updated_at
- ) VALUES (?, ?, ?, ?, ?, ?)`,
- ).bind(
- modelConfigId,
- providerId,
- input.initialModelId,
- input.initialModelId.slice(0, 80),
- now,
- now,
- ));
- if (input.isEnabled && !(await getDefaultAiTranscriptionModelId(context.env.storage.db, workspaceId))) {
- statements.push(aiTranscriptionDefaultStatement(
- context.env.storage.db,
- workspaceId,
- modelConfigId,
- now,
- ));
- }
- }
- statements.push(auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_provider.create",
- "ai_transcription_provider",
- providerId,
- { isEnabled: input.isEnabled, provider: input.provider, modelId: input.initialModelId ?? null },
- ));
- await context.env.storage.db.batch(statements);
- return context.json(await readTranscriptionSettings(context, dependencies), 201);
- } catch (error) {
- return withAiError(context, error, "ai_transcription_provider_create_failed");
- }
- },
- );
-
- app.put(
- "/api/v1/ai/transcription-providers/:providerId",
- zValidator("json", AiTranscriptionProviderUpdateSchema, invalidTranscriptionSettings),
- async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- try {
- const input = context.req.valid("json");
- const providerId = context.req.param("providerId");
- const workspaceId = getWorkspaceId(context);
- const existing = await getAiTranscriptionProvider(context.env.storage.db, workspaceId, providerId);
- if (!existing) return notFound(context, "Speech service not found.");
- const now = isoNow();
- const apiKeyEncrypted = input.apiKey
- ? await encryptSecret(input.apiKey, requireEncryptionKey(context))
- : existing.api_key_encrypted;
- await context.env.storage.db.batch([
- context.env.storage.db.prepare(
- `UPDATE ai_transcription_providers SET
- provider = ?, display_name = ?, base_url = ?, api_key_encrypted = ?,
- is_enabled = ?, updated_at = ?
- WHERE id = ? AND workspace_id = ?`,
- ).bind(
- input.provider,
- input.displayName,
- normalizeAiBaseUrl(input.baseUrl),
- apiKeyEncrypted,
- input.isEnabled ? 1 : 0,
- now,
- providerId,
- workspaceId,
- ),
- auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_provider.update",
- "ai_transcription_provider",
- providerId,
- { isEnabled: input.isEnabled, provider: input.provider, apiKeyUpdated: Boolean(input.apiKey) },
- ),
- ]);
- return context.json(await readTranscriptionSettings(context, dependencies));
- } catch (error) {
- return withAiError(context, error, "ai_transcription_provider_update_failed");
- }
- },
- );
-
- app.delete("/api/v1/ai/transcription-providers/:providerId", async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- const providerId = context.req.param("providerId");
- const workspaceId = getWorkspaceId(context);
- const existing = await getAiTranscriptionProvider(context.env.storage.db, workspaceId, providerId);
- if (!existing) return notFound(context, "Speech service not found.");
- await context.env.storage.db.batch([
- context.env.storage.db.prepare(
- `DELETE FROM ai_transcription_providers WHERE id = ? AND workspace_id = ?`,
- ).bind(providerId, workspaceId),
- auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_provider.delete",
- "ai_transcription_provider",
- providerId,
- {},
- ),
- ]);
- return context.json(await readTranscriptionSettings(context, dependencies));
- });
-
- app.post(
- "/api/v1/ai/transcription-providers/:providerId/models",
- zValidator("json", AiTranscriptionModelCreateSchema, invalidTranscriptionSettings),
- async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- const providerId = context.req.param("providerId");
- const workspaceId = getWorkspaceId(context);
- const provider = await getAiTranscriptionProvider(context.env.storage.db, workspaceId, providerId);
- if (!provider) return notFound(context, "Speech service not found.");
- const input = context.req.valid("json");
- const modelConfigId = createId("atm");
- const now = isoNow();
- const statements = [
- context.env.storage.db.prepare(
- `INSERT INTO ai_transcription_models (
- id, provider_id, model_id, display_name, created_at, updated_at
- ) VALUES (?, ?, ?, ?, ?, ?)`,
- ).bind(
- modelConfigId,
- providerId,
- input.modelId,
- input.displayName ?? input.modelId.slice(0, 80),
- now,
- now,
- ),
- ];
- if (provider.is_enabled && !(await getDefaultAiTranscriptionModelId(context.env.storage.db, workspaceId))) {
- statements.push(aiTranscriptionDefaultStatement(
- context.env.storage.db,
- workspaceId,
- modelConfigId,
- now,
- ));
- }
- statements.push(auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_model.create",
- "ai_transcription_model",
- modelConfigId,
- { providerId, modelId: input.modelId },
- ));
- try {
- await context.env.storage.db.batch(statements);
- return context.json(await readTranscriptionSettings(context, dependencies), 201);
- } catch (error) {
- return withAiError(context, error, "ai_transcription_model_create_failed");
- }
- },
- );
-
- app.delete(
- "/api/v1/ai/transcription-providers/:providerId/models/:modelConfigId",
- async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- const workspaceId = getWorkspaceId(context);
- const providerId = context.req.param("providerId");
- const modelConfigId = context.req.param("modelConfigId");
- const model = await getAiTranscriptionModel(context.env.storage.db, workspaceId, modelConfigId);
- if (!model || model.provider_id !== providerId) {
- return notFound(context, "Speech model not found.");
- }
- await context.env.storage.db.batch([
- context.env.storage.db.prepare(
- `DELETE FROM ai_transcription_models WHERE id = ? AND provider_id = ?`,
- ).bind(modelConfigId, providerId),
- auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_model.delete",
- "ai_transcription_model",
- modelConfigId,
- { providerId, modelId: model.model_id },
- ),
- ]);
- return context.json(await readTranscriptionSettings(context, dependencies));
- },
- );
-
- app.put(
- "/api/v1/ai/transcription-default-model",
- zValidator("json", AiTranscriptionDefaultModelUpdateSchema, invalidTranscriptionSettings),
- async (context) => {
- const denied = denyMutation(context, dependencies);
- if (denied) return denied;
- const input = context.req.valid("json");
- const workspaceId = getWorkspaceId(context);
- if (input.modelConfigId) {
- const model = await context.env.storage.db.prepare(
- `SELECT models.id
- FROM ai_transcription_models AS models
- JOIN ai_transcription_providers AS providers ON providers.id = models.provider_id
- WHERE models.id = ? AND providers.workspace_id = ? AND providers.is_enabled = 1
- LIMIT 1`,
- ).bind(input.modelConfigId, workspaceId).first<{ id: string }>();
- if (!model) {
- return apiError(
- context,
- "ai_transcription_default_unavailable",
- "The selected speech model belongs to a disabled or missing service.",
- 400,
- );
- }
- }
- const now = isoNow();
- await context.env.storage.db.batch([
- aiTranscriptionDefaultStatement(context.env.storage.db, workspaceId, input.modelConfigId, now),
- auditStatement(
- context.env.storage.db,
- "user",
- context.get("auth").actorId,
- "workspace.ai_transcription_default.update",
- "workspace",
- workspaceId,
- { modelConfigId: input.modelConfigId },
- ),
- ]);
- return context.json(await readTranscriptionSettings(context, dependencies));
- },
- );
-
- app.post("/api/v1/ai/transcription-settings/prepare", async (context) => {
- const denied = requireUser(context);
- if (denied) return denied;
- if (dependencies.isDemoMode(context.env)) {
- return forbidden(context, "Speech transcription credentials are unavailable in demo mode.");
- }
- try {
- return context.json(await prepareAiTranscriptionCredentials(
- context.env.storage.db,
- getWorkspaceId(context),
- context.env,
- ));
- } catch (error) {
- return withAiError(context, error, "ai_transcription_prepare_failed");
- }
- });
-
app.post(
"/api/v1/ai/providers",
zValidator("json", AiProviderConfigCreateSchema),
diff --git a/apps/api/src/ai-service.ts b/apps/api/src/ai-service.ts
index 620b3717..7b11b2ab 100644
--- a/apps/api/src/ai-service.ts
+++ b/apps/api/src/ai-service.ts
@@ -8,10 +8,6 @@ import type {
AiProvider,
AiProviderConfig,
AiSettings,
- AiTranscriptionModel,
- AiTranscriptionProvider,
- AiTranscriptionStandard,
- AiTranscriptionSettings,
AiTargetLanguage,
AiTone,
} from "@edgeever/shared";
@@ -228,197 +224,6 @@ export const getAiSettings = async (
export const normalizeAiBaseUrl = (value: string) => value.trim().replace(/\/+$/, "");
-export type AiTranscriptionProviderRow = {
- id: string;
- workspace_id: string;
- provider: AiTranscriptionStandard;
- display_name: string;
- base_url: string;
- api_key_encrypted: string;
- is_enabled: number;
- created_at: string;
- updated_at: string;
-};
-
-export type AiTranscriptionModelRow = {
- id: string;
- provider_id: string;
- model_id: string;
- display_name: string;
- created_at: string;
- updated_at: string;
-};
-
-const selectTranscriptionProviderSql = `SELECT
- id, workspace_id, provider, display_name, base_url, api_key_encrypted, is_enabled, created_at, updated_at
- FROM ai_transcription_providers`;
-
-const selectTranscriptionModelSql = `SELECT
- id, provider_id, model_id, display_name, created_at, updated_at
- FROM ai_transcription_models`;
-
-export const getAiTranscriptionProvider = (
- db: DatabaseAdapter,
- workspaceId: string,
- providerId: string,
-) => db.prepare(
- `${selectTranscriptionProviderSql} WHERE id = ? AND workspace_id = ? LIMIT 1`,
-).bind(providerId, workspaceId).first<AiTranscriptionProviderRow>();
-
-export const getAiTranscriptionModel = (
- db: DatabaseAdapter,
- workspaceId: string,
- modelConfigId: string,
-) => db.prepare(
- `${selectTranscriptionModelSql}
- WHERE id = ? AND provider_id IN (
- SELECT id FROM ai_transcription_providers WHERE workspace_id = ?
- )
- LIMIT 1`,
-).bind(modelConfigId, workspaceId).first<AiTranscriptionModelRow>();
-
-export const getDefaultAiTranscriptionModelId = async (db: DatabaseAdapter, workspaceId: string) => {
- const row = await db.prepare(
- `SELECT default_model_id
- FROM ai_transcription_workspace_settings
- WHERE workspace_id = ?
- LIMIT 1`,
- ).bind(workspaceId).first<{ default_model_id: string | null }>();
- return row?.default_model_id ?? null;
-};
-
-export const aiTranscriptionDefaultStatement = (
- db: DatabaseAdapter,
- workspaceId: string,
- modelConfigId: string | null,
- now: string,
-) => db.prepare(
- `INSERT INTO ai_transcription_workspace_settings (
- workspace_id, default_model_id, created_at, updated_at
- ) VALUES (?, ?, ?, ?)
- ON CONFLICT(workspace_id) DO UPDATE SET
- default_model_id = excluded.default_model_id,
- updated_at = excluded.updated_at`,
-).bind(workspaceId, modelConfigId, now, now);
-
-const mapAiTranscriptionModel = (row: AiTranscriptionModelRow): AiTranscriptionModel => ({
- id: row.id,
- providerId: row.provider_id,
- modelId: row.model_id,
- displayName: row.display_name,
-});
-
-const mapAiTranscriptionProvider = (
- row: AiTranscriptionProviderRow,
- models: AiTranscriptionModelRow[],
- credentialsUnavailable = false,
-): AiTranscriptionProvider => ({
- id: row.id,
- provider: row.provider,
- displayName: row.display_name,
- baseUrl: row.base_url,
- isEnabled: Boolean(row.is_enabled),
- hasApiKey: Boolean(row.api_key_encrypted),
- models: models.filter((model) => model.provider_id === row.id).map(mapAiTranscriptionModel),
- ...(credentialsUnavailable ? { credentialsUnavailable: true } : {}),
-});
-
-const withAiTranscriptionCredentialAvailability = async (
- row: AiTranscriptionProviderRow,
- models: AiTranscriptionModelRow[],
- encryptionConfigured: boolean,
- environment?: AiCredentialEnvironment,
-): Promise<AiTranscriptionProvider> => {
- const mapped = mapAiTranscriptionProvider(row, models);
- if (!encryptionConfigured || !environment || !row.api_key_encrypted) return mapped;
- try {
- await decryptAiCredential(row.api_key_encrypted, environment);
- return mapped;
- } catch {
- return mapAiTranscriptionProvider(row, models, true);
- }
-};
-
-export const getAiTranscriptionSettings = async (
- db: DatabaseAdapter,
- workspaceId: string,
- encryptionConfigured: boolean,
- readOnly: boolean,
- environment?: AiCredentialEnvironment,
-): Promise<AiTranscriptionSettings> => {
- const [providersResult, modelsResult, defaultModelId] = await Promise.all([
- db.prepare(
- `${selectTranscriptionProviderSql} WHERE workspace_id = ? ORDER BY created_at ASC, id ASC`,
- ).bind(workspaceId).all<AiTranscriptionProviderRow>(),
- db.prepare(
- `${selectTranscriptionModelSql}
- WHERE provider_id IN (
- SELECT id FROM ai_transcription_providers WHERE workspace_id = ?
- )
- ORDER BY created_at ASC, id ASC`,
- ).bind(workspaceId).all<AiTranscriptionModelRow>(),
- getDefaultAiTranscriptionModelId(db, workspaceId),
- ]);
- const providers = await Promise.all(providersResult.results.map((provider) =>
- withAiTranscriptionCredentialAvailability(provider, modelsResult.results, encryptionConfigured, environment)));
- const defaultProvider = providers.find((provider) =>
- provider.models.some((model) => model.id === defaultModelId));
- const enabled = Boolean(
- encryptionConfigured
- && defaultProvider?.isEnabled
- && defaultProvider.hasApiKey
- && !defaultProvider.credentialsUnavailable
- && defaultProvider.models.some((model) => model.id === defaultModelId && model.modelId),
- );
- return {
- providers,
- defaultModelId,
- enabled,
- encryptionConfigured,
- readOnly,
- };
-};
-
-export const prepareAiTranscriptionCredentials = async (
- db: DatabaseAdapter,
- workspaceId: string,
- environment: AiCredentialEnvironment,
-) => {
- const row = await db.prepare(
- `SELECT
- models.model_id,
- providers.provider,
- providers.base_url,
- providers.api_key_encrypted
- FROM ai_transcription_workspace_settings AS settings
- JOIN ai_transcription_models AS models ON models.id = settings.default_model_id
- JOIN ai_transcription_providers AS providers ON providers.id = models.provider_id
- WHERE settings.workspace_id = ?
- AND providers.workspace_id = ?
- AND providers.is_enabled = 1
- LIMIT 1`,
- ).bind(workspaceId, workspaceId).first<{
- model_id: string;
- provider: AiTranscriptionStandard;
- base_url: string;
- api_key_encrypted: string;
- }>();
- if (!row?.base_url || !row.model_id || !row.api_key_encrypted || !row.provider) {
- return { enabled: false as const };
- }
- try {
- return {
- enabled: true as const,
- provider: row.provider,
- baseUrl: normalizeAiBaseUrl(row.base_url),
- modelId: row.model_id,
- apiKey: await decryptAiCredential(row.api_key_encrypted, environment),
- };
- } catch {
- return { enabled: false as const };
- }
-};
-
const loadAiRuntime = () => import("./ai-runtime");
export const createAiModel = async (config: {
diff --git a/apps/api/src/index.ts b/apps/api/src/index.ts
index 977b592f..7662a973 100644
--- a/apps/api/src/index.ts
+++ b/apps/api/src/index.ts
@@ -90,7 +90,6 @@ import {
import { registerPluginDistributionRoutes } from "./plugin-distribution-routes";
import { registerSyncRoutes } from "./sync-routes";
import { registerMemoRoutes } from "./memo-routes";
-import { registerVideoTranscriptRoutes } from "./video-transcript-routes";
import { registerScheduledTaskRoutes } from "./scheduled-task-routes";
import { registerWorkspaceExtensionRoutes } from "./workspace-extension-routes";
import { registerBackupRoutes } from "./backup-routes";
@@ -342,7 +341,6 @@ registerTemplateRoutes(app, {
getMemoDetail: (...args) => getMemoDetail(...args),
});
-registerVideoTranscriptRoutes(app, { isDemoMode: (...args) => isDemoMode(...args) });
registerMemoRoutes(app, {
clampNumber: (...args) => clampNumber(...args),
createImageResource: (...args) => createImageResource(...args),
diff --git a/apps/api/src/memo-list-service.ts b/apps/api/src/memo-list-service.ts
index ef778573..761c2ad4 100644
--- a/apps/api/src/memo-list-service.ts
+++ b/apps/api/src/memo-list-service.ts
@@ -6,7 +6,6 @@ import {
getDiagramSummary,
getInfographicSummary,
getTableSummary,
- getVideoNoteSummary,
type MemoSummary,
} from "@edgeever/shared";
import { clampNumber, parseJsonArray } from "./entity-utils";
@@ -75,7 +74,6 @@ export const mapMemoSummary = (row: MemoSummaryRow): MemoSummary => ({
...getDiagramSummary(row.content_markdown),
...getInfographicSummary(row.content_markdown),
...getTableSummary(row.content_markdown),
- ...getVideoNoteSummary(row.content_markdown),
tags: parseJsonArray(row.tags_json),
isPinned: Boolean(row.is_pinned),
isArchived: Boolean(row.is_archived),
diff --git a/apps/api/src/memo-routes.ts b/apps/api/src/memo-routes.ts
index d9ca731d..fdd8bb6e 100644
--- a/apps/api/src/memo-routes.ts
+++ b/apps/api/src/memo-routes.ts
@@ -151,12 +151,10 @@ export const registerMemoRoutes = (
try {
const input = context.req.valid("json");
- const { videoTranscript, ...memoInput } = input;
- void videoTranscript;
const memo = await dependencies.createMemo(
context.env.storage.db,
getWorkspaceId(context),
- memoInput,
+ input,
getAuditActor(context),
getActorLabel(context),
);
diff --git a/apps/api/src/video-transcript-routes.ts b/apps/api/src/video-transcript-routes.ts
deleted file mode 100644
index 11ccac1d..00000000
--- a/apps/api/src/video-transcript-routes.ts
+++ /dev/null
@@ -1,82 +0,0 @@
-import { VideoTranscriptFinishSchema } from "@edgeever/shared";
-import { zValidator } from "@hono/zod-validator";
-import type { Hono } from "hono";
-import type { AppContext, AppEnv, Bindings } from "./api-context";
-import { apiError, databaseNotReady, forbidden, notFound } from "./http-errors";
-import { getWorkspaceId, requireUser } from "./request-auth";
-import {
- claimVideoTranscriptJob,
- extractVideoTranscript,
- finishVideoTranscriptJob,
- listClaimableVideoTranscriptJobs,
-} from "./video-transcript-service";
-
-const requireInteractiveUser = (context: AppContext, isDemoMode: (env: Bindings) => boolean) => {
- const denied = requireUser(context);
- if (denied) return denied;
- if (isDemoMode(context.env)) {
- return forbidden(context, "Speech transcription is unavailable in demo mode.");
- }
- return null;
-};
-
-export const registerVideoTranscriptRoutes = (
- app: Hono<AppEnv>,
- dependencies: { isDemoMode: (env: Bindings) => boolean },
-) => {
- app.get("/api/v1/video-transcript-jobs", async (context) => {
- const denied = requireInteractiveUser(context, dependencies.isDemoMode);
- if (denied) return denied;
- const jobs = await listClaimableVideoTranscriptJobs(context.env.storage.db, getWorkspaceId(context));
- return context.json({ jobs });
- });
-
- app.post("/api/v1/video-transcript-jobs/:memoId/extract", async (context) => {
- const denied = requireInteractiveUser(context, dependencies.isDemoMode);
- if (denied) return denied;
- const result = await extractVideoTranscript(
- context.env.storage.db,
- getWorkspaceId(context),
- context.req.param("memoId"),
- );
- if (result.status === "missing") {
- return notFound(context, "This note is not a video note.");
- }
- if (result.status === "unavailable") return databaseNotReady(context);
- return context.json({ status: result.status });
- });
-
- app.post("/api/v1/video-transcript-jobs/:memoId/claim", async (context) => {
- const denied = requireInteractiveUser(context, dependencies.isDemoMode);
- if (denied) return denied;
- const job = await claimVideoTranscriptJob(
- context.env.storage.db,
- getWorkspaceId(context),
- context.req.param("memoId"),
- );
- if (!job) {
- return apiError(context, "video_transcript_not_claimed", "This transcript job is not waiting.", 409);
- }
- return context.json({ job });
- });
-
- app.post(
- "/api/v1/video-transcript-jobs/:memoId/finish",
- zValidator("json", VideoTranscriptFinishSchema),
- async (context) => {
- const denied = requireInteractiveUser(context, dependencies.isDemoMode);
- if (denied) return denied;
- const input = context.req.valid("json");
- const finished = await finishVideoTranscriptJob(
- context.env.storage.db,
- getWorkspaceId(context),
- context.req.param("memoId"),
- { status: input.status, errorCode: input.errorCode ?? null, claimedAt: input.claimedAt },
- );
- if (!finished) {
- return apiError(context, "video_transcript_not_claimed", "This transcript job is not claimed.", 409);
- }
- return context.json({ ok: true });
- },
- );
-};
diff --git a/apps/api/src/video-transcript-service.ts b/apps/api/src/video-transcript-service.ts
deleted file mode 100644
index 98ba0142..00000000
--- a/apps/api/src/video-transcript-service.ts
+++ /dev/null
@@ -1,184 +0,0 @@
-import { parseVideoNoteMarker } from "@edgeever/shared";
-import { isDatabaseNotReadyError } from "./auth-state";
-import { isoNow } from "./entity-utils";
-import type { DatabaseAdapter } from "./storage-contract";
-
-const STALE_MS = 20 * 60 * 1000;
-
-export type VideoTranscriptJob = {
- memoId: string;
- platform: "youtube" | "bilibili";
- videoId: string;
- sourceUrl: string;
- durationSeconds: number;
- placeholderText: string;
- transcriptLabel: string;
- contentHash: string;
- status: string;
- claimedAt: string | null;
-};
-
-type VideoTranscriptJobRow = {
- memo_id: string;
- platform: "youtube" | "bilibili";
- video_id: string;
- source_url: string;
- duration_seconds: number;
- placeholder_text: string;
- transcript_label: string;
- content_hash: string;
- status: string;
- claimed_at: string | null;
-};
-
-const JOB_COLUMNS = `memo_id, platform, video_id, source_url, duration_seconds,
- placeholder_text, transcript_label, content_hash, status, claimed_at`;
-
-const mapJob = (row: VideoTranscriptJobRow): VideoTranscriptJob => ({
- memoId: row.memo_id,
- platform: row.platform,
- videoId: row.video_id,
- sourceUrl: row.source_url,
- durationSeconds: row.duration_seconds,
- placeholderText: row.placeholder_text,
- transcriptLabel: row.transcript_label,
- contentHash: row.content_hash,
- status: row.status,
- claimedAt: row.claimed_at,
-});
-
-const changed = (result: { meta?: { changes?: unknown } }) => Number(result.meta?.changes) === 1;
-
-const staleBefore = (now: string) => new Date(Date.parse(now) - STALE_MS).toISOString();
-
-const claimableSql = `(
- status = 'pending'
- OR (status = 'transcribing' AND claimed_at IS NOT NULL AND claimed_at < ?)
-)`;
-
-export type VideoTranscriptExtractResult =
- | { status: "queued" | "running" }
- | { status: "missing" }
- | { status: "unavailable" };
-
-export const extractVideoTranscript = async (
- db: DatabaseAdapter,
- workspaceId: string,
- memoId: string,
- now = isoNow(),
-): Promise<VideoTranscriptExtractResult> => {
- if (!memoId) return { status: "missing" };
- const memo = await db.prepare(
- `SELECT c.content_markdown, c.content_hash
- FROM memos m
- JOIN memo_contents c ON c.memo_id = m.id
- WHERE m.workspace_id = ? AND m.id = ? AND m.is_deleted = 0
- LIMIT 1`,
- ).bind(workspaceId, memoId).first<{ content_markdown: string; content_hash: string }>();
- const identity = parseVideoNoteMarker(memo?.content_markdown);
- if (!memo?.content_hash || !identity) return { status: "missing" };
- try {
- const result = await db.prepare(
- `INSERT INTO video_transcript_jobs (
- memo_id, workspace_id, platform, video_id, source_url, duration_seconds,
- placeholder_text, transcript_label, content_hash, status, created_at, updated_at
- ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 'pending', ?, ?)
- ON CONFLICT(memo_id) DO UPDATE SET
- platform = excluded.platform,
- video_id = excluded.video_id,
- source_url = excluded.source_url,
- duration_seconds = excluded.duration_seconds,
- placeholder_text = excluded.placeholder_text,
- transcript_label = excluded.transcript_label,
- content_hash = excluded.content_hash,
- status = 'pending',
- claimed_at = NULL,
- error_code = NULL,
- updated_at = excluded.updated_at
- WHERE video_transcript_jobs.workspace_id = excluded.workspace_id
- AND (
- video_transcript_jobs.status IN ('ready', 'failed')
- OR (
- video_transcript_jobs.status = 'transcribing'
- AND video_transcript_jobs.claimed_at IS NOT NULL
- AND video_transcript_jobs.claimed_at < ?
- )
- )`,
- ).bind(
- memoId,
- workspaceId,
- identity.platform,
- identity.videoId,
- identity.sourceUrl,
- identity.durationSeconds,
- identity.placeholderText,
- identity.transcriptLabel,
- memo.content_hash,
- now,
- now,
- staleBefore(now),
- ).run();
- return { status: changed(result) ? "queued" : "running" };
- } catch (error) {
- if (isDatabaseNotReadyError(error)) return { status: "unavailable" };
- throw error;
- }
-};
-
-export const listClaimableVideoTranscriptJobs = async (
- db: DatabaseAdapter,
- workspaceId: string,
- now = isoNow(),
-) => {
- const rows = await db.prepare(
- `SELECT ${JOB_COLUMNS}
- FROM video_transcript_jobs
- WHERE workspace_id = ? AND ${claimableSql}
- ORDER BY created_at ASC
- LIMIT 5`,
- ).bind(workspaceId, staleBefore(now)).all<VideoTranscriptJobRow>();
- return (rows.results ?? []).map(mapJob);
-};
-
-export const claimVideoTranscriptJob = async (
- db: DatabaseAdapter,
- workspaceId: string,
- memoId: string,
- now = isoNow(),
-) => {
- const result = await db.prepare(
- `UPDATE video_transcript_jobs
- SET status = 'transcribing', claimed_at = ?, error_code = NULL, updated_at = ?
- WHERE workspace_id = ? AND memo_id = ? AND ${claimableSql}`,
- ).bind(now, now, workspaceId, memoId, staleBefore(now)).run();
- if (!changed(result)) return null;
- const row = await db.prepare(
- `SELECT ${JOB_COLUMNS}
- FROM video_transcript_jobs
- WHERE workspace_id = ? AND memo_id = ? AND status = 'transcribing' AND claimed_at = ?
- LIMIT 1`,
- ).bind(workspaceId, memoId, now).first<VideoTranscriptJobRow>();
- return row ? mapJob(row) : null;
-};
-
-export const finishVideoTranscriptJob = async (
- db: DatabaseAdapter,
- workspaceId: string,
- memoId: string,
- input: { status: "ready" | "failed"; errorCode: string | null; claimedAt: string },
- now = isoNow(),
-) => {
- const result = await db.prepare(
- `UPDATE video_transcript_jobs
- SET status = ?, error_code = ?, updated_at = ?
- WHERE workspace_id = ? AND memo_id = ? AND status = 'transcribing' AND claimed_at = ?`,
- ).bind(
- input.status,
- input.status === "failed" ? input.errorCode : null,
- now,
- workspaceId,
- memoId,
- input.claimedAt,
- ).run();
- return changed(result);
-};
diff --git a/apps/api/src/video-transcript.test.mjs b/apps/api/src/video-transcript.test.mjs
deleted file mode 100644
index 884e45b1..00000000
--- a/apps/api/src/video-transcript.test.mjs
+++ /dev/null
@@ -1,336 +0,0 @@
-import { serializeVideoNoteMarker } from "@edgeever/shared";
-import { afterEach, expect, test } from "bun:test";
-import { Database } from "bun:sqlite";
-import { globSync, readFileSync } from "node:fs";
-import { Hono } from "hono";
-import { mapMemoSummary } from "./memo-list-service.ts";
-import { createMemoRecord } from "./memo-service.ts";
-import { createSelfHostedStorageAdapter } from "./self-hosted-storage-adapter.ts";
-import { registerMemoRoutes } from "./memo-routes.ts";
-import { registerVideoTranscriptRoutes } from "./video-transcript-routes.ts";
-import { extractVideoTranscript } from "./video-transcript-service.ts";
-
-const opened = [];
-afterEach(() => opened.splice(0).forEach((db) => db.close()));
-
-const user = {
- kind: "user",
- actorType: "user",
- actorId: "usr_1",
- username: "member",
- displayName: "Member",
- scopes: [],
- workspaceId: "paw",
- role: "owner",
-};
-
-const jobInput = {
- platform: "youtube",
- videoId: "abcdefghijk",
- sourceUrl: "https://www.youtube.com/watch?v=abcdefghijk",
- durationSeconds: 42,
- placeholderText: "这一集没有可用字幕",
- transcriptLabel: "字幕实录",
-};
-
-const setup = () => {
- const sqlite = new Database(":memory:");
- opened.push(sqlite);
- for (const file of globSync("migrations/*.sql").sort()) sqlite.exec(readFileSync(file, "utf8"));
- sqlite.exec(`INSERT INTO workspaces(id, name) VALUES ('paw', 'Paw');
- INSERT INTO notebooks(id, workspace_id, name) VALUES ('inbox', 'paw', 'Inbox');`);
- const db = createSelfHostedStorageAdapter(sqlite, "/tmp/video-transcript-unused").db;
- return { sqlite, db };
-};
-
-const createApp = (auth = user, { demoMode = false } = {}) => {
- const app = new Hono();
- app.use("/api/v1/*", async (context, next) => {
- context.set("auth", auth);
- await next();
- });
- const unused = async () => {
- throw new Error("unused");
- };
- registerVideoTranscriptRoutes(app, { isDemoMode: () => demoMode });
- registerMemoRoutes(app, {
- clampNumber: (value) => value,
- createImageResource: async () => ({ id: "res_1" }),
- createMemo: (...args) => createMemoRecord(...args),
- createMemoEditSession: unused,
- deleteMemo: async () => {},
- deleteMemos: async () => 0,
- emptyTrash: async () => 0,
- getMemoDetail: async () => null,
- listMemos: async () => ({ memos: [] }),
- listMemoRevisions: async () => [],
- mergeMemos: unused,
- moveMemos: async () => 0,
- restoreMemo: unused,
- restoreMemoRevision: unused,
- updateMemo: unused,
- });
- return app;
-};
-
-const environmentFor = (db) => ({
- storage: { db, resources: {} },
-});
-
-const postMemo = (app, db, body) => app.request("/api/v1/memos", {
- method: "POST",
- headers: { "content-type": "application/json" },
- body: JSON.stringify({ notebookId: "inbox", title: "Video", contentMarkdown: "hello\n", ...body }),
-}, environmentFor(db));
-
-const videoMarkdown = () => `这一集没有可用字幕\n\n${serializeVideoNoteMarker(jobInput)}\n`;
-
-const extract = (app, db, memoId) => app.request(
- `/api/v1/video-transcript-jobs/${memoId}/extract`,
- { method: "POST" },
- environmentFor(db),
-);
-
-test("a missing transcript table does not pretend extraction started", async () => {
- const db = {
- prepare(sql) {
- return {
- bind() { return this; },
- async first() {
- return String(sql).includes("FROM memos")
- ? { content_markdown: videoMarkdown(), content_hash: "a".repeat(64) }
- : null;
- },
- async run() { throw new Error("no such table: video_transcript_jobs"); },
- };
- },
- };
- await expect(extractVideoTranscript(db, "paw", "memo_1")).resolves.toEqual({ status: "unavailable" });
-});
-
-test("saving a video note waits for an explicit extract", async () => {
- const { sqlite, db } = setup();
- const app = createApp();
- const created = await postMemo(app, db, { videoTranscript: jobInput, contentMarkdown: videoMarkdown() });
- expect(created.status).toBe(201);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(0);
- const memoId = (await created.json()).memo.id;
- expect(mapMemoSummary({
- id: memoId,
- notebook_id: "inbox",
- title: "Video",
- excerpt: "",
- content_markdown: videoMarkdown(),
- tags_json: "[]",
- is_pinned: 0,
- is_archived: 0,
- is_deleted: 0,
- created_at: "2026-10-07T00:00:00.000Z",
- updated_at: "2026-10-07T00:00:00.000Z",
- deleted_at: null,
- revision: 1,
- }).videoNote).toBe(true);
-
- const started = await extract(app, db, memoId);
- expect(started.status).toBe(200);
- expect(await started.json()).toEqual({ status: "queued" });
- const row = sqlite.query(
- "SELECT platform, video_id, source_url, duration_seconds, status, content_hash FROM video_transcript_jobs",
- ).get();
- expect(row).toMatchObject({
- platform: "youtube",
- video_id: "abcdefghijk",
- source_url: jobInput.sourceUrl,
- duration_seconds: 42,
- status: "pending",
- });
- expect(row.content_hash).toHaveLength(64);
-
- const again = await extract(app, db, memoId);
- expect(await again.json()).toEqual({ status: "running" });
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(1);
-
- const plain = await postMemo(app, db, {});
- expect(plain.status).toBe(201);
- expect((await extract(app, db, (await plain.json()).memo.id)).status).toBe(404);
-
- const rejected = await postMemo(app, db, {
- videoTranscript: { ...jobInput, sourceUrl: "https://user:[email protected]/watch?v=abcdefghijk" },
- });
- expect(rejected.status).toBe(201);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(1);
-});
-
-test("a cover upload keeps the video note without starting transcription", async () => {
- const { sqlite, db } = setup();
- const app = createApp();
- const form = new FormData();
- form.append("notebookId", "inbox");
- form.append("title", "Video");
- form.append("contentMarkdown", `![封面](/api/v1/resources/EDGEVERRESOURCEID/blob)\n\n${videoMarkdown()}`);
- form.append("tags", JSON.stringify(["web-clip"]));
- form.append("videoTranscript", JSON.stringify(jobInput));
- form.append("file", new File([Uint8Array.from([1, 2, 3])], "cover.jpg", { type: "image/jpeg" }));
- const created = await app.request("/api/v1/memos/with-image", { method: "POST", body: form }, environmentFor(db));
- expect(created.status).toBe(201);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(0);
- const memoId = (await created.json()).memo.id;
- expect((await extract(app, db, memoId)).status).toBe(200);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(1);
- const markdown = sqlite.query("SELECT content_markdown FROM memo_contents").get().content_markdown;
- expect(markdown).not.toContain("EDGEVERRESOURCEID");
-});
-
-test("one desktop claims the job, a second waits, and a stale claim can be taken", async () => {
- const { sqlite, db } = setup();
- const app = createApp();
- const created = await postMemo(app, db, { contentMarkdown: videoMarkdown() });
- expect(created.status).toBe(201);
- const memoId = (await created.json()).memo.id;
- expect((await extract(app, db, memoId)).status).toBe(200);
- const environment = environmentFor(db);
- const claim = () => app.request(`/api/v1/video-transcript-jobs/${memoId}/claim`, { method: "POST" }, environment);
- const first = await claim();
- expect(first.status).toBe(200);
- const claimed = await first.json();
- expect(claimed.job.memoId).toBe(memoId);
- expect(claimed.job.claimedAt).toEqual(expect.any(String));
- expect(JSON.stringify(claimed)).not.toContain("apiKey");
-
- const second = await claim();
- expect(second.status).toBe(409);
- expect(sqlite.query("SELECT claimed_at FROM video_transcript_jobs").get().claimed_at).toBe(claimed.job.claimedAt);
-
- const listed = await app.request("/api/v1/video-transcript-jobs", {}, environment);
- expect(await listed.json()).toEqual({ jobs: [] });
-
- sqlite.query("UPDATE video_transcript_jobs SET claimed_at = ?").run("2020-01-01T00:00:00.000Z");
- const reclaimed = await claim();
- expect(reclaimed.status).toBe(200);
- const next = await reclaimed.json();
- expect(next.job.claimedAt).not.toBe("2020-01-01T00:00:00.000Z");
-
- const finished = await app.request(`/api/v1/video-transcript-jobs/${memoId}/finish`, {
- method: "POST",
- headers: { "content-type": "application/json" },
- body: JSON.stringify({ status: "ready", claimedAt: next.job.claimedAt }),
- }, environment);
- expect(finished.status).toBe(200);
- expect(sqlite.query("SELECT status, error_code FROM video_transcript_jobs").get()).toEqual({
- status: "ready",
- error_code: null,
- });
-
- const late = await app.request(`/api/v1/video-transcript-jobs/${memoId}/finish`, {
- method: "POST",
- headers: { "content-type": "application/json" },
- body: JSON.stringify({ status: "failed", errorCode: "download_failed", claimedAt: next.job.claimedAt }),
- }, environment);
- expect(late.status).toBe(409);
-});
-
-test("a failed job stays failed and is not offered again", async () => {
- const { sqlite, db } = setup();
- const app = createApp();
- const environment = environmentFor(db);
- const created = await postMemo(app, db, { contentMarkdown: videoMarkdown() });
- expect(created.status).toBe(201);
- const memoId = (await created.json()).memo.id;
- expect((await extract(app, db, memoId)).status).toBe(200);
- const claimed = await (await app.request(
- `/api/v1/video-transcript-jobs/${memoId}/claim`,
- { method: "POST" },
- environment,
- )).json();
- const failed = await app.request(`/api/v1/video-transcript-jobs/${memoId}/finish`, {
- method: "POST",
- headers: { "content-type": "application/json" },
- body: JSON.stringify({
- status: "failed",
- errorCode: "download_failed",
- claimedAt: claimed.job.claimedAt,
- }),
- }, environment);
- expect(failed.status).toBe(200);
- expect(await (await app.request("/api/v1/video-transcript-jobs", {}, environment)).json()).toEqual({ jobs: [] });
- expect((await app.request(`/api/v1/video-transcript-jobs/${memoId}/claim`, { method: "POST" }, environment)).status).toBe(409);
-});
-
-test("a failed extract can be started again, and a fresh claim cannot", async () => {
- const { sqlite, db } = setup();
- const app = createApp();
- const environment = environmentFor(db);
- const created = await postMemo(app, db, { contentMarkdown: videoMarkdown() });
- const memoId = (await created.json()).memo.id;
- expect(await (await extract(app, db, memoId)).json()).toEqual({ status: "queued" });
- const claimed = await (await app.request(
- `/api/v1/video-transcript-jobs/${memoId}/claim`,
- { method: "POST" },
- environment,
- )).json();
- expect((await app.request(`/api/v1/video-transcript-jobs/${memoId}/finish`, {
- method: "POST",
- headers: { "content-type": "application/json" },
- body: JSON.stringify({
- status: "failed",
- errorCode: "download_failed",
- claimedAt: claimed.job.claimedAt,
- }),
- }, environment)).status).toBe(200);
- expect(await (await extract(app, db, memoId)).json()).toEqual({ status: "queued" });
- expect(sqlite.query("SELECT status, error_code, claimed_at FROM video_transcript_jobs").get()).toEqual({
- status: "pending",
- error_code: null,
- claimed_at: null,
- });
-
- expect((await app.request(`/api/v1/video-transcript-jobs/${memoId}/claim`, { method: "POST" }, environment)).status).toBe(200);
- expect(await (await extract(app, db, memoId)).json()).toEqual({ status: "running" });
- expect(sqlite.query("SELECT status FROM video_transcript_jobs").get().status).toBe("transcribing");
- sqlite.query("UPDATE video_transcript_jobs SET claimed_at = ?").run("2020-01-01T00:00:00.000Z");
- expect(await (await extract(app, db, memoId)).json()).toEqual({ status: "queued" });
- expect(sqlite.query("SELECT status, claimed_at FROM video_transcript_jobs").get()).toEqual({
- status: "pending",
- claimed_at: null,
- });
-});
-
-test("clipper tokens cannot start, list, or claim transcription", async () => {
- const { sqlite, db } = setup();
- const clipper = createApp({
- ...user,
- kind: "agent",
- actorType: "agent",
- actorId: "tok_clip",
- scopes: ["write:memos", "read:memos"],
- });
- const created = await postMemo(clipper, db, { contentMarkdown: videoMarkdown(), videoTranscript: jobInput });
- expect(created.status).toBe(201);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(0);
- const environment = environmentFor(db);
- const memoId = (await created.json()).memo.id;
- expect((await extract(clipper, db, memoId)).status).toBe(403);
- expect((await clipper.request("/api/v1/video-transcript-jobs", {}, environment)).status).toBe(403);
- expect((await clipper.request(`/api/v1/video-transcript-jobs/${memoId}/claim`, { method: "POST" }, environment)).status).toBe(403);
- expect(sqlite.query("SELECT COUNT(*) AS count FROM video_transcript_jobs").get().count).toBe(0);
-
- const demo = createApp(user, { demoMode: true });
- expect((await extract(demo, db, memoId)).status).toBe(403);
- expect((await demo.request("/api/v1/video-transcript-jobs", {}, environment)).status).toBe(403);
-});
-
-test("the API does not import the desktop transcription package", () => {
- for (const name of [
- "video-transcript-service.ts",
- "video-transcript-routes.ts",
- "ai-routes.ts",
- "memo-routes.ts",
- ]) {
- const source = readFileSync(new URL(`./${name}`, import.meta.url), "utf8");
- expect(source).not.toContain("@ai-sdk/openai");
- expect(source).not.toContain("transcribe(");
- }
- const service = readFileSync(new URL("./ai-service.ts", import.meta.url), "utf8");
- expect(service).not.toContain("@ai-sdk/openai");
- expect(service).not.toContain(".transcription(");
-});
diff --git a/apps/desktop/package.json b/apps/desktop/package.json
index 2edc3b88..b1b9a9e9 100644
--- a/apps/desktop/package.json
+++ b/apps/desktop/package.json
@@ -18,10 +18,8 @@
"electron-builder": "^26.0.12"
},
"dependencies": {
- "@ai-sdk/openai": "4.0.36",
"@agentclientprotocol/sdk": "^1.5.1",
"@edgeever/public-network": "workspace:*",
- "ai": "7.0.58",
"croner": "^10.0.1",
"electron-updater": "^6.6.2",
"fflate": "0.8.3",
diff --git a/apps/desktop/src/main/index.mjs b/apps/desktop/src/main/index.mjs
index f901b447..7b9d7e9b 100644
--- a/apps/desktop/src/main/index.mjs
+++ b/apps/desktop/src/main/index.mjs
@@ -48,13 +48,6 @@ import {
} from "./windows-update-trust.mjs";
import electronUpdater from "electron-updater";
import { createPluginPublicNetworkRuntime } from "./plugin-public-network.mjs";
-import { createYtDlpManager, startYtDlpMaintenance } from "./yt-dlp-tool.mjs";
-import {
- createSpeechTranscriptionRunner,
- readCookieBrowserPreference,
- startSpeechTranscription,
- writeCookieBrowserPreference,
-} from "./speech-transcription.mjs";
import { createAiDirectRuntime } from "./ai-direct.mjs";
import { createAcpHostRuntime, registerAcpIpc } from "./acp-host.mjs";
import { shouldQuitAfterAllWindowsClosed } from "./window-lifecycle.mjs";
@@ -134,10 +127,6 @@ let updateState = "idle";
let updateCheckInFlight = null;
let updateDownloadInFlight = null;
let updateCheckTimer = null;
-let stopYtDlpMaintenance = null;
-let stopSpeechTranscription = null;
-let ytDlpMaintenanceTimer = null;
-let ytDlpManager = null;
let lastUpdateCheckAt = 0;
let downloadedUpdateVersion = null;
let promptedUpdateVersion = null;
@@ -185,7 +174,6 @@ const sessionTokenPath = () => join(app.getPath("userData"), "session-token");
const crashMarkerPath = () => join(app.getPath("userData"), "last-session-active");
const installationMarkerPath = () => join(app.getPath("userData"), "installation-confirmed");
const logPath = () => join(app.getPath("userData"), "logs", "desktop.log");
-const videoCookieBrowserPath = () => join(app.getPath("userData"), "video-cookie-browser.json");
let desktopSessionToken = "";
const sidecarDataDirectory = (accountId = null) => {
return accountId
@@ -1530,15 +1518,6 @@ const startApplication = async () => {
app.quit();
return;
}
- ytDlpManager = createYtDlpManager({
- directory: join(app.getPath("userData"), "tools"),
- platform: process.platform,
- arch: process.arch,
- fetch: net.fetch.bind(net),
- onDiagnostic: (event, details) => {
- void writeDiagnostic(event, details);
- },
- });
await loadConfiguredApiBaseUrl();
await loadDesktopSessionToken();
app.setAsDefaultProtocolClient("edgeever");
@@ -1572,17 +1551,6 @@ const startApplication = async () => {
});
ipcMain.handle("desktop:sidecar-status", () => ({ available: Boolean(sidecar), path: sidecarPath, scope: sidecarScopeKey }));
ipcMain.handle("desktop:system-info", () => desktopRuntimeSystemInfo());
- ipcMain.handle("desktop:yt-dlp-status", () => ytDlpManager?.status() ?? {
- state: "idle",
- version: null,
- path: "",
- errorCode: null,
- httpStatus: null,
- });
- ipcMain.handle("desktop:video-cookie-browser-get", () => readCookieBrowserPreference(videoCookieBrowserPath()));
- ipcMain.handle("desktop:video-cookie-browser-set", (_event, browser) => (
- writeCookieBrowserPreference(videoCookieBrowserPath(), browser)
- ));
ipcMain.handle("desktop:set-account-scope", async (_event, accountId) => {
const normalizedAccountId = typeof accountId === "string" && accountId.trim() ? accountId.trim() : null;
const nextScopeKey = accountScopeKey(configuredApiBaseUrl, normalizedAccountId);
@@ -1947,22 +1915,6 @@ const startApplication = async () => {
// user-visible critical path so the first installed launch opens promptly.
await ejectMountedMacInstallers();
await confirmMacInstallation();
- // The standalone binary is about 35 MB. Start only after the window and the
- // first-run installer dialog, and never await it: the workspace must stay usable.
- ytDlpMaintenanceTimer = setTimeout(() => {
- ytDlpMaintenanceTimer = null;
- if (!ytDlpManager || stopYtDlpMaintenance) return;
- stopYtDlpMaintenance = startYtDlpMaintenance(ytDlpManager, { intervalMs: updateCheckIntervalMs });
- // Transcription waits on this same timer and is not awaited. A pass that
- // finds nothing, or a download that fails, must not block the window.
- stopSpeechTranscription = startSpeechTranscription(createSpeechTranscriptionRunner({
- getSession: () => ({ baseUrl: configuredApiBaseUrl, token: desktopSessionToken }),
- ytDlpStatus: () => ytDlpManager.status(),
- cookieBrowser: () => readCookieBrowserPreference(videoCookieBrowserPath()),
- audioDirectory: () => join(app.getPath("userData"), "tools", "transcript-audio"),
- }));
- }, 3_000);
- ytDlpMaintenanceTimer.unref?.();
void enableMacShareExtension({
platform: process.platform,
packaged: app.isPackaged,
@@ -2052,14 +2004,6 @@ app.on("before-quit", (event) => {
clearInterval(updateCheckTimer);
updateCheckTimer = null;
}
- if (ytDlpMaintenanceTimer) {
- clearTimeout(ytDlpMaintenanceTimer);
- ytDlpMaintenanceTimer = null;
- }
- stopYtDlpMaintenance?.();
- stopYtDlpMaintenance = null;
- stopSpeechTranscription?.();
- stopSpeechTranscription = null;
tray?.destroy();
void (async () => {
await stopSidecar();
diff --git a/apps/desktop/src/main/speech-transcription.mjs b/apps/desktop/src/main/speech-transcription.mjs
deleted file mode 100644
index 1bc74f55..00000000
--- a/apps/desktop/src/main/speech-transcription.mjs
+++ /dev/null
@@ -1,412 +0,0 @@
-import { execFile } from "node:child_process";
-import { mkdir, readdir, readFile, stat, unlink, writeFile } from "node:fs/promises";
-import { dirname, join } from "node:path";
-
-export const MAX_AUDIO_BYTES = 24 * 1024 * 1024;
-export const MAX_DURATION_SECONDS = 30 * 60;
-export const SPEECH_TRANSCRIPTION_INTERVAL_MS = 20_000;
-const DOWNLOAD_TIMEOUT_MS = 180_000;
-const TRANSCRIBE_TIMEOUT_MS = 180_000;
-
-const PUBLIC_VIDEO_HOSTS = new Set([
- "youtube.com",
- "www.youtube.com",
- "m.youtube.com",
- "youtu.be",
- "bilibili.com",
- "www.bilibili.com",
- "m.bilibili.com",
-]);
-
-const FAILURE_CODES = new Set([
- "duration_limit",
- "unsupported_source",
- "audio_too_large",
- "download_failed",
- "transcribe_failed",
- "empty_transcript",
- "note_changed",
- "note_write_failed",
- "credentials_unavailable",
- "disabled",
-]);
-
-const fail = (code) => {
- const error = new Error(code);
- error.code = code;
- return error;
-};
-
-export const isPublicVideoSourceUrl = (value) => {
- try {
- const url = new URL(value);
- return url.protocol === "https:"
- && !url.username
- && !url.password
- && PUBLIC_VIDEO_HOSTS.has(url.hostname);
- } catch {
- return false;
- }
-};
-
-export const transcriptAudioBaseName = (memoId) => {
- const value = String(memoId ?? "");
- if (!value || !/^[A-Za-z0-9_-]+$/.test(value)) return null;
- return value;
-};
-
-// Keep this aligned with packages/shared/src/video-cookie-browser.ts.
-export const COOKIE_BROWSERS = Object.freeze([
- "chrome",
- "chromium",
- "edge",
- "firefox",
- "safari",
- "brave",
- "opera",
- "vivaldi",
- "whale",
-]);
-export const DEFAULT_COOKIE_BROWSER = "chrome";
-
-export const normalizeCookieBrowser = (value) => {
- const name = String(value ?? "").trim().toLowerCase();
- if (name === "none") return "none";
- return COOKIE_BROWSERS.includes(name) ? name : DEFAULT_COOKIE_BROWSER;
-};
-
-export const readCookieBrowserPreference = async (filePath) => {
- try {
- const parsed = JSON.parse(await readFile(filePath, "utf8"));
- return normalizeCookieBrowser(parsed?.browser);
- } catch {
- return DEFAULT_COOKIE_BROWSER;
- }
-};
-
-export const writeCookieBrowserPreference = async (filePath, browser) => {
- const name = String(browser ?? "").trim().toLowerCase();
- if (name !== "none" && !COOKIE_BROWSERS.includes(name)) {
- throw Object.assign(new Error("unsupported_browser"), { code: "unsupported_browser" });
- }
- await mkdir(dirname(filePath), { recursive: true });
- await writeFile(filePath, `${JSON.stringify({ browser: name })}\n`, { mode: 0o600 });
- return { browser: name };
-};
-
-export const bestAudioArguments = (sourceUrl, outputTemplate, browser = DEFAULT_COOKIE_BROWSER) => {
- const selected = normalizeCookieBrowser(browser);
- return [
- "--no-playlist",
- "--no-progress",
- "--no-mtime",
- "--no-cache-dir",
- "--restrict-filenames",
- "-f",
- "bestaudio",
- "--max-filesize",
- "24M",
- ...(selected === "none" ? [] : ["--cookies-from-browser", selected]),
- "-o",
- outputTemplate,
- "--",
- sourceUrl,
- ];
-};
-
-export const formatTranscriptTimestamp = (seconds) => {
- const total = Math.max(0, Math.floor(Number(seconds) || 0));
- const hours = Math.floor(total / 3600);
- const minutes = Math.floor((total % 3600) / 60);
- const remain = total % 60;
- const clock = `${String(minutes).padStart(2, "0")}:${String(remain).padStart(2, "0")}`;
- return hours > 0 ? `${hours}:${clock}` : clock;
-};
-
-const escapeHtml = (value) => String(value)
- .replace(/&/g, "&amp;")
- .replace(/</g, "&lt;")
- .replace(/>/g, "&gt;");
-
-const plain = (value) => escapeHtml(String(value).replace(/\s+/g, " ").trim())
- .replace(/\]\(/g, "] (")
- .replace(/\[/g, "[")
- .replace(/\]/g, "]");
-
-export const videoTimestampUrl = (sourceUrl, platform, seconds) => {
- const time = Math.max(0, Math.floor(Number(seconds) || 0));
- if (platform === "youtube") {
- return sourceUrl.includes("/shorts/")
- ? `${sourceUrl}?t=${time}`
- : `${sourceUrl}&t=${time}s`;
- }
- return sourceUrl.includes("?")
- ? `${sourceUrl}&t=${time}`
- : `${sourceUrl}?t=${time}`;
-};
-
-export const cuesFromTranscription = (result) => {
- const segments = Array.isArray(result?.segments) ? result.segments : [];
- return segments.flatMap((segment) => {
- const text = String(segment?.text ?? "").replace(/\s+/g, " ").trim();
- const start = Number(segment?.startSecond);
- const end = Number(segment?.endSecond);
- if (!text || !Number.isFinite(start) || !Number.isFinite(end)) return [];
- return [{ start: Math.max(0, start), end: Math.max(start, end), text }];
- });
-};
-
-export const buildTranscriptMarkdown = ({ label, text, cues, sourceUrl, platform }) => {
- const summary = plain(label ?? "");
- if (!summary) return null;
- const lines = (cues ?? []).map((cue) => {
- const body = plain(cue.text);
- if (!body) return "";
- return `- [${formatTranscriptTimestamp(cue.start)}](${videoTimestampUrl(sourceUrl, platform, cue.start)}) ${body}`;
- }).filter(Boolean);
- const inner = lines.length ? lines.join("\n") : plain(text ?? "");
- if (!inner) return null;
- return ["<details>", `<summary>${summary}</summary>`, "", inner, "", "</details>"].join("\n");
-};
-
-const VIDEO_NOTE_MARKER = /<!--\s*edgeever-video-v1:[A-Za-z0-9_=-]+\s*-->/g;
-
-export const applyTranscript = (markdown, placeholder, block) => {
- const source = String(markdown ?? "");
- const markers = source.match(VIDEO_NOTE_MARKER) ?? [];
- const body = markers.length ? source.replace(VIDEO_NOTE_MARKER, "") : source;
- const needle = String(placeholder ?? "").trim();
- const addition = String(block ?? "").trim();
- if (!addition) return source;
- let next = body;
- if (needle) {
- const lines = body.split("\n");
- const index = lines.findIndex((line) => line.trim() === needle);
- if (index >= 0) {
- lines.splice(index, 1, addition);
- next = lines.join("\n");
- } else {
- next = `${body.trim()}\n\n${addition}`;
- }
- } else {
- next = `${body.trim()}\n\n${addition}`;
- }
- const cleaned = `${next.replace(/\n{3,}/g, "\n\n").trim()}\n`;
- if (!markers.length) return cleaned;
- return `${cleaned.trim()}\n\n${markers.join("\n")}\n`;
-};
-
-export const transcriptionRequest = ({ baseUrl, apiKey, modelId, audio, abortSignal }) => ({
- provider: { baseURL: baseUrl, apiKey },
- modelId,
- audio,
- maxRetries: 0,
- abortSignal,
- providerOptions: { openai: { timestampGranularities: ["segment"] } },
-});
-
-export const transcribeLocalAudio = async (input, deps = {}) => {
- const request = transcriptionRequest(input);
- const createProvider = deps.createProvider ?? (await import("@ai-sdk/openai")).createOpenAI;
- const transcribe = deps.transcribe ?? (await import("ai")).transcribe;
- const provider = createProvider(request.provider);
- return transcribe({
- model: provider.transcription(request.modelId),
- audio: request.audio,
- maxRetries: request.maxRetries,
- abortSignal: request.abortSignal,
- providerOptions: request.providerOptions,
- });
-};
-
-const instanceUrl = (baseUrl, path) => `${String(baseUrl).replace(/\/+$/, "")}${path}`;
-
-const requestJson = async (fetchImpl, session, path, init = {}) => {
- const headers = {
- Accept: "application/json",
- Authorization: `Bearer ${session.token}`,
- };
- if (init.body) headers["content-type"] = "application/json";
- const response = await fetchImpl(instanceUrl(session.baseUrl, path), {
- method: init.method ?? "GET",
- headers,
- body: init.body,
- });
- const payload = await response.json().catch(() => null);
- return { ok: response.ok, status: response.status, payload };
-};
-
-const removeAudioFiles = async (directory, baseName) => {
- if (!directory || !baseName) return;
- const names = await readdir(directory).catch(() => []);
- await Promise.all(names
- .filter((name) => name === baseName || name.startsWith(`${baseName}.`))
- .map((name) => unlink(join(directory, name)).catch(() => undefined)));
-};
-
-const downloadBestAudio = async ({ execFileImpl, binaryPath, sourceUrl, directory, baseName, timeoutMs, browser }) => {
- await mkdir(directory, { recursive: true });
- const template = join(directory, `${baseName}.%(ext)s`);
- const selected = typeof browser === "function" ? await browser() : browser;
- await new Promise((resolve, reject) => {
- execFileImpl(binaryPath, bestAudioArguments(sourceUrl, template, selected), {
- timeout: timeoutMs,
- windowsHide: true,
- maxBuffer: 4 * 1024 * 1024,
- }, (error) => {
- if (error) reject(fail("download_failed"));
- else resolve();
- });
- });
- const names = await readdir(directory);
- const audioName = names.find((name) => name.startsWith(`${baseName}.`) && !name.endsWith(".part") && !name.endsWith(".ytdl"));
- if (!audioName) throw fail("download_failed");
- return join(directory, audioName);
-};
-
-const failureCode = (error) => FAILURE_CODES.has(error?.code) ? error.code : "transcribe_failed";
-
-const processClaimedJob = async (deps, session, job) => {
- const maxAudioBytes = deps.maxAudioBytes ?? MAX_AUDIO_BYTES;
- const maxDurationSeconds = deps.maxDurationSeconds ?? MAX_DURATION_SECONDS;
- const memoId = job.memoId;
- const baseName = transcriptAudioBaseName(memoId);
- let directory = null;
- let settled = false;
- const finish = (status, errorCode = null) => requestJson(
- deps.fetchImpl,
- session,
- `/api/v1/video-transcript-jobs/${encodeURIComponent(memoId)}/finish`,
- {
- method: "POST",
- body: JSON.stringify({
- status,
- errorCode,
- claimedAt: job.claimedAt,
- }),
- },
- );
- try {
- directory = await deps.audioDirectory();
- const memoResponse = await requestJson(deps.fetchImpl, session, `/api/v1/memos/${encodeURIComponent(memoId)}`);
- const memo = memoResponse.payload?.memo;
- if (!memoResponse.ok || !memo) throw fail("note_write_failed");
- if (memo.contentHash !== job.contentHash) throw fail("note_changed");
- if (Number(job.durationSeconds) > maxDurationSeconds) throw fail("duration_limit");
- if (!isPublicVideoSourceUrl(job.sourceUrl) || !baseName) throw fail("unsupported_source");
- const tool = deps.ytDlpStatus();
- if (tool?.state !== "ready" || !tool.path) throw fail("download_failed");
- const prepared = await requestJson(deps.fetchImpl, session, "/api/v1/ai/transcription-settings/prepare", {
- method: "POST",
- body: "{}",
- });
- const standard = prepared.payload?.provider ?? "openai-compatible";
- if (!prepared.ok || !prepared.payload?.enabled || !prepared.payload.apiKey || !prepared.payload.baseUrl || !prepared.payload.modelId || standard !== "openai-compatible") {
- throw fail(prepared.ok ? "disabled" : "credentials_unavailable");
- }
- const audioPath = await downloadBestAudio({
- execFileImpl: deps.execFileImpl,
- binaryPath: tool.path,
- sourceUrl: job.sourceUrl,
- directory,
- baseName,
- timeoutMs: deps.downloadTimeoutMs ?? DOWNLOAD_TIMEOUT_MS,
- browser: deps.cookieBrowser,
- });
- const audioStat = await stat(audioPath);
- if (audioStat.size > maxAudioBytes) throw fail("audio_too_large");
- if (audioStat.size <= 0) throw fail("download_failed");
- const audio = await readFile(audioPath);
- const result = await transcribeLocalAudio({
- baseUrl: prepared.payload.baseUrl,
- apiKey: prepared.payload.apiKey,
- modelId: prepared.payload.modelId,
- audio,
- abortSignal: AbortSignal.timeout(deps.transcribeTimeoutMs ?? TRANSCRIBE_TIMEOUT_MS),
- }, deps);
- const cues = cuesFromTranscription(result);
- const block = buildTranscriptMarkdown({
- label: job.transcriptLabel,
- text: result?.text,
- cues,
- sourceUrl: job.sourceUrl,
- platform: job.platform,
- });
- if (!block) throw fail("empty_transcript");
- const saved = await requestJson(deps.fetchImpl, session, `/api/v1/memos/${encodeURIComponent(memoId)}`, {
- method: "PATCH",
- body: JSON.stringify({
- contentMarkdown: applyTranscript(memo.contentMarkdown, job.placeholderText, block),
- expectedRevision: memo.revision,
- expectedContentHash: memo.contentHash,
- }),
- });
- if (saved.status === 409) throw fail("note_changed");
- if (!saved.ok) throw fail("note_write_failed");
- const done = await finish("ready");
- if (!done.ok) throw fail("note_write_failed");
- settled = true;
- return { outcome: "ready", memoId };
- } catch (error) {
- if (!settled) await finish("failed", failureCode(error)).catch(() => undefined);
- return { outcome: "failed", reason: failureCode(error), memoId };
- } finally {
- await removeAudioFiles(directory, baseName);
- }
-};
-
-export const runSpeechTranscriptionPass = async (deps) => {
- const fetchImpl = deps.fetchImpl ?? fetch;
- const session = deps.getSession() ?? {};
- if (!session.baseUrl || !session.token) return { outcome: "skipped", reason: "signed_out" };
- const bound = { ...deps, fetchImpl, execFileImpl: deps.execFileImpl ?? execFile };
- try {
- const settings = await requestJson(fetchImpl, session, "/api/v1/ai/transcription-settings");
- if (!settings.ok || settings.payload?.enabled !== true) return { outcome: "skipped", reason: "disabled" };
- const tool = bound.ytDlpStatus();
- if (tool?.state !== "ready" || !tool.path) return { outcome: "skipped", reason: "yt_dlp" };
- const listed = await requestJson(fetchImpl, session, "/api/v1/video-transcript-jobs");
- if (!listed.ok) return { outcome: "skipped", reason: "list_failed" };
- const waiting = listed.payload?.jobs?.[0];
- if (!waiting?.memoId) return { outcome: "skipped", reason: "idle" };
- const claimed = await requestJson(
- fetchImpl,
- session,
- `/api/v1/video-transcript-jobs/${encodeURIComponent(waiting.memoId)}/claim`,
- { method: "POST", body: "{}" },
- );
- if (!claimed.ok || !claimed.payload?.job?.claimedAt) return { outcome: "skipped", reason: "claimed" };
- return processClaimedJob(bound, session, claimed.payload.job);
- } catch {
- return { outcome: "failed", reason: "transcribe_failed" };
- }
-};
-
-export const createSpeechTranscriptionRunner = (options) => ({
- pass: () => runSpeechTranscriptionPass(options),
-});
-
-export const startSpeechTranscription = (
- runner,
- { intervalMs = SPEECH_TRANSCRIPTION_INTERVAL_MS, schedule = setInterval, cancel = clearInterval } = {},
-) => {
- let stopped = false;
- let running = false;
- const tick = () => {
- if (stopped || running) return;
- running = true;
- Promise.resolve()
- .then(() => runner.pass())
- .catch(() => undefined)
- .finally(() => {
- running = false;
- });
- };
- const timer = schedule(tick, intervalMs);
- timer?.unref?.();
- tick();
- return () => {
- stopped = true;
- cancel(timer);
- };
-};
diff --git a/apps/desktop/src/main/speech-transcription.test.mjs b/apps/desktop/src/main/speech-transcription.test.mjs
deleted file mode 100644
index e8ac22ae..00000000
--- a/apps/desktop/src/main/speech-transcription.test.mjs
+++ /dev/null
@@ -1,454 +0,0 @@
-import { expect, test } from "bun:test";
-import { mkdir, readFile, rm, writeFile } from "node:fs/promises";
-import { join } from "node:path";
-import { tmpdir } from "node:os";
-import { VIDEO_COOKIE_BROWSERS } from "../../../../packages/shared/src/video-cookie-browser.ts";
-import {
- applyTranscript,
- bestAudioArguments,
- buildTranscriptMarkdown,
- COOKIE_BROWSERS,
- DEFAULT_COOKIE_BROWSER,
- readCookieBrowserPreference,
- writeCookieBrowserPreference,
- createSpeechTranscriptionRunner,
- isPublicVideoSourceUrl,
- runSpeechTranscriptionPass,
- startSpeechTranscription,
- transcriptionRequest,
-} from "./speech-transcription.mjs";
-
-const session = { baseUrl: "http://127.0.0.1:8787", token: "desktop-session" };
-const apiKey = "asr-test-key";
-const sourceUrl = "https://www.youtube.com/watch?v=abcdefghijk";
-const placeholder = "这一集没有可用字幕";
-
-const waitingJob = {
- memoId: "memo_abc",
- platform: "youtube",
- videoId: "abcdefghijk",
- sourceUrl,
- durationSeconds: 12,
- placeholderText: placeholder,
- transcriptLabel: "字幕实录",
- contentHash: "hash-1",
- status: "pending",
- claimedAt: null,
-};
-
-const memo = {
- id: "memo_abc",
- contentMarkdown: `# Title\n\n${placeholder}\n`,
- contentHash: "hash-1",
- revision: 0,
-};
-
-const jsonResponse = (body, status = 200) => new Response(JSON.stringify(body), {
- status,
- headers: { "content-type": "application/json" },
-});
-
-const installFetch = (state) => async (url, init = {}) => {
- const path = new URL(url).pathname;
- const body = init.body ? JSON.parse(init.body) : null;
- state.calls.push({
- path,
- method: init.method ?? "GET",
- body,
- authorization: init.headers?.Authorization ?? init.headers?.authorization ?? "",
- });
- if (path === "/api/v1/ai/transcription-settings" && (init.method ?? "GET") === "GET") {
- return jsonResponse({ enabled: state.enabled });
- }
- if (path === "/api/v1/video-transcript-jobs" && (init.method ?? "GET") === "GET") {
- return jsonResponse({ jobs: state.jobs });
- }
- if (path.endsWith("/claim")) {
- if (!state.jobs.length) return jsonResponse({ error: { code: "video_transcript_not_claimed" } }, 409);
- const job = { ...state.jobs[0], status: "transcribing", claimedAt: "2026-10-06T00:00:00.000Z" };
- state.jobs = [];
- return jsonResponse({ job });
- }
- if (path.endsWith("/finish")) {
- state.finished = body;
- return jsonResponse({ ok: true });
- }
- if (path === "/api/v1/ai/transcription-settings/prepare") {
- state.prepared = true;
- return jsonResponse({
- enabled: true,
- provider: state.speechProvider ?? "openai-compatible",
- baseUrl: "https://api.example.test/v1",
- modelId: "whisper-1",
- apiKey,
- });
- }
- if (path === "/api/v1/memos/memo_abc" && (init.method ?? "GET") === "GET") {
- return jsonResponse({ memo: state.memo });
- }
- if (path === "/api/v1/memos/memo_abc" && init.method === "PATCH") {
- state.patched = body;
- return jsonResponse({ memo: { ...state.memo, revision: 1 } }, state.patchStatus ?? 200);
- }
- return jsonResponse({ error: { code: "unexpected" } }, 500);
-};
-
-const baseState = (overrides = {}) => ({
- enabled: true,
- jobs: [{ ...waitingJob }],
- memo: { ...memo },
- calls: [],
- finished: null,
- patched: null,
- prepared: false,
- ...overrides,
-});
-
-const tempDir = async () => {
- const directory = join(tmpdir(), `edgeever-transcript-${crypto.randomUUID()}`);
- await mkdir(directory, { recursive: true });
- return directory;
-};
-
-const readyTool = { state: "ready", path: "/usr/local/bin/yt-dlp" };
-
-test("downloads bestaudio with the chosen browser cookies and without extraction or a shell", () => {
- const args = bestAudioArguments(sourceUrl, "/tmp/memo_abc.%(ext)s");
- expect(args).toContain("-f");
- expect(args).toContain("bestaudio");
- expect(args).toContain("--max-filesize");
- expect(args).toContain("--cookies-from-browser");
- expect(args[args.indexOf("--cookies-from-browser") + 1]).toBe("chrome");
- expect(args).toContain("--");
- expect(args.at(-1)).toBe(sourceUrl);
- expect(args).not.toContain("-x");
- expect(args).not.toContain("--audio-format");
- const firefox = bestAudioArguments(sourceUrl, "/tmp/memo_abc.%(ext)s", "Firefox");
- expect(firefox[firefox.indexOf("--cookies-from-browser") + 1]).toBe("firefox");
- const off = bestAudioArguments(sourceUrl, "/tmp/memo_abc.%(ext)s", "none");
- expect(off.join(" ")).not.toContain("cookie");
- const rejected = bestAudioArguments(sourceUrl, "/tmp/memo_abc.%(ext)s", "chrome:Profile --proxy");
- expect(rejected[rejected.indexOf("--cookies-from-browser") + 1]).toBe("chrome");
- expect(COOKIE_BROWSERS).toEqual([...VIDEO_COOKIE_BROWSERS]);
- expect(DEFAULT_COOKIE_BROWSER).toBe("chrome");
- expect(isPublicVideoSourceUrl(sourceUrl)).toBe(true);
- expect(isPublicVideoSourceUrl("https://www.youtube.com/shorts/abcdefghijk")).toBe(true);
- expect(isPublicVideoSourceUrl("https://www.bilibili.com/video/BV1xx411c7xx?p=2")).toBe(true);
- expect(isPublicVideoSourceUrl("http://www.youtube.com/watch?v=abcdefghijk")).toBe(false);
- expect(isPublicVideoSourceUrl("https://user:[email protected]/watch?v=abcdefghijk")).toBe(false);
- expect(isPublicVideoSourceUrl("https://evil.youtube.com.example/watch?v=abcdefghijk")).toBe(false);
-});
-
-test("stores only an allowed browser name and defaults to chrome", async () => {
- const directory = await tempDir();
- const filePath = join(directory, "video-cookie-browser.json");
- expect(await readCookieBrowserPreference(filePath)).toBe("chrome");
- expect(await writeCookieBrowserPreference(filePath, "Edge")).toEqual({ browser: "edge" });
- expect(await readCookieBrowserPreference(filePath)).toBe("edge");
- expect(await readFile(filePath, "utf8")).toBe('{"browser":"edge"}\n');
- expect(await writeCookieBrowserPreference(filePath, "none")).toEqual({ browser: "none" });
- expect(await readCookieBrowserPreference(filePath)).toBe("none");
- await expect(writeCookieBrowserPreference(filePath, "chrome:Default")).rejects.toThrow("unsupported_browser");
- expect(await readCookieBrowserPreference(filePath)).toBe("none");
- await writeFile(filePath, "{\"browser\":\"not-a-browser\"}\n");
- expect(await readCookieBrowserPreference(filePath)).toBe("chrome");
- await rm(directory, { recursive: true, force: true });
-});
-
-test("writes timestamped lines when segments exist and plain text when they do not", () => {
- const timed = buildTranscriptMarkdown({
- label: "字幕实录",
- text: "ignored when cues exist",
- cues: [{ start: 1.2, end: 3, text: "hello & there" }],
- sourceUrl,
- platform: "youtube",
- });
- expect(timed).toContain("<summary>字幕实录</summary>");
- expect(timed).toContain("[00:01](https://www.youtube.com/watch?v=abcdefghijk&t=1s)");
- expect(timed).toContain("hello &amp; there");
- const plain = buildTranscriptMarkdown({
- label: "字幕实录",
- text: "only words",
- cues: [],
- sourceUrl,
- platform: "youtube",
- });
- expect(plain).toContain("only words");
- expect(plain).not.toContain("](http");
- const replaced = applyTranscript(`# Title\n\n${placeholder}\n`, placeholder, timed);
- expect(replaced).not.toContain(placeholder);
- expect(replaced).toContain("<details>");
- const marker = "<!-- edgeever-video-v1:abc -->";
- const kept = applyTranscript(`# Title\n\n${placeholder}\n\n${marker}\n`, placeholder, "字幕");
- expect(kept.trim().endsWith(marker)).toBe(true);
- expect(kept.indexOf("字幕")).toBeLessThan(kept.indexOf(marker));
-});
-
-test("asks the OpenAI-compatible client for segment timestamps and does not retry", () => {
- const audio = new Uint8Array([1, 2, 3]);
- const signal = AbortSignal.timeout(1000);
- expect(transcriptionRequest({
- baseUrl: "https://api.example.test/v1",
- apiKey,
- modelId: "whisper-1",
- audio,
- abortSignal: signal,
- })).toEqual({
- provider: { baseURL: "https://api.example.test/v1", apiKey },
- modelId: "whisper-1",
- audio,
- maxRetries: 0,
- abortSignal: signal,
- providerOptions: { openai: { timestampGranularities: ["segment"] } },
- });
-});
-
-test("skips the job when speech recognition is off or yt-dlp is not ready", async () => {
- const disabled = baseState({ enabled: false });
- const disabledResult = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(disabled),
- audioDirectory: async () => "/tmp/unused",
- });
- expect(disabledResult).toEqual({ outcome: "skipped", reason: "disabled" });
- expect(disabled.calls.map((call) => call.path)).toEqual(["/api/v1/ai/transcription-settings"]);
-
- const waiting = baseState();
- const toolResult = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => ({ state: "downloading", path: null }),
- fetchImpl: installFetch(waiting),
- audioDirectory: async () => "/tmp/unused",
- });
- expect(toolResult).toEqual({ outcome: "skipped", reason: "yt_dlp" });
- expect(waiting.calls.map((call) => call.path)).toEqual(["/api/v1/ai/transcription-settings"]);
- expect(waiting.prepared).toBe(false);
-});
-
-test("does not download when the note changed or the video is too long", async () => {
- const directory = await tempDir();
- const changed = baseState({ memo: { ...memo, contentHash: "edited" } });
- let spawned = 0;
- const changedResult = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(changed),
- audioDirectory: async () => directory,
- execFileImpl: () => { spawned += 1; },
- transcribe: async () => { throw new Error("should not transcribe"); },
- });
- expect(changedResult.reason).toBe("note_changed");
- expect(spawned).toBe(0);
- expect(changed.prepared).toBe(false);
- expect(changed.finished).toMatchObject({ status: "failed", errorCode: "note_changed" });
-
- const long = baseState({ jobs: [{ ...waitingJob, durationSeconds: 30 * 60 + 1 }] });
- const longResult = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(long),
- audioDirectory: async () => directory,
- execFileImpl: () => { spawned += 1; },
- });
- expect(longResult.reason).toBe("duration_limit");
- expect(spawned).toBe(0);
- expect(long.prepared).toBe(false);
- await rm(directory, { recursive: true, force: true });
-});
-
-test("does not call the OpenAI transcriber for another compatibility standard", async () => {
- const state = baseState({ speechProvider: "future-vendor" });
- let spawned = 0;
- let transcribed = false;
- const result = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(state),
- audioDirectory: async () => "/tmp/unused",
- execFileImpl: () => { spawned += 1; },
- transcribe: async () => { transcribed = true; },
- });
- expect(result.reason).toBe("disabled");
- expect(spawned).toBe(0);
- expect(transcribed).toBe(false);
- expect(state.prepared).toBe(true);
- expect(state.finished).toMatchObject({ status: "failed", errorCode: "disabled" });
-});
-
-test("transcribes the local file and writes the same note", async () => {
- const directory = await tempDir();
- const state = baseState();
- let seenAudio = null;
- const providerCalls = [];
- const result = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- cookieBrowser: async () => "safari",
- fetchImpl: installFetch(state),
- audioDirectory: async () => directory,
- maxAudioBytes: 32,
- execFileImpl: (_binary, args, options, callback) => {
- expect(options.shell).toBeUndefined();
- expect(args).toContain("bestaudio");
- expect(args).not.toContain("-x");
- expect(args[args.indexOf("--cookies-from-browser") + 1]).toBe("safari");
- return writeFile(join(directory, "memo_abc.m4a"), Uint8Array.from([9, 8, 7, 6])).then(() => callback(null));
- },
- createProvider: (settings) => {
- providerCalls.push(settings);
- return { transcription: (modelId) => ({ modelId }) };
- },
- transcribe: async ({ model, audio, maxRetries, providerOptions }) => {
- seenAudio = audio;
- expect(model).toEqual({ modelId: "whisper-1" });
- expect(maxRetries).toBe(0);
- expect(providerOptions).toEqual({ openai: { timestampGranularities: ["segment"] } });
- return {
- text: "hello there",
- segments: [{ text: " hello there", startSecond: 1.4, endSecond: 2 }],
- };
- },
- });
- expect(result).toEqual({ outcome: "ready", memoId: "memo_abc" });
- expect(providerCalls).toEqual([{ baseURL: "https://api.example.test/v1", apiKey }]);
- expect(Array.from(seenAudio)).toEqual([9, 8, 7, 6]);
- expect(state.patched.contentMarkdown).toContain("[00:01](https://www.youtube.com/watch?v=abcdefghijk&t=1s)");
- expect(state.patched.contentMarkdown).not.toContain(placeholder);
- expect(state.patched.expectedRevision).toBe(0);
- expect(state.patched.expectedContentHash).toBe("hash-1");
- expect(state.finished).toEqual({
- status: "ready",
- errorCode: null,
- claimedAt: "2026-10-06T00:00:00.000Z",
- });
- expect(JSON.stringify(state.finished)).not.toContain(apiKey);
- expect(state.calls.every((call) => call.authorization === "Bearer desktop-session")).toBe(true);
- expect(state.calls.some((call) => call.path.includes("audio/transcriptions"))).toBe(false);
- const names = await readFile(join(directory, "memo_abc.m4a")).then(() => ["left"], () => []);
- expect(names).toEqual([]);
- await rm(directory, { recursive: true, force: true });
-});
-
-test("leaves the note in place when recognition fails and does not store the token", async () => {
- const directory = await tempDir();
- const state = baseState();
- const result = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(state),
- audioDirectory: async () => directory,
- execFileImpl: (_binary, _args, _options, callback) => (
- writeFile(join(directory, "memo_abc.webm"), Uint8Array.from([1])).then(() => callback(null))
- ),
- createProvider: () => ({ transcription: () => ({}) }),
- transcribe: async () => {
- throw new Error(`provider rejected ${apiKey}`);
- },
- });
- expect(result.reason).toBe("transcribe_failed");
- expect(state.patched).toBeNull();
- expect(state.finished).toMatchObject({ status: "failed", errorCode: "transcribe_failed" });
- expect(JSON.stringify(state.finished)).not.toContain(apiKey);
- await expect(readFile(join(directory, "memo_abc.webm"))).rejects.toThrow();
- await rm(directory, { recursive: true, force: true });
-});
-
-test("refuses an oversized download before calling the recognizer", async () => {
- const directory = await tempDir();
- const state = baseState();
- let transcribed = false;
- const result = await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(state),
- audioDirectory: async () => directory,
- maxAudioBytes: 3,
- execFileImpl: (_binary, _args, _options, callback) => (
- writeFile(join(directory, "memo_abc.m4a"), Uint8Array.from([1, 2, 3, 4])).then(() => callback(null))
- ),
- transcribe: async () => { transcribed = true; },
- });
- expect(result.reason).toBe("audio_too_large");
- expect(transcribed).toBe(false);
- expect(state.patched).toBeNull();
- await rm(directory, { recursive: true, force: true });
-});
-
-test("saves text without timestamps when the recognizer returns no segments", async () => {
- const directory = await tempDir();
- const state = baseState();
- await runSpeechTranscriptionPass({
- getSession: () => session,
- ytDlpStatus: () => readyTool,
- fetchImpl: installFetch(state),
- audioDirectory: async () => directory,
- execFileImpl: (_binary, _args, _options, callback) => (
- writeFile(join(directory, "memo_abc.m4a"), Uint8Array.from([1])).then(() => callback(null))
- ),
- createProvider: () => ({ transcription: () => ({}) }),
- transcribe: async () => ({ text: "plain words", segments: [] }),
- });
- expect(state.patched.contentMarkdown).toContain("plain words");
- expect(state.patched.contentMarkdown).not.toContain("](http");
- expect(state.finished.status).toBe("ready");
- await rm(directory, { recursive: true, force: true });
-});
-
-test("does not start a second pass while one is still running, and the app does not await it", async () => {
- let release = () => {};
- const gate = new Promise((resolve) => { release = resolve; });
- let calls = 0;
- let scheduled = null;
- const stop = startSpeechTranscription({
- pass: () => {
- calls += 1;
- return gate;
- },
- }, {
- intervalMs: 20,
- schedule: (callback) => {
- scheduled = callback;
- return 4;
- },
- cancel: () => { scheduled = null; },
- });
- await Promise.resolve();
- expect(calls).toBe(1);
- scheduled();
- await Promise.resolve();
- expect(calls).toBe(1);
- release();
- await gate;
- await Promise.resolve();
- await Promise.resolve();
- scheduled();
- await Promise.resolve();
- expect(calls).toBe(2);
- stop();
- expect(scheduled).toBeNull();
-
- const main = await readFile(new URL("./index.mjs", import.meta.url), "utf8");
- const windowAt = main.indexOf("await createWindow()");
- const startAt = main.indexOf("startSpeechTranscription(");
- expect(startAt).toBeGreaterThan(windowAt);
- expect(main).not.toContain("await startSpeechTranscription");
- expect(main).not.toContain("await runSpeechTranscriptionPass");
- expect(main).toContain("stopSpeechTranscription?.()");
-});
-
-test("the installed provider can build a transcription model without a network call", async () => {
- const { createOpenAI } = await import("@ai-sdk/openai");
- const provider = createOpenAI({ baseURL: "https://example.test/v1", apiKey: "not-a-real-token" });
- expect(typeof provider.transcription).toBe("function");
- expect(provider.transcription("whisper-1")).toBeTruthy();
-});
-
-test("the runner factory passes the session through", async () => {
- const runner = createSpeechTranscriptionRunner({
- getSession: () => ({}),
- ytDlpStatus: () => readyTool,
- audioDirectory: async () => "/tmp/unused",
- });
- await expect(runner.pass()).resolves.toEqual({ outcome: "skipped", reason: "signed_out" });
-});
diff --git a/apps/desktop/src/main/yt-dlp-tool.mjs b/apps/desktop/src/main/yt-dlp-tool.mjs
deleted file mode 100644
index 8de4bd6c..00000000
--- a/apps/desktop/src/main/yt-dlp-tool.mjs
+++ /dev/null
@@ -1,297 +0,0 @@
-import { createHash, timingSafeEqual } from "node:crypto";
-import { execFile } from "node:child_process";
-import { existsSync } from "node:fs";
-import { chmod, mkdir, open, readFile, rename, unlink, writeFile } from "node:fs/promises";
-import { join } from "node:path";
-import { Readable } from "node:stream";
-import { restrictDirectory, restrictFile } from "./file-permissions.mjs";
-
-const RELEASES_LATEST_URL = "https://github.com/yt-dlp/yt-dlp/releases/latest";
-const USER_AGENT = "EdgeEverDesktop yt-dlp-fetch";
-const MIN_BINARY_BYTES = 100_000;
-const MAX_BINARY_BYTES = 80 * 1024 * 1024;
-const MAX_SUMS_BYTES = 1024 * 1024;
-
-const toolError = (code, httpStatus = null) => {
- const error = new Error(code);
- error.code = code;
- error.httpStatus = httpStatus;
- return error;
-};
-
-// Official standalone builds. The 2.9 MB `yt-dlp` zipapp still needs a system Python.
-export const ytDlpAssetName = (platform, arch) => {
- if (platform === "darwin") return "yt-dlp_macos";
- if (platform === "win32" && arch === "arm64") return "yt-dlp_arm64.exe";
- if (platform === "win32") return "yt-dlp.exe";
- if (platform === "linux" && arch === "arm64") return "yt-dlp_linux_aarch64";
- if (platform === "linux" && arch === "x64") return "yt-dlp_linux";
- return null;
-};
-
-export const ytDlpExecutableName = (platform) => (platform === "win32" ? "yt-dlp.exe" : "yt-dlp");
-
-export const parseSha256Sums = (text) => {
- const sums = new Map();
- for (const line of String(text).split(/\r?\n/)) {
- const match = /^([a-fA-F0-9]{64})\s+\*?(\S+)\s*$/.exec(line.trim());
- if (match) sums.set(match[2], match[1].toLowerCase());
- }
- return sums;
-};
-
-export const releaseTagFromUrl = (url) => {
- const match = /\/releases\/(?:tag|download)\/([^/?#]+)/.exec(String(url));
- if (!match) return null;
- try {
- return decodeURIComponent(match[1]);
- } catch {
- return match[1];
- }
-};
-
-const hashesMatch = (actual, expected) => {
- const left = Buffer.from(actual, "hex");
- const right = Buffer.from(expected, "hex");
- return left.length === right.length && left.length > 0 && timingSafeEqual(left, right);
-};
-
-const readInstalledVersion = async (binaryPath) => {
- const stdout = await new Promise((resolve, reject) => {
- execFile(binaryPath, ["--version"], { timeout: 15_000, windowsHide: true }, (error, output) => {
- if (error) reject(error);
- else resolve(output);
- });
- });
- return String(stdout).trim().split(/\s+/)[0] ?? "";
-};
-
-const fetchResponse = async (fetchImpl, url, timeoutMs) => {
- let response;
- try {
- response = await fetchImpl(url, {
- redirect: "follow",
- headers: { "user-agent": USER_AGENT },
- signal: AbortSignal.timeout(timeoutMs),
- });
- } catch (error) {
- throw toolError(error?.name === "TimeoutError" ? "timeout" : "network");
- }
- if (!response.ok) throw toolError("http", response.status);
- return response;
-};
-
-const fetchBytes = async (fetchImpl, url, { maxBytes, timeoutMs }) => {
- const response = await fetchResponse(fetchImpl, url, timeoutMs);
- const bytes = Buffer.from(await response.arrayBuffer());
- if (bytes.byteLength > maxBytes) throw toolError("unexpected_size");
- return bytes;
-};
-
-// Stream to disk and hash one chunk at a time. Buffering the whole binary and
-// hashing it in one call stalls the Electron main process and freezes the window.
-const downloadHashed = async (fetchImpl, url, destination, { maxBytes, minBytes, timeoutMs }) => {
- const response = await fetchResponse(fetchImpl, url, timeoutMs);
- const hash = createHash("sha256");
- let size = 0;
- const consume = (chunk) => {
- const bytes = Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk);
- size += bytes.byteLength;
- if (size > maxBytes) throw toolError("unexpected_size");
- hash.update(bytes);
- return bytes;
- };
- const writeAll = async (handle, bytes) => {
- let offset = 0;
- while (offset < bytes.byteLength) {
- const { bytesWritten } = await handle.write(bytes, offset, bytes.byteLength - offset);
- if (!bytesWritten) throw toolError("storage");
- offset += bytesWritten;
- }
- };
- const webBody = response.body;
- if (webBody && typeof webBody.getReader === "function") {
- const handle = await open(destination, "w", 0o600);
- try {
- for await (const chunk of Readable.fromWeb(webBody)) {
- await writeAll(handle, consume(chunk));
- }
- } finally {
- await handle.close();
- }
- } else {
- await writeFile(destination, consume(Buffer.from(await response.arrayBuffer())));
- }
- if (size < minBytes) throw toolError("unexpected_size");
- return hash.digest("hex");
-};
-
-const latestReleaseTag = async (fetchImpl) => {
- const response = await fetchResponse(fetchImpl, RELEASES_LATEST_URL, 20_000);
- const page = await response.text();
- if (page.length > MAX_SUMS_BYTES) throw toolError("unexpected_size");
- const fromUrl = releaseTagFromUrl(response.url || "");
- const fromPage = /\/releases\/tag\/(\d{4}\.\d{2}\.\d{2}(?:\.\d+)?)/.exec(page)?.[1] ?? null;
- const tag = fromUrl && /^\d{4}\.\d{2}\.\d{2}/.test(fromUrl) ? fromUrl : fromPage;
- if (!tag || !/^\d{4}\.\d{2}\.\d{2}(?:\.\d+)?$/.test(tag)) throw toolError("unrecognized_release");
- return tag;
-};
-
-export const createYtDlpManager = ({
- directory,
- platform,
- arch,
- fetch: fetchImpl = globalThis.fetch,
- readVersion = readInstalledVersion,
- onDiagnostic = () => {},
-}) => {
- const asset = ytDlpAssetName(platform, arch);
- const binaryPath = join(directory, ytDlpExecutableName(platform));
- const partialPath = `${binaryPath}.partial`;
- const statePath = join(directory, "yt-dlp.json");
- let current = {
- state: asset ? "idle" : "failed",
- version: null,
- path: binaryPath,
- errorCode: asset ? null : "unsupported_platform",
- httpStatus: null,
- };
- let pending = null;
-
- const publish = (next) => {
- current = next;
- };
-
- const readState = async () => {
- if (!existsSync(binaryPath)) return null;
- try {
- const parsed = JSON.parse(await readFile(statePath, "utf8"));
- const version = typeof parsed?.version === "string" ? parsed.version : "";
- return version ? { version } : null;
- } catch {
- return null;
- }
- };
-
- const install = async (tag) => {
- const releaseBase = `https://github.com/yt-dlp/yt-dlp/releases/download/${encodeURIComponent(tag)}`;
- const sums = await fetchBytes(fetchImpl, `${releaseBase}/SHA2-256SUMS`, {
- maxBytes: MAX_SUMS_BYTES,
- timeoutMs: 20_000,
- });
- const expected = parseSha256Sums(sums.toString("utf8")).get(asset);
- if (!expected) throw toolError("checksum_missing");
- await mkdir(directory, { recursive: true });
- await restrictDirectory(directory);
- const actual = await downloadHashed(fetchImpl, `${releaseBase}/${asset}`, partialPath, {
- maxBytes: MAX_BINARY_BYTES,
- minBytes: MIN_BINARY_BYTES,
- timeoutMs: 120_000,
- });
- if (!hashesMatch(actual, expected)) throw toolError("checksum_mismatch");
- await chmod(partialPath, platform === "win32" ? 0o644 : 0o755);
- let reported = "";
- try {
- reported = await readVersion(partialPath);
- } catch {
- throw toolError("not_executable");
- }
- if (reported !== tag) throw toolError("version_mismatch");
- try {
- await rename(partialPath, binaryPath);
- } catch {
- await unlink(binaryPath).catch(() => {});
- await rename(partialPath, binaryPath);
- }
- await writeFile(statePath, `${JSON.stringify({
- version: tag,
- sha256: actual,
- asset,
- installedAt: new Date().toISOString(),
- })}\n`);
- await restrictFile(statePath);
- };
-
- const run = async () => {
- if (!asset) return current;
- const installed = await readState();
- publish({
- state: installed ? "ready" : "downloading",
- version: installed?.version ?? null,
- path: binaryPath,
- errorCode: null,
- httpStatus: null,
- });
- try {
- const tag = await latestReleaseTag(fetchImpl);
- if (installed?.version === tag) return current;
- publish({
- state: "downloading",
- version: installed?.version ?? null,
- path: binaryPath,
- errorCode: null,
- httpStatus: null,
- });
- await install(tag);
- publish({
- state: "ready",
- version: tag,
- path: binaryPath,
- errorCode: null,
- httpStatus: null,
- });
- onDiagnostic(installed ? "yt-dlp.updated" : "yt-dlp.installed", { version: tag });
- } catch (error) {
- const known = new Set([
- "checksum_mismatch",
- "checksum_missing",
- "version_mismatch",
- "not_executable",
- "timeout",
- "network",
- "http",
- "unexpected_size",
- "unrecognized_release",
- "storage",
- ]);
- const errorCode = known.has(error?.code)
- ? error.code
- : ["ENOENT", "EACCES", "EPERM", "ENOSPC", "EROFS"].includes(error?.code)
- ? "storage"
- : "network";
- const httpStatus = Number.isInteger(error?.httpStatus) ? error.httpStatus : null;
- publish({
- state: installed ? "ready" : "failed",
- version: installed?.version ?? null,
- path: binaryPath,
- errorCode,
- httpStatus,
- });
- onDiagnostic("yt-dlp.failed", { version: installed?.version ?? null, errorCode, httpStatus });
- } finally {
- await unlink(partialPath).catch(() => {});
- }
- return current;
- };
-
- return {
- status: () => ({ ...current }),
- ensure: () => {
- if (pending) return pending;
- pending = run().finally(() => {
- pending = null;
- });
- return pending;
- },
- };
-};
-
-export const startYtDlpMaintenance = (manager, { intervalMs, schedule = setInterval, cancel = clearInterval } = {}) => {
- void manager.ensure();
- if (!intervalMs) return () => {};
- const timer = schedule(() => {
- void manager.ensure();
- }, intervalMs);
- timer.unref?.();
- return () => cancel(timer);
-};
diff --git a/apps/desktop/src/main/yt-dlp-tool.test.mjs b/apps/desktop/src/main/yt-dlp-tool.test.mjs
deleted file mode 100644
index 948a97b1..00000000
--- a/apps/desktop/src/main/yt-dlp-tool.test.mjs
+++ /dev/null
@@ -1,217 +0,0 @@
-import { createHash } from "node:crypto";
-import { readFileSync } from "node:fs";
-import { mkdtemp, readFile, rm, stat } from "node:fs/promises";
-import { tmpdir } from "node:os";
-import { join } from "node:path";
-import { describe, expect, test } from "bun:test";
-import {
- createYtDlpManager,
- parseSha256Sums,
- releaseTagFromUrl,
- startYtDlpMaintenance,
- ytDlpAssetName,
- ytDlpExecutableName,
-} from "./yt-dlp-tool.mjs";
-
-const sha256 = (bytes) => createHash("sha256").update(bytes).digest("hex");
-
-describe("yt-dlp release selection", () => {
- test("picks the standalone binary for each desktop target", () => {
- expect(ytDlpAssetName("darwin", "arm64")).toBe("yt-dlp_macos");
- expect(ytDlpAssetName("darwin", "x64")).toBe("yt-dlp_macos");
- expect(ytDlpAssetName("win32", "x64")).toBe("yt-dlp.exe");
- expect(ytDlpAssetName("win32", "arm64")).toBe("yt-dlp_arm64.exe");
- expect(ytDlpAssetName("linux", "x64")).toBe("yt-dlp_linux");
- expect(ytDlpAssetName("linux", "arm64")).toBe("yt-dlp_linux_aarch64");
- expect(ytDlpAssetName("freebsd", "x64")).toBeNull();
- expect(ytDlpExecutableName("win32")).toBe("yt-dlp.exe");
- expect(ytDlpExecutableName("darwin")).toBe("yt-dlp");
- });
-
- test("reads GNU checksum lines and release tags", () => {
- const sums = parseSha256Sums("abc\n0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef yt-dlp_macos\n");
- expect(sums.get("yt-dlp_macos")).toBe("0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef");
- expect(releaseTagFromUrl("https://github.com/yt-dlp/yt-dlp/releases/tag/2026.08.19")).toBe("2026.08.19");
- expect(releaseTagFromUrl("https://github.com/yt-dlp/yt-dlp/releases/latest")).toBeNull();
- });
-});
-
-const response = ({ url, status = 200, body = "", stream = false }) => {
- const payload = Buffer.from(body);
- const result = {
- ok: status >= 200 && status < 300,
- status,
- url,
- text: async () => (typeof body === "string" ? body : payload.toString("utf8")),
- arrayBuffer: async () => Uint8Array.from(payload).buffer,
- };
- if (stream) {
- result.body = new ReadableStream({
- start(controller) {
- const chunk = 32_768;
- for (let offset = 0; offset < payload.byteLength; offset += chunk) {
- controller.enqueue(payload.subarray(offset, offset + chunk));
- }
- controller.close();
- },
- });
- }
- return result;
-};
-
-const fakeRelease = ({ tag, asset, body, checksum, stream = false }) => {
- const calls = [];
- const fetchImpl = async (url) => {
- calls.push(url);
- if (url.endsWith("/releases/latest")) {
- return response({
- url: `https://github.com/yt-dlp/yt-dlp/releases/tag/${tag}`,
- body: `<a href="/releases/tag/${tag}">`,
- });
- }
- if (url.endsWith("/SHA2-256SUMS")) {
- return response({ url, body: `${checksum ?? sha256(body)} ${asset}\n` });
- }
- if (url.endsWith(`/${asset}`)) return response({ url, body, stream });
- return response({ url, status: 404, body: "" });
- };
- return { calls, fetchImpl };
-};
-
-describe("yt-dlp install and update", () => {
- test("downloads a missing binary, skips the same version, and replaces a newer release", async () => {
- const directory = await mkdtemp(join(tmpdir(), "edgeever-yt-dlp-"));
- const diagnostics = [];
- let tag = "2026.08.19";
- let body = Buffer.alloc(120_000, 7);
- const reported = () => tag;
- const release = () => fakeRelease({ tag, asset: "yt-dlp_macos", body });
- let current = release();
- const manager = createYtDlpManager({
- directory,
- platform: "darwin",
- arch: "arm64",
- fetch: (url, init) => current.fetchImpl(url, init),
- readVersion: async () => reported(),
- onDiagnostic: (event, details) => diagnostics.push({ event, details }),
- });
-
- expect(manager.status()).toMatchObject({ state: "idle", version: null, errorCode: null });
- await manager.ensure();
- expect(manager.status()).toMatchObject({ state: "ready", version: "2026.08.19", errorCode: null });
- expect(await readFile(join(directory, "yt-dlp"))).toEqual(body);
- expect((await stat(join(directory, "yt-dlp"))).mode & 0o111).toBeGreaterThan(0);
- expect(diagnostics.map((item) => item.event)).toEqual(["yt-dlp.installed"]);
-
- const callsAfterInstall = current.calls.length;
- current = release();
- await manager.ensure();
- expect(current.calls.some((url) => url.endsWith("/yt-dlp_macos"))).toBe(false);
- expect(current.calls.length).toBeLessThan(callsAfterInstall);
-
- tag = "2026.09.27";
- body = Buffer.alloc(120_000, 9);
- current = release();
- await manager.ensure();
- expect(manager.status()).toMatchObject({ state: "ready", version: "2026.09.27", errorCode: null });
- expect(await readFile(join(directory, "yt-dlp"))).toEqual(body);
- expect(diagnostics.at(-1)).toEqual({ event: "yt-dlp.updated", details: { version: "2026.09.27" } });
- await rm(directory, { recursive: true, force: true });
- });
-
- test("keeps the installed binary when a newer download fails its checksum", async () => {
- const directory = await mkdtemp(join(tmpdir(), "edgeever-yt-dlp-"));
- let tag = "2026.08.19";
- let body = Buffer.alloc(120_000, 3);
- let checksum = null;
- const manager = createYtDlpManager({
- directory,
- platform: "darwin",
- arch: "arm64",
- fetch: (url) => fakeRelease({ tag, asset: "yt-dlp_macos", body, checksum }).fetchImpl(url),
- readVersion: async () => tag,
- });
- await manager.ensure();
- const installed = await readFile(join(directory, "yt-dlp"));
-
- tag = "2026.09.27";
- body = Buffer.alloc(120_000, 4);
- checksum = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa";
- await manager.ensure();
- expect(manager.status()).toMatchObject({
- state: "ready",
- version: "2026.08.19",
- errorCode: "checksum_mismatch",
- });
- expect(await readFile(join(directory, "yt-dlp"))).toEqual(installed);
- await rm(directory, { recursive: true, force: true });
- });
-
- test("reports a failed first download without leaving a binary", async () => {
- const directory = await mkdtemp(join(tmpdir(), "edgeever-yt-dlp-"));
- const manager = createYtDlpManager({
- directory,
- platform: "darwin",
- arch: "arm64",
- fetch: async () => response({ url: "https://github.com/yt-dlp/yt-dlp/releases/latest", status: 503 }),
- readVersion: async () => "2026.08.19",
- });
- await manager.ensure();
- expect(manager.status()).toMatchObject({ state: "failed", version: null, errorCode: "http", httpStatus: 503 });
- await expect(stat(join(directory, "yt-dlp"))).rejects.toThrow();
- await rm(directory, { recursive: true, force: true });
- });
-
- test("writes a streamed release body in chunks", async () => {
- const directory = await mkdtemp(join(tmpdir(), "edgeever-yt-dlp-"));
- const body = Buffer.alloc(120_000, 11);
- const manager = createYtDlpManager({
- directory,
- platform: "darwin",
- arch: "arm64",
- fetch: (url) => fakeRelease({ tag: "2026.08.19", asset: "yt-dlp_macos", body, stream: true }).fetchImpl(url),
- readVersion: async () => "2026.08.19",
- });
- await manager.ensure();
- expect(manager.status()).toMatchObject({ state: "ready", version: "2026.08.19", errorCode: null });
- expect(await readFile(join(directory, "yt-dlp"))).toEqual(body);
- await rm(directory, { recursive: true, force: true });
- });
-
- test("starts the download after the window is open and does not await it", () => {
- const main = readFileSync(new URL("./index.mjs", import.meta.url), "utf8");
- const windowAt = main.indexOf("await createWindow()");
- const confirmAt = main.indexOf("await confirmMacInstallation()");
- const startAt = main.indexOf("startYtDlpMaintenance(");
- expect(windowAt).toBeGreaterThan(0);
- expect(confirmAt).toBeGreaterThan(windowAt);
- expect(startAt).toBeGreaterThan(confirmAt);
- expect(main).toContain("ytDlpMaintenanceTimer = setTimeout");
- expect(main).toContain("clearTimeout(ytDlpMaintenanceTimer)");
- expect(main).not.toContain("await startYtDlpMaintenance");
- expect(main).not.toContain("await ytDlpManager.ensure");
- });
-
- test("schedules a later check and can stop it", () => {
- const calls = [];
- let scheduled = null;
- const stop = startYtDlpMaintenance({
- ensure: () => {
- calls.push("now");
- return Promise.resolve();
- },
- }, {
- intervalMs: 50,
- schedule: (callback) => {
- scheduled = callback;
- return 7;
- },
- cancel: (timer) => {
- calls.push(timer);
- },
- });
- scheduled();
- stop();
- expect(calls).toEqual(["now", "now", 7]);
- });
-});
diff --git a/apps/desktop/src/preload/index.cjs b/apps/desktop/src/preload/index.cjs
index 5b1f12cf..1bdd05c4 100644
--- a/apps/desktop/src/preload/index.cjs
+++ b/apps/desktop/src/preload/index.cjs
@@ -30,9 +30,6 @@ contextBridge.exposeInMainWorld("edgeeverDesktop", Object.freeze({
recoveredAfterAbnormalExit: ipcRenderer.sendSync("desktop:recovered-after-abnormal-exit-sync"),
sidecarStatus: () => ipcRenderer.invoke("desktop:sidecar-status"),
systemInfo: () => ipcRenderer.invoke("desktop:system-info"),
- ytDlpStatus: () => ipcRenderer.invoke("desktop:yt-dlp-status"),
- videoCookieBrowser: () => ipcRenderer.invoke("desktop:video-cookie-browser-get"),
- setVideoCookieBrowser: (browser) => ipcRenderer.invoke("desktop:video-cookie-browser-set", browser),
setAccountScope: (accountId) => ipcRenderer.invoke("desktop:set-account-scope", accountId),
apiBaseUrl: ipcRenderer.sendSync("desktop:api-base-url-sync"),
setApiBaseUrl: (value) => ipcRenderer.invoke("desktop:set-api-base-url", value),
diff --git a/apps/extension/src/background.ts b/apps/extension/src/background.ts
index c77ae973..9d4b2163 100644
--- a/apps/extension/src/background.ts
+++ b/apps/extension/src/background.ts
@@ -267,7 +267,6 @@ const imageNoteClient = (settings: ExtensionSettings): ImageNoteClient => ({
form.append("title", body.title);
form.append("contentMarkdown", body.contentMarkdown);
form.append("tags", JSON.stringify(body.tags));
- if (body.videoTranscript) form.append("videoTranscript", JSON.stringify(body.videoTranscript));
form.append("file", new File([buffer], body.filename, { type: body.mimeType }));
const created = await edgeEverFormRequest<{ memo: { id: string }; resourceId: string }>(
settings,
diff --git a/apps/extension/src/image-clip.ts b/apps/extension/src/image-clip.ts
index 3addaedf..6f601d70 100644
--- a/apps/extension/src/image-clip.ts
+++ b/apps/extension/src/image-clip.ts
@@ -289,14 +289,6 @@ export type ImageNoteClient = {
filename: string;
mimeType: string;
bytes: Uint8Array;
- videoTranscript?: {
- platform: "youtube" | "bilibili";
- videoId: string;
- sourceUrl: string;
- durationSeconds: number;
- placeholderText: string;
- transcriptLabel: string;
- };
}) => Promise<{ memoId: string; resourceId: string }>;
};
diff --git a/apps/extension/src/video/video-note.ts b/apps/extension/src/video/video-note.ts
index 4078ccae..f83dcd8a 100644
--- a/apps/extension/src/video/video-note.ts
+++ b/apps/extension/src/video/video-note.ts
@@ -27,35 +27,6 @@ export type VideoNoteLabels = {
const COVER_PLACEHOLDER = "EDGEVERRESOURCEID";
-export type VideoTranscriptRequest = {
- platform: "youtube" | "bilibili";
- videoId: string;
- sourceUrl: string;
- durationSeconds: number;
- placeholderText: string;
- transcriptLabel: string;
-};
-
-export const videoNoteIdentity = (
- capture: VideoCapture,
- labels: VideoNoteLabels,
-): VideoTranscriptRequest => ({
- platform: capture.platform,
- videoId: capture.videoId,
- sourceUrl: capture.sourceUrl,
- durationSeconds: Math.max(0, Math.floor(capture.duration || 0)),
- placeholderText: labels.noCaptions,
- transcriptLabel: labels.transcript,
-});
-
-const videoNoteMarker = (identity: VideoTranscriptRequest) => {
- const bytes = new TextEncoder().encode(JSON.stringify(identity));
- let binary = "";
- for (const byte of bytes) binary += String.fromCharCode(byte);
- const encoded = btoa(binary).replaceAll("+", "-").replaceAll("/", "_").replace(/=+$/g, "");
- return `<!-- edgeever-video-v1:${encoded} -->`;
-};
-
export const videoNoteTitle = (title: string, fallback: string) => {
const clean = title.replace(/[\r\n]+/g, " ").replace(/\s+/g, " ").trim();
const value = clean || fallback;
@@ -150,7 +121,6 @@ export const buildVideoNote = (input: {
"",
);
}
- lines.push(videoNoteMarker(videoNoteIdentity(input.capture, input.labels)));
return { title, markdown: lines.join("\n").replace(/\n{3,}/g, "\n\n").trim() + "\n" };
};
@@ -265,7 +235,6 @@ export const persistVideoNote = async (input: {
title: string;
contentMarkdown: string;
tags: string[];
- videoTranscript?: VideoTranscriptRequest;
}) => Promise<unknown>;
createWithImage?: (body: {
notebookId: string;
@@ -275,7 +244,6 @@ export const persistVideoNote = async (input: {
filename: string;
mimeType: string;
bytes: Uint8Array;
- videoTranscript?: VideoTranscriptRequest;
}) => Promise<unknown>;
}) => {
const withCover = Boolean(input.capture.thumbnail && input.createWithImage);
diff --git a/apps/extension/video-clip.test.mjs b/apps/extension/video-clip.test.mjs
index cfac358a..138415e9 100644
--- a/apps/extension/video-clip.test.mjs
+++ b/apps/extension/video-clip.test.mjs
@@ -13,7 +13,6 @@ import { VIDEO_DOCUMENT_PATTERNS } from "./src/video/patterns.ts";
import { readBilibiliVideoInPage } from "./src/video/read-bilibili-in-page.ts";
import { readYouTubeVideoInPage } from "./src/video/read-youtube-in-page.ts";
import { chunkTranscript, cleanCueText, sectionsFromOutline } from "./src/video/transcript.ts";
-import { parseVideoNoteMarker } from "../../packages/shared/src/video-note.ts";
import {
buildVideoNote,
persistVideoNote,
@@ -534,8 +533,6 @@ test("assembles the Chinese note from chapters and keeps the transcript outside
- [00:25](https://www.bilibili.com/video/BV1xx411c7xx?p=1&t=25) 很多人收藏之后就没有再打开。
</details>
-
-<!-- edgeever-video-v1:eyJwbGF0Zm9ybSI6ImJpbGliaWxpIiwidmlkZW9JZCI6IkJWMXh4NDExYzd4eCIsInNvdXJjZVVybCI6Imh0dHBzOi8vd3d3LmJpbGliaWxpLmNvbS92aWRlby9CVjF4eDQxMWM3eHg_cD0xIiwiZHVyYXRpb25TZWNvbmRzIjoxNDU1LCJwbGFjZWhvbGRlclRleHQiOiLov5nkuIDpm4bmsqHmnInlj6_nlKjlrZfluZUiLCJ0cmFuc2NyaXB0TGFiZWwiOiLlrZfluZXlrp7lvZUifQ -->
`);
expect(note.markdown).not.toContain("模型不该重写大纲");
expect(note.markdown).not.toContain("hdslb.com");
@@ -660,14 +657,10 @@ test("omits the cover when the image upload fails and never hotlinks the thumbna
expect(memos[0].contentMarkdown).toContain("caption");
expect(memos[0].videoTranscript).toBeUndefined();
expect(uploaded[0].videoTranscript).toBeUndefined();
- expect(parseVideoNoteMarker(memos[0].contentMarkdown)).toMatchObject({
- platform: "youtube",
- videoId: CURRENT_ID,
- sourceUrl: `https://www.youtube.com/watch?v=${CURRENT_ID}`,
- });
+ expect(memos[0].contentMarkdown).not.toContain("edgeever-video-v1");
});
-test("marks every video note and leaves transcription for a later click", async () => {
+test("saves a video without captions as a source note without queuing transcription", async () => {
const capture = {
platform: "youtube",
videoId: CURRENT_ID,
@@ -693,14 +686,8 @@ test("marks every video note and leaves transcription for a later click", async
expect(uploaded[0].videoTranscript).toBeUndefined();
expect(memos[0].videoTranscript).toBeUndefined();
expect(memos[0].contentMarkdown).toContain("这一集没有可用字幕");
- expect(parseVideoNoteMarker(memos[0].contentMarkdown)).toEqual({
- platform: "youtube",
- videoId: CURRENT_ID,
- sourceUrl: `https://www.youtube.com/watch?v=${CURRENT_ID}`,
- durationSeconds: 95,
- placeholderText: "这一集没有可用字幕",
- transcriptLabel: "字幕实录",
- });
+ expect(memos[0].contentMarkdown).toContain(`https://www.youtube.com/watch?v=${CURRENT_ID}`);
+ expect(memos[0].contentMarkdown).not.toContain("edgeever-video-v1");
});
test("keeps the video menu hosts away from the other page commands", () => {
diff --git a/apps/web/src/components/MemoListPane.tsx b/apps/web/src/components/MemoListPane.tsx
index ba37828e..ccaa8fa3 100644
--- a/apps/web/src/components/MemoListPane.tsx
+++ b/apps/web/src/components/MemoListPane.tsx
@@ -50,7 +50,6 @@ import {
Printer,
Pencil,
Copy,
- Captions,
} from "lucide-react";
import { WeChatIcon } from "./WeChatIcon";
import { Button } from "@/components/ui/button";
@@ -523,11 +522,9 @@ export const MemoListPane = ({
const [moveTargetNotebookId, setMoveTargetNotebookId] = useState("");
const [memoIdCopyNotice, setMemoIdCopyNotice] = useState<{ status: "copied" | "error"; id: string } | null>(null);
const [wechatCopyNotice, setWechatCopyNotice] = useState<"copied" | "error" | null>(null);
- const [extractNotice, setExtractNotice] = useState<"started" | "running" | "error" | null>(null);
const [wechatCopyPending, setWechatCopyPending] = useState(false);
const wechatCopyPendingRef = useRef(false);
const wechatCopyNoticeTimerRef = useRef<number | null>(null);
- const extractNoticeTimerRef = useRef<number | null>(null);
const [fileDragActive, setFileDragActive] = useState(false);
const filterOptions = useMemo(() => getMemoFilterOptions(t), [t]);
@@ -546,9 +543,6 @@ export const MemoListPane = ({
if (wechatCopyNoticeTimerRef.current !== null) {
window.clearTimeout(wechatCopyNoticeTimerRef.current);
}
- if (extractNoticeTimerRef.current !== null) {
- window.clearTimeout(extractNoticeTimerRef.current);
- }
}, []);
const listScrollRef = useRef<HTMLDivElement | null>(null);
const [listScrollElement, setListScrollElement] = useState<HTMLDivElement | null>(null);
@@ -668,24 +662,6 @@ export const MemoListPane = ({
onRequestDocumentAction(memo.id, action, printWindow);
};
- const showExtractNotice = (status: "started" | "running" | "error") => {
- if (extractNoticeTimerRef.current !== null) window.clearTimeout(extractNoticeTimerRef.current);
- setExtractNotice(status);
- extractNoticeTimerRef.current = window.setTimeout(() => setExtractNotice(null), status === "error" ? 2600 : 2200);
- };
-
- const handleExtractTranscript = async () => {
- const memo = memoContextMenu?.memo;
- if (!memo || !memo.videoNote || isLocalMemoId(memo.id)) return;
- setMemoContextMenu(null);
- try {
- const result = await api.extractVideoTranscript(memo.id);
- showExtractNotice(result.status === "running" ? "running" : "started");
- } catch {
- showExtractNotice("error");
- }
- };
-
const handleCopyContextMemoId = async () => {
const memo = memoContextMenu?.memo;
if (!memo || isLocalMemoId(memo.id)) return;
@@ -1598,16 +1574,6 @@ export const MemoListPane = ({
<FileIcon className="h-4 w-4" />
{t("memoList.openMemo")}
</DropdownMenuItem>
- {view !== "trash" && memoContextMenu.memo.videoNote && (
- <DropdownMenuItem
- className="flex h-9 w-full items-center gap-2 px-3 text-left text-xs text-slate-700 hover:bg-slate-50 cursor-pointer outline-none"
- disabled={isLocalMemoId(memoContextMenu.memo.id)}
- onClick={() => void handleExtractTranscript()}
- >
- <Captions className="h-4 w-4 text-slate-500" />
- {t("memoList.extractTranscript")}
- </DropdownMenuItem>
- )}
<DropdownMenuItem
className="flex h-9 w-full items-center gap-2 px-3 text-left text-xs text-slate-700 hover:bg-slate-50 cursor-pointer outline-none"
onClick={() => {
@@ -1786,16 +1752,6 @@ export const MemoListPane = ({
</ClipboardCopyNotice>
)}
- {extractNotice && (
- <ClipboardCopyNotice status={extractNotice === "error" ? "error" : "copied"}>
- {t(extractNotice === "started"
- ? "memoList.extractTranscriptStarted"
- : extractNotice === "running"
- ? "memoList.extractTranscriptRunning"
- : "memoList.extractTranscriptFailed")}
- </ClipboardCopyNotice>
- )}
-
{selectionMode && (
<MobileSelectionActionBar
canDelete={selectedMemoIds.size > 0 && !isDeleting}
diff --git a/apps/web/src/components/SettingsPane.tsx b/apps/web/src/components/SettingsPane.tsx
index 70fb4592..2bf54008 100644
--- a/apps/web/src/components/SettingsPane.tsx
+++ b/apps/web/src/components/SettingsPane.tsx
@@ -1,5 +1,4 @@
import {
- AudioLines,
ChevronLeft,
ChevronRight,
Database,
@@ -38,7 +37,6 @@ import { PasswordCard } from "./settings/PasswordCard";
import { UserManagementCard } from "./settings/UserManagementCard";
import { ObjectStorageCard } from "./settings/ObjectStorageCard";
import { AiModelCard } from "./settings/AiModelCard";
-import { SpeechTranscriptionCard } from "./settings/SpeechTranscriptionCard";
import { DesktopAcpAgentCard } from "./settings/DesktopAcpAgentCard";
import { ThemeToggle } from "./ThemeToggle";
import type { AuthUser } from "@edgeever/shared";
@@ -75,7 +73,7 @@ const SettingsGroup = ({ children }: { children: ReactNode }) => (
</div>
);
-type TabKey = "general" | "shortcuts" | "users" | "data" | "ai" | "speech" | "mcp" | "advanced" | "account" | "system";
+type TabKey = "general" | "shortcuts" | "users" | "data" | "ai" | "mcp" | "advanced" | "account" | "system";
interface TabItem {
key: TabKey;
@@ -132,11 +130,6 @@ export const SettingsPane = ({
label: t("settings.tabs.mcp"),
icon: KeyRound,
},
- {
- key: "speech",
- label: t("settings.tabs.speech"),
- icon: AudioLines,
- },
{
key: "data",
label: t("settings.tabs.data"),
@@ -243,12 +236,6 @@ export const SettingsPane = ({
<AiModelCard />
</SettingsGroup>
);
- case "speech":
- return (
- <SettingsGroup>
- <SpeechTranscriptionCard demoMode={demoMode} />
- </SettingsGroup>
- );
case "mcp":
return (
<SettingsGroup>
diff --git a/apps/web/src/components/settings/SpeechProviderCard.tsx b/apps/web/src/components/settings/SpeechProviderCard.tsx
deleted file mode 100644
index c4be09b2..00000000
--- a/apps/web/src/components/settings/SpeechProviderCard.tsx
+++ /dev/null
@@ -1,325 +0,0 @@
-import { useEffect, useState, type FormEvent, type ReactNode } from "react";
-import { useMutation } from "@tanstack/react-query";
-import type { AiTranscriptionProvider, AiTranscriptionSettings, AiTranscriptionStandard } from "@edgeever/shared";
-import { Loader2, MoreHorizontal, Pencil, Plus, Trash2 } from "lucide-react";
-import { useTranslation } from "react-i18next";
-import { aiErrorMessage, trimAiText } from "@/components/settings/ai-provider-options";
-import { Button } from "@/components/ui/button";
-import {
- Dialog,
- DialogContent,
- DialogFooter,
- DialogHeader,
- DialogTitle,
-} from "@/components/ui/dialog";
-import {
- DropdownMenu,
- DropdownMenuContent,
- DropdownMenuItem,
- DropdownMenuSeparator,
- DropdownMenuTrigger,
-} from "@/components/ui/dropdown-menu";
-import { Input } from "@/components/ui/input";
-import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from "@/components/ui/select";
-import { Switch } from "@/components/ui/switch";
-import { Tooltip, TooltipContent, TooltipProvider, TooltipTrigger } from "@/components/ui/tooltip";
-import { api } from "@/lib/api";
-
-export const SpeechProviderCard = ({
- provider: saved,
- defaultModelId,
- readOnly,
- onChanged,
-}: {
- provider: AiTranscriptionProvider;
- defaultModelId: string | null;
- readOnly: boolean;
- onChanged: (settings: AiTranscriptionSettings) => void;
-}) => {
- const { t } = useTranslation();
- const [provider, setProvider] = useState<AiTranscriptionStandard>(saved.provider);
- const [displayName, setDisplayName] = useState(saved.displayName);
- const [baseUrl, setBaseUrl] = useState(saved.baseUrl);
- const [apiKey, setApiKey] = useState("");
- const [modelId, setModelId] = useState("");
- const [showConnection, setShowConnection] = useState(false);
- const [showAddModel, setShowAddModel] = useState(false);
-
- useEffect(() => {
- setProvider(saved.provider);
- setDisplayName(saved.displayName);
- setBaseUrl(saved.baseUrl);
- }, [saved.baseUrl, saved.displayName, saved.provider]);
-
- const saveMutation = useMutation({
- mutationFn: () => api.updateAiTranscriptionProvider(saved.id, {
- provider,
- displayName: displayName.trim(),
- baseUrl: baseUrl.trim(),
- isEnabled: saved.isEnabled,
- ...(apiKey.trim() ? { apiKey: apiKey.trim() } : {}),
- }),
- onSuccess: (settings) => {
- setApiKey("");
- setShowConnection(false);
- onChanged(settings);
- },
- });
- const toggleMutation = useMutation({
- mutationFn: (isEnabled: boolean) => api.updateAiTranscriptionProvider(saved.id, {
- provider: saved.provider,
- displayName: saved.displayName,
- baseUrl: saved.baseUrl,
- isEnabled,
- }),
- onSuccess: onChanged,
- });
- const deleteMutation = useMutation({
- mutationFn: () => api.deleteAiTranscriptionProvider(saved.id),
- onSuccess: onChanged,
- });
- const addModelMutation = useMutation({
- mutationFn: () => api.addAiTranscriptionModel(saved.id, { modelId: trimAiText(modelId) }),
- onSuccess: (settings) => {
- setModelId("");
- setShowAddModel(false);
- onChanged(settings);
- },
- });
- const deleteModelMutation = useMutation({
- mutationFn: (modelConfigId: string) => api.deleteAiTranscriptionModel(saved.id, modelConfigId),
- onSuccess: onChanged,
- });
-
- const cardBusy = toggleMutation.isPending || deleteMutation.isPending || deleteModelMutation.isPending;
- const cardError = toggleMutation.error ?? deleteMutation.error ?? addModelMutation.error ?? deleteModelMutation.error;
- const resetConnectionForm = () => {
- setProvider(saved.provider);
- setDisplayName(saved.displayName);
- setBaseUrl(saved.baseUrl);
- setApiKey("");
- saveMutation.reset();
- };
- const handleConnectionChange = (open: boolean) => {
- setShowConnection(open);
- if (!open) resetConnectionForm();
- };
- const handleAddModelChange = (open: boolean) => {
- setShowAddModel(open);
- if (!open) {
- setModelId("");
- addModelMutation.reset();
- }
- };
- const deleteProvider = () => {
- if (window.confirm(t("speechTranscription.deleteProviderConfirm", { name: saved.displayName }))) {
- deleteMutation.mutate();
- }
- };
- const deleteModel = (model: AiTranscriptionProvider["models"][number]) => {
- if (window.confirm(t("speechTranscription.deleteModelConfirm", { name: model.displayName }))) {
- deleteModelMutation.mutate(model.id);
- }
- };
-
- return (
- <section>
- <div className="flex flex-col gap-2.5 p-3.5 sm:p-4">
- <div className="flex items-center justify-between gap-3">
- <div className="flex min-w-0 flex-wrap items-center gap-2">
- <span className="truncate text-xs font-normal text-slate-900">{saved.displayName}</span>
- {saved.credentialsUnavailable ? (
- <span className="rounded-md bg-amber-50 px-2 py-0.5 text-xs font-normal text-amber-800">
- {t("aiModel.savedCredentialsUnavailableBadge")}
- </span>
- ) : null}
- <span className="max-w-full truncate font-mono text-xs text-slate-400">{formatBaseUrl(saved.baseUrl)}</span>
- <span className="max-w-full truncate text-xs text-slate-400">{t(`speechTranscription.standards.${saved.provider}`)}</span>
- </div>
- <div className="flex shrink-0 items-center gap-1.5">
- <span className="hidden text-xs text-slate-400 sm:inline">
- {saved.isEnabled ? t("speechTranscription.serviceEnabledStatus") : t("speechTranscription.serviceDisabledStatus")}
- </span>
- <Switch
- checked={saved.isEnabled}
- disabled={readOnly || toggleMutation.isPending}
- aria-label={t("speechTranscription.serviceEnabled")}
- onCheckedChange={(checked) => toggleMutation.mutate(checked)}
- />
- <TooltipProvider>
- <Tooltip>
- <TooltipTrigger asChild>
- <Button
- type="button"
- variant="ghost"
- size="icon"
- className="h-8 w-8 text-slate-500 hover:text-slate-900"
- disabled={readOnly}
- onClick={() => setShowAddModel(true)}
- aria-label={t("speechTranscription.addModel")}
- >
- <Plus className="h-4 w-4" />
- </Button>
- </TooltipTrigger>
- <TooltipContent side="bottom">{t("speechTranscription.addModel")}</TooltipContent>
- </Tooltip>
- <DropdownMenu>
- <Tooltip>
- <TooltipTrigger asChild>
- <DropdownMenuTrigger asChild>
- <Button
- type="button"
- variant="ghost"
- size="icon"
- className="h-8 w-8 text-slate-500 hover:text-slate-900"
- disabled={cardBusy}
- aria-label={t("speechTranscription.serviceActions")}
- >
- <MoreHorizontal className="h-4 w-4" />
- </Button>
- </DropdownMenuTrigger>
- </TooltipTrigger>
- <TooltipContent side="bottom">{t("speechTranscription.serviceActions")}</TooltipContent>
- </Tooltip>
- <DropdownMenuContent align="end">
- <DropdownMenuItem disabled={readOnly} onSelect={() => { resetConnectionForm(); setShowConnection(true); }}>
- <Pencil className="mr-2 h-4 w-4" />{t("speechTranscription.editConnection")}
- </DropdownMenuItem>
- <DropdownMenuSeparator />
- <DropdownMenuItem className="text-rose-600 focus:text-rose-700" disabled={readOnly} onSelect={deleteProvider}>
- <Trash2 className="mr-2 h-4 w-4" />{t("common.delete")}
- </DropdownMenuItem>
- </DropdownMenuContent>
- </DropdownMenu>
- </TooltipProvider>
- </div>
- </div>
- <div className="flex min-w-0 flex-wrap items-center gap-1.5 pt-0.5">
- <TooltipProvider>
- {saved.models.length ? saved.models.map((model) => (
- <span
- key={model.id}
- className="inline-flex h-6.5 max-w-full min-w-0 items-center gap-1.5 rounded-md border border-slate-200/80 bg-slate-50/80 pl-2 pr-1 text-xs text-slate-700"
- >
- <span className="min-w-0 truncate font-normal">{model.displayName}</span>
- {model.id === defaultModelId ? (
- <span className="shrink-0 rounded border border-slate-200 bg-slate-100 px-1 py-0.5 text-xs font-normal text-slate-700">
- {t("speechTranscription.defaultBadge")}
- </span>
- ) : null}
- <Tooltip>
- <TooltipTrigger asChild>
- <Button
- type="button"
- variant="ghost"
- size="icon"
- className="h-4 w-4 shrink-0 text-slate-400 hover:bg-card hover:text-rose-600"
- disabled={readOnly || deleteModelMutation.isPending}
- onClick={() => deleteModel(model)}
- aria-label={`${t("speechTranscription.removeModel")}: ${model.displayName}`}
- >
- <Trash2 className="h-3 w-3" />
- </Button>
- </TooltipTrigger>
- <TooltipContent side="bottom">{t("speechTranscription.removeModel")}</TooltipContent>
- </Tooltip>
- </span>
- )) : (
- <span className="text-xs text-slate-400">{t("speechTranscription.noModels")}</span>
- )}
- </TooltipProvider>
- </div>
- </div>
- {cardError ? (
- <p className="border-t px-4 py-3 text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(cardError, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : null}
- <Dialog open={showConnection} onOpenChange={handleConnectionChange}>
- <DialogContent className="max-h-[calc(100vh-2rem)] overflow-y-auto sm:max-w-lg">
- <form className="grid gap-4" onSubmit={(event: FormEvent) => { event.preventDefault(); if (readOnly) return; saveMutation.mutate(); }}>
- <DialogHeader>
- <DialogTitle className="text-xs font-normal">{t("speechTranscription.editConnection")}</DialogTitle>
- </DialogHeader>
- <div className="grid gap-4 sm:grid-cols-2">
- <Field label={t("speechTranscription.displayName")}>
- <Input className="h-9 text-xs" value={displayName} onChange={(event) => setDisplayName(event.target.value)} required maxLength={80} disabled={readOnly} />
- </Field>
- <Field label={t("speechTranscription.standard")}>
- <Select value={provider} onValueChange={(value) => setProvider(value as AiTranscriptionStandard)} disabled={readOnly}>
- <SelectTrigger className="h-9 bg-card text-xs font-normal"><SelectValue /></SelectTrigger>
- <SelectContent>
- <SelectItem value="openai-compatible">{t("speechTranscription.standards.openai-compatible")}</SelectItem>
- </SelectContent>
- </Select>
- </Field>
- </div>
- <Field label={t("speechTranscription.baseUrl")} hint={t("speechTranscription.baseUrlHint")}>
- <Input className="h-9 text-xs" value={baseUrl} onChange={(event) => setBaseUrl(event.target.value)} required inputMode="url" autoComplete="off" spellCheck={false} maxLength={500} disabled={readOnly} />
- </Field>
- <Field label={t("speechTranscription.apiToken")} hint={saved.hasApiKey ? t("speechTranscription.apiTokenSavedHint") : t("speechTranscription.apiTokenHint")}>
- <Input
- className="h-9 text-xs"
- type="password"
- value={apiKey}
- onChange={(event) => setApiKey(event.target.value)}
- placeholder={saved.hasApiKey ? t("speechTranscription.apiTokenStoredPlaceholder") : ""}
- autoComplete="new-password"
- maxLength={4096}
- disabled={readOnly}
- />
- </Field>
- {saveMutation.isError ? (
- <p className="text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(saveMutation.error, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : null}
- <DialogFooter className="gap-2 sm:space-x-0">
- <Button type="button" variant="outline" className="text-xs font-normal" onClick={() => handleConnectionChange(false)}>{t("common.cancel")}</Button>
- <Button type="submit" variant="solid" className="text-xs font-normal" disabled={readOnly || saveMutation.isPending}>
- {saveMutation.isPending ? <Loader2 className="h-4 w-4 animate-spin" /> : null}{t("common.save")}
- </Button>
- </DialogFooter>
- </form>
- </DialogContent>
- </Dialog>
- <Dialog open={showAddModel} onOpenChange={handleAddModelChange}>
- <DialogContent className="sm:max-w-md">
- <form className="grid gap-4" onSubmit={(event: FormEvent) => { event.preventDefault(); if (readOnly || !trimAiText(modelId)) return; addModelMutation.mutate(); }}>
- <DialogHeader>
- <DialogTitle className="text-xs font-normal">{t("speechTranscription.addModel")}</DialogTitle>
- </DialogHeader>
- <Field label={t("speechTranscription.modelId")}>
- <Input className="h-9 text-xs" value={modelId} onChange={(event) => setModelId(event.target.value)} placeholder={t("speechTranscription.modelIdPlaceholder")} autoComplete="off" spellCheck={false} maxLength={200} required disabled={readOnly} />
- </Field>
- {addModelMutation.isError ? (
- <p className="text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(addModelMutation.error, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : null}
- <DialogFooter className="gap-2 sm:space-x-0">
- <Button type="button" variant="outline" className="text-xs font-normal" onClick={() => handleAddModelChange(false)}>{t("common.cancel")}</Button>
- <Button type="submit" variant="solid" className="text-xs font-normal" disabled={readOnly || addModelMutation.isPending || !trimAiText(modelId)}>
- {addModelMutation.isPending ? <Loader2 className="h-4 w-4 animate-spin" /> : null}{t("speechTranscription.addModel")}
- </Button>
- </DialogFooter>
- </form>
- </DialogContent>
- </Dialog>
- </section>
- );
-};
-
-const formatBaseUrl = (baseUrl: string) => {
- try {
- return new URL(baseUrl).host;
- } catch {
- return baseUrl;
- }
-};
-
-const Field = ({ label, hint, children }: { label: string; hint?: string; children: ReactNode }) => (
- <label className="grid gap-1.5 text-xs font-normal leading-5 text-slate-700">
- {label}{children}{hint ? <span className="text-xs font-normal leading-4 text-slate-500">{hint}</span> : null}
- </label>
-);
diff --git a/apps/web/src/components/settings/SpeechTranscriptionCard.test.mjs b/apps/web/src/components/settings/SpeechTranscriptionCard.test.mjs
deleted file mode 100644
index 364986f2..00000000
--- a/apps/web/src/components/settings/SpeechTranscriptionCard.test.mjs
+++ /dev/null
@@ -1,133 +0,0 @@
-import { describe, expect, test } from "bun:test";
-import { readFileSync } from "node:fs";
-import { createElement } from "react";
-import { renderToStaticMarkup } from "react-dom/server";
-import { createInstance } from "i18next";
-import { I18nextProvider } from "react-i18next";
-import { QueryClient, QueryClientProvider } from "@tanstack/react-query";
-import { zhCN } from "../../../../../packages/shared/src/i18n/zh-CN.ts";
-import { SpeechTranscriptionCard } from "./SpeechTranscriptionCard.tsx";
-
-const saved = {
- providers: [
- {
- id: "atp_groq",
- provider: "openai-compatible",
- displayName: "Groq",
- baseUrl: "https://api.groq.com/openai/v1",
- isEnabled: true,
- hasApiKey: true,
- models: [{
- id: "atm_turbo",
- providerId: "atp_groq",
- modelId: "whisper-large-v3-turbo",
- displayName: "whisper-large-v3-turbo",
- }],
- },
- {
- id: "atp_local",
- provider: "openai-compatible",
- displayName: "本机",
- baseUrl: "http://127.0.0.1:8080/v1",
- isEnabled: true,
- hasApiKey: true,
- models: [{
- id: "atm_local",
- providerId: "atp_local",
- modelId: "local-whisper",
- displayName: "local-whisper",
- }],
- },
- ],
- defaultModelId: "atm_turbo",
- enabled: true,
- encryptionConfigured: true,
- readOnly: false,
-};
-
-const renderCard = async (settings, demoMode = false) => {
- const i18n = createInstance();
- await i18n.init({ lng: "zh-CN", resources: { "zh-CN": { translation: zhCN } } });
- const queryClient = new QueryClient({
- defaultOptions: { queries: { retry: false, staleTime: Infinity } },
- });
- queryClient.setQueryData(["ai-transcription-settings"], settings);
- return renderToStaticMarkup(createElement(
- QueryClientProvider,
- { client: queryClient },
- createElement(I18nextProvider, { i18n }, createElement(SpeechTranscriptionCard, { demoMode })),
- ));
-};
-
-describe("speech transcription settings card", () => {
- test("lists every saved speech service and the one used for extraction", async () => {
- const html = await renderCard(saved);
- expect(html).toContain("语音识别模型配置");
- expect(html).not.toContain("语音识别服务当前主要用于视频字幕识别");
- expect(html).toContain("默认语音模型");
- expect(html).toContain("添加模型服务");
- expect(html).toContain("Groq");
- expect(html).toContain("本机");
- expect(html).toContain("api.groq.com");
- expect(html).toContain("127.0.0.1:8080");
- expect(html).toContain("whisper-large-v3-turbo");
- expect(html).toContain("local-whisper");
- expect(html).toContain("默认");
- expect(html).toContain("OpenAI 兼容标准");
- expect(html).not.toContain("asr-token");
- expect(html).not.toContain("title=");
- });
-
- test("shows an empty list when no speech service is saved", async () => {
- const html = await renderCard({
- providers: [],
- defaultModelId: null,
- enabled: false,
- encryptionConfigured: true,
- readOnly: false,
- });
- expect(html).toContain("还没有语音识别服务");
- expect(html).toContain("默认语音模型");
- expect(html).toContain("yt-dlp 复用的浏览器登录状态");
- expect(html).not.toContain("提取字幕下载音频时");
- expect(html).toContain("仅在本地用于读取当前浏览器已登录的媒体流以获取音轨与字幕");
- expect(html).toContain("使用规范与版权说明");
- expect(html).toContain("字幕提取与音视频转录功能仅供个人学习、离线研读及辅助记录使用");
- expect(html).toContain("音视频素材及其原始台词、字幕等衍生内容的知识产权均归原作者及发布平台所有");
- expect(html).toContain("相关解析与下载逻辑完全在您的本地设备运行");
- expect(html).toContain("yt-dlp 版本");
- expect(html).toContain("尚未下载");
- expect(html).toContain("yt-dlp 路径");
- expect(html).toContain("未知");
- });
-
- test("is its own settings page under models and agents", () => {
- const pane = readFileSync(new URL("../SettingsPane.tsx", import.meta.url), "utf8");
- const card = readFileSync(new URL("./SpeechTranscriptionCard.tsx", import.meta.url), "utf8");
- const provider = readFileSync(new URL("./SpeechProviderCard.tsx", import.meta.url), "utf8");
- const aiTab = pane.slice(pane.indexOf('case "ai"'), pane.indexOf('case "speech"'));
- const speechTab = pane.slice(pane.indexOf('case "speech"'), pane.indexOf('case "mcp"'));
- expect(aiTab).toContain("<AiModelCard />");
- expect(aiTab).not.toContain("SpeechTranscriptionCard");
- expect(speechTab).toContain("<SpeechTranscriptionCard demoMode={demoMode} />");
- expect(pane.indexOf('key: "ai"')).toBeLessThan(pane.indexOf('key: "mcp"'));
- expect(pane.indexOf('key: "mcp"')).toBeLessThan(pane.indexOf('key: "speech"'));
- expect(card).toContain("createAiTranscriptionProvider");
- expect(card).not.toContain("speechTranscription.description");
- expect(card).toContain("speechTranscription.cookieBrowser");
- expect(card).not.toContain("cookieBrowserHint");
- const localToolStart = card.lastIndexOf("rounded-lg border border-slate-200 bg-card divide-y", card.indexOf('id="video-cookie-browser"'));
- const localToolEnd = card.indexOf('t("systemInfo.ytDlpPath")');
- expect(localToolStart).toBeGreaterThan(-1);
- expect(localToolStart).toBeLessThan(localToolEnd);
- expect(card.slice(localToolStart, localToolEnd)).toContain('t("systemInfo.ytDlpVersion")');
- expect(card).toContain("speechTranscription.standard");
- expect(card).toContain('value="openai-compatible"');
- expect(provider).toContain("apiKey.trim() ? { apiKey: apiKey.trim() } : {}");
- expect(card).not.toContain("updateAiTranscriptionSettings");
- expect(provider).not.toContain("@ai-sdk/openai");
- expect(provider).not.toContain("transcribe(");
- expect(card).not.toContain("title=");
- expect(provider).not.toContain("title=");
- });
-});
diff --git a/apps/web/src/components/settings/SpeechTranscriptionCard.tsx b/apps/web/src/components/settings/SpeechTranscriptionCard.tsx
deleted file mode 100644
index 1655d908..00000000
--- a/apps/web/src/components/settings/SpeechTranscriptionCard.tsx
+++ /dev/null
@@ -1,410 +0,0 @@
-import { useState, type FormEvent, type ReactNode } from "react";
-import { useMutation, useQuery, useQueryClient } from "@tanstack/react-query";
-import { VIDEO_COOKIE_BROWSERS, type AiTranscriptionSettings, type AiTranscriptionStandard, type VideoCookieBrowser } from "@edgeever/shared";
-import { AudioLines, Loader2, Plus, ShieldCheck, TriangleAlert } from "lucide-react";
-import { useTranslation } from "react-i18next";
-import { aiErrorMessage } from "@/components/settings/ai-provider-options";
-import { SpeechProviderCard } from "@/components/settings/SpeechProviderCard";
-import { Button } from "@/components/ui/button";
-import { Card, CardContent, CardHeader, CardTitle } from "@/components/ui/card";
-import {
- Dialog,
- DialogContent,
- DialogFooter,
- DialogHeader,
- DialogTitle,
-} from "@/components/ui/dialog";
-import { Input } from "@/components/ui/input";
-import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from "@/components/ui/select";
-import { Tooltip, TooltipContent, TooltipProvider, TooltipTrigger } from "@/components/ui/tooltip";
-import { api } from "@/lib/api";
-import {
- SETTINGS_CARD_HEADER_CLASSNAME,
- SETTINGS_CARD_ICON_CLASSNAME,
- SETTINGS_CARD_TITLE_CLASSNAME,
- SETTINGS_ITEM_TITLE_CLASSNAME,
-} from "./settings-ui";
-
-const emptyDraft = {
- provider: "openai-compatible" as AiTranscriptionStandard,
- displayName: "",
- baseUrl: "",
- apiKey: "",
- initialModelId: "",
-};
-
-type YtDlpStatus = {
- state: "idle" | "checking" | "downloading" | "ready" | "failed";
- version: string | null;
- path: string;
- errorCode: string | null;
- httpStatus: number | null;
-};
-
-const ytDlpReason = (
- t: (key: string, options?: Record<string, string>) => string,
- status: YtDlpStatus,
-) => {
- if (!status.errorCode) return "";
- if (status.errorCode === "http") {
- return t("systemInfo.ytDlpErrors.http", { status: String(status.httpStatus ?? "") });
- }
- const key = `systemInfo.ytDlpErrors.${status.errorCode}`;
- const translated = t(key);
- return translated === key ? t("systemInfo.ytDlpFailed") : translated;
-};
-
-const ytDlpVersionValue = (
- t: (key: string, options?: Record<string, string>) => string,
- status: YtDlpStatus | undefined,
-) => {
- if (!status || status.state === "idle" || status.state === "checking") {
- return status?.version || t("systemInfo.ytDlpMissing");
- }
- const reason = ytDlpReason(t, status);
- if (status.state === "downloading") {
- return status.version
- ? t("systemInfo.ytDlpUpdating", { version: status.version })
- : t("systemInfo.ytDlpDownloading");
- }
- if (status.state === "failed") return reason || t("systemInfo.ytDlpFailed");
- if (status.errorCode && status.version) {
- return t("systemInfo.ytDlpUpdateFailed", {
- version: status.version,
- reason: reason || t("systemInfo.ytDlpFailed"),
- });
- }
- return status.version || t("systemInfo.ytDlpMissing");
-};
-
-export const SpeechTranscriptionCard = ({ demoMode }: { demoMode: boolean }) => {
- const { t } = useTranslation();
- const queryClient = useQueryClient();
- const desktop = typeof window !== "undefined" && window.edgeeverDesktop?.isAvailable === true;
- const settingsQuery = useQuery({
- queryKey: ["ai-transcription-settings"],
- queryFn: api.getAiTranscriptionSettings,
- });
- const cookieBrowserQuery = useQuery({
- queryKey: ["video-cookie-browser"],
- queryFn: () => window.edgeeverDesktop!.videoCookieBrowser(),
- enabled: desktop,
- });
- const ytDlpStatusQuery = useQuery({
- queryKey: ["desktop-yt-dlp-status"],
- queryFn: () => window.edgeeverDesktop!.ytDlpStatus(),
- enabled: desktop,
- refetchInterval: 30_000,
- retry: 1,
- });
- const [showAdd, setShowAdd] = useState(false);
- const [draft, setDraft] = useState(emptyDraft);
- const [previewBrowser, setPreviewBrowser] = useState<VideoCookieBrowser | null>(null);
-
- const settings = settingsQuery.data;
- const providers = settings?.providers ?? [];
- const readOnly = demoMode || Boolean(settings?.readOnly);
- const encryptionConfigured = settings?.encryptionConfigured ?? false;
- const canAdd = !readOnly && encryptionConfigured;
- const hasUnavailableCredentials = providers.some((provider) => provider.credentialsUnavailable);
- const allModels = providers.flatMap((provider) =>
- provider.models.map((model) => ({ ...model, providerName: provider.displayName, providerEnabled: provider.isEnabled })));
- const defaultModelAvailable = !settings?.defaultModelId
- || allModels.some((model) => model.id === settings.defaultModelId && model.providerEnabled);
- const applySettings = (saved: AiTranscriptionSettings) => {
- queryClient.setQueryData(["ai-transcription-settings"], saved);
- };
- const createMutation = useMutation({
- mutationFn: () => api.createAiTranscriptionProvider({
- provider: draft.provider,
- displayName: draft.displayName.trim(),
- baseUrl: draft.baseUrl.trim(),
- apiKey: draft.apiKey.trim(),
- isEnabled: true,
- ...(draft.initialModelId.trim() ? { initialModelId: draft.initialModelId.trim() } : {}),
- }),
- onSuccess: (saved) => {
- applySettings(saved);
- setShowAdd(false);
- setDraft(emptyDraft);
- },
- });
- const defaultMutation = useMutation({
- mutationFn: api.updateDefaultAiTranscriptionModel,
- onSuccess: applySettings,
- });
- const cookieBrowserMutation = useMutation({
- mutationFn: (browser: VideoCookieBrowser) => window.edgeeverDesktop!.setVideoCookieBrowser(browser),
- onSuccess: (saved) => {
- queryClient.setQueryData(["video-cookie-browser"], saved.browser);
- },
- });
- const browserNotice = desktop
- ? (cookieBrowserMutation.isSuccess ? cookieBrowserMutation.data.browser : null)
- : previewBrowser;
- const ytDlpStatus = ytDlpStatusQuery.data;
- const ytDlpFailed = ytDlpStatusQuery.isError || ytDlpStatus?.state === "failed" || (ytDlpStatus?.state === "ready" && Boolean(ytDlpStatus.errorCode));
- const ytDlpVersion = ytDlpStatusQuery.isError ? t("systemInfo.ytDlpFailed") : ytDlpVersionValue(t, ytDlpStatus);
- const ytDlpPath = ytDlpStatus?.path || t("systemInfo.unknown");
- const addDisabledReason = readOnly
- ? t("speechTranscription.demoDisabled")
- : !encryptionConfigured
- ? t("aiModel.encryptionKeyMissing")
- : undefined;
- const handleAddDialogChange = (open: boolean) => {
- setShowAdd(open);
- if (!open) {
- setDraft(emptyDraft);
- createMutation.reset();
- }
- };
-
- return (
- <Card className="w-full min-w-0 overflow-hidden shadow-none">
- <CardHeader className={SETTINGS_CARD_HEADER_CLASSNAME}>
- <CardTitle className={SETTINGS_CARD_TITLE_CLASSNAME}>
- <AudioLines className={SETTINGS_CARD_ICON_CLASSNAME} />
- {t("speechTranscription.title")}
- </CardTitle>
- </CardHeader>
- <CardContent className="grid gap-5 p-4 pt-0 sm:px-5 sm:pb-5">
- {settingsQuery.isLoading ? (
- <p className="flex items-center gap-2 text-xs leading-5 text-slate-500">
- <Loader2 className="h-4 w-4 animate-spin" />
- {t("common.loading")}
- </p>
- ) : settingsQuery.isError ? (
- <p className="text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(settingsQuery.error, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : (
- <>
- {!encryptionConfigured ? (
- <p className="flex items-start gap-2 rounded-lg border border-amber-200 bg-amber-50 p-3 text-xs leading-5 text-amber-800">
- <TriangleAlert className="mt-0.5 h-4 w-4 shrink-0" />
- {t("aiModel.encryptionKeyMissing")}
- </p>
- ) : null}
- {hasUnavailableCredentials ? (
- <p className="flex items-start gap-2 rounded-lg border border-amber-200 bg-amber-50 p-3 text-xs leading-5 text-amber-800">
- <TriangleAlert className="mt-0.5 h-4 w-4 shrink-0" />
- {t("speechTranscription.savedCredentialsUnavailable")}
- </p>
- ) : null}
- <div className="space-y-2">
- <div className="flex items-center justify-between gap-3">
- <div className={SETTINGS_ITEM_TITLE_CLASSNAME}>{t("speechTranscription.defaultModel")}</div>
- <div className="w-56 max-w-[60%] shrink-0 sm:w-72">
- <Select
- value={settings?.defaultModelId ?? "none"}
- onValueChange={(value) => defaultMutation.mutate(value === "none" ? null : value)}
- disabled={readOnly || defaultMutation.isPending}
- >
- <SelectTrigger className="h-8 bg-card text-xs"><SelectValue /></SelectTrigger>
- <SelectContent>
- <SelectItem value="none">{t("speechTranscription.noDefaultModel")}</SelectItem>
- {allModels.map((model) => (
- <SelectItem key={model.id} value={model.id} disabled={!model.providerEnabled}>
- {model.displayName} · {model.providerName}
- </SelectItem>
- ))}
- </SelectContent>
- </Select>
- </div>
- </div>
- {!defaultModelAvailable ? (
- <p className="flex items-center gap-1.5 text-xs text-amber-700">
- <TriangleAlert className="h-3.5 w-3.5 shrink-0" />
- {t("speechTranscription.defaultUnavailable")}
- </p>
- ) : null}
- {defaultMutation.isError ? (
- <p className="text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(defaultMutation.error, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : null}
- </div>
- <section className="grid gap-3">
- <div className="flex items-center justify-end">
- <DisabledActionTooltip label={!canAdd ? addDisabledReason : undefined}>
- <Button type="button" variant="outline" size="sm" className="h-8 gap-1.5 bg-card text-xs font-normal" disabled={!canAdd} onClick={() => handleAddDialogChange(true)}>
- <Plus className="h-3.5 w-3.5" />{t("speechTranscription.addProvider")}
- </Button>
- </DisabledActionTooltip>
- </div>
- {providers.length ? (
- <div className="overflow-hidden rounded-lg border border-slate-200 divide-y divide-slate-100 bg-card">
- {providers.map((provider) => (
- <SpeechProviderCard
- key={provider.id}
- provider={provider}
- defaultModelId={settings?.defaultModelId ?? null}
- readOnly={readOnly}
- onChanged={applySettings}
- />
- ))}
- </div>
- ) : (
- <p className="rounded-lg border border-dashed border-slate-200 p-6 text-center text-xs text-slate-400">{t("speechTranscription.noProviders")}</p>
- )}
- </section>
- {readOnly ? <p className="text-xs leading-5 text-slate-500">{t("speechTranscription.demoDisabled")}</p> : null}
- <section className="overflow-hidden rounded-lg border border-slate-200 bg-card divide-y divide-slate-100">
- <div className="grid gap-2 px-4 py-3">
- <div className="flex items-center justify-between gap-3">
- <label className={`${SETTINGS_ITEM_TITLE_CLASSNAME} min-w-0`} htmlFor="video-cookie-browser">
- {t("speechTranscription.cookieBrowser")}
- </label>
- <div className="w-56 max-w-[60%] shrink-0 sm:w-72">
- <Select
- value={(desktop ? cookieBrowserQuery.data : previewBrowser) ?? "chrome"}
- onValueChange={(value) => {
- const browser = value as VideoCookieBrowser;
- if (desktop) cookieBrowserMutation.mutate(browser);
- else setPreviewBrowser(browser);
- }}
- disabled={desktop && (cookieBrowserQuery.isLoading || cookieBrowserMutation.isPending)}
- >
- <SelectTrigger id="video-cookie-browser" className="h-8 bg-card text-xs">
- <SelectValue />
- </SelectTrigger>
- <SelectContent>
- {VIDEO_COOKIE_BROWSERS.map((browser) => (
- <SelectItem key={browser} value={browser}>
- {t(`speechTranscription.cookieBrowsers.${browser}`)}
- </SelectItem>
- ))}
- <SelectItem value="none">{t("speechTranscription.cookieBrowserNone")}</SelectItem>
- </SelectContent>
- </Select>
- </div>
- </div>
- <p className="text-xs leading-4 text-slate-500">
- {t("speechTranscription.cookieBrowserNotice")}
- </p>
- {browserNotice ? (
- <p className="text-xs font-medium text-emerald-700">
- {browserNotice === "none"
- ? t("speechTranscription.cookieBrowserOff")
- : t("speechTranscription.cookieBrowserApplied", {
- browser: t(`speechTranscription.cookieBrowsers.${browserNotice}`),
- })}
- </p>
- ) : null}
- {desktop && (cookieBrowserQuery.isError || cookieBrowserMutation.isError) ? (
- <p className="text-xs font-medium text-rose-600" role="alert">{t("speechTranscription.cookieBrowserFailed")}</p>
- ) : null}
- </div>
- <div className="flex items-center justify-between gap-3 px-4 py-3">
- <p className={SETTINGS_ITEM_TITLE_CLASSNAME}>{t("systemInfo.ytDlpVersion")}</p>
- <p className={`text-right text-xs font-medium leading-5 ${ytDlpFailed ? "text-rose-600" : "text-slate-900"}`}>{ytDlpVersion}</p>
- </div>
- <div className="flex items-start justify-between gap-3 px-4 py-3">
- <p className={SETTINGS_ITEM_TITLE_CLASSNAME}>{t("systemInfo.ytDlpPath")}</p>
- <p className="max-w-[60%] break-all text-right font-mono text-xs leading-5 text-slate-900">{ytDlpPath}</p>
- </div>
- </section>
- <section className="rounded-lg border border-slate-200 bg-slate-50/60 p-4 text-xs leading-5 text-slate-600">
- <div className="flex items-center gap-1.5 font-medium text-slate-800">
- <ShieldCheck className="h-4 w-4 text-emerald-600 shrink-0" />
- <span>{t("speechTranscription.complianceNoticeTitle")}</span>
- </div>
- <ul className="mt-2 list-disc space-y-1 pl-4 text-slate-500">
- <li>{t("speechTranscription.complianceNoticePersonalUse")}</li>
- <li>{t("speechTranscription.complianceNoticeCopyright")}</li>
- <li>{t("speechTranscription.complianceNoticeLocalSecurity")}</li>
- </ul>
- </section>
- <Dialog open={showAdd} onOpenChange={handleAddDialogChange}>
- <DialogContent className="max-h-[calc(100vh-2rem)] overflow-y-auto sm:max-w-lg">
- <form
- className="grid gap-4"
- onSubmit={(event: FormEvent) => {
- event.preventDefault();
- if (!canAdd || createMutation.isPending) return;
- createMutation.mutate();
- }}
- >
- <DialogHeader>
- <DialogTitle className="text-xs font-normal">{t("speechTranscription.addProvider")}</DialogTitle>
- </DialogHeader>
- <div className="grid gap-4 sm:grid-cols-2">
- <Field label={t("speechTranscription.displayName")}>
- <Input className="h-9 text-xs" value={draft.displayName} onChange={(event) => setDraft({ ...draft, displayName: event.target.value })} required maxLength={80} />
- </Field>
- <SpeechStandardField
- value={draft.provider}
- onChange={(provider) => setDraft({ ...draft, provider })}
- />
- </div>
- <Field label={t("speechTranscription.baseUrl")} hint={t("speechTranscription.baseUrlHint")}>
- <Input className="h-9 text-xs" value={draft.baseUrl} onChange={(event) => setDraft({ ...draft, baseUrl: event.target.value })} placeholder={t("speechTranscription.baseUrlPlaceholder")} required inputMode="url" autoComplete="off" spellCheck={false} maxLength={500} />
- </Field>
- <Field label={t("speechTranscription.modelId")}>
- <Input className="h-9 text-xs" value={draft.initialModelId} onChange={(event) => setDraft({ ...draft, initialModelId: event.target.value })} placeholder={t("speechTranscription.modelIdPlaceholder")} required autoComplete="off" spellCheck={false} maxLength={200} />
- </Field>
- <Field label={t("speechTranscription.apiToken")} hint={t("speechTranscription.apiTokenHint")}>
- <Input className="h-9 text-xs" type="password" value={draft.apiKey} onChange={(event) => setDraft({ ...draft, apiKey: event.target.value })} required autoComplete="new-password" maxLength={4096} />
- </Field>
- {createMutation.isError ? (
- <p className="text-xs font-medium text-rose-600" role="alert">
- {aiErrorMessage(createMutation.error, t("speechTranscription.failed"), t("aiModel.encryptionKeyMissing"), t("speechTranscription.savedCredentialsUnavailable"))}
- </p>
- ) : null}
- <DialogFooter className="gap-2 sm:space-x-0">
- <Button type="button" variant="outline" className="text-xs font-normal" onClick={() => handleAddDialogChange(false)}>{t("common.cancel")}</Button>
- <Button type="submit" variant="solid" className="text-xs font-normal" disabled={!canAdd || createMutation.isPending}>
- {createMutation.isPending ? <Loader2 className="h-4 w-4 animate-spin" /> : null}{t("speechTranscription.addProvider")}
- </Button>
- </DialogFooter>
- </form>
- </DialogContent>
- </Dialog>
- </>
- )}
- </CardContent>
- </Card>
- );
-};
-
-const SpeechStandardField = ({
- value,
- onChange,
- disabled = false,
-}: {
- value: AiTranscriptionStandard;
- onChange: (value: AiTranscriptionStandard) => void;
- disabled?: boolean;
-}) => {
- const { t } = useTranslation();
- return (
- <Field label={t("speechTranscription.standard")}>
- <Select value={value} onValueChange={(next) => onChange(next as AiTranscriptionStandard)} disabled={disabled}>
- <SelectTrigger className="h-9 bg-card text-xs font-normal"><SelectValue /></SelectTrigger>
- <SelectContent>
- <SelectItem value="openai-compatible">{t("speechTranscription.standards.openai-compatible")}</SelectItem>
- </SelectContent>
- </Select>
- </Field>
- );
-};
-
-const Field = ({ label, hint, children }: { label: string; hint?: string; children: ReactNode }) => (
- <label className="grid gap-1.5 text-xs font-normal leading-5 text-slate-700">
- {label}{children}{hint ? <span className="text-xs font-normal leading-4 text-slate-500">{hint}</span> : null}
- </label>
-);
-
-const DisabledActionTooltip = ({ label, children }: { label?: string; children: ReactNode }) => {
- if (!label) return children;
- return (
- <TooltipProvider delayDuration={0} skipDelayDuration={0}>
- <Tooltip>
- <TooltipTrigger asChild>
- <span className="inline-flex" tabIndex={0}>{children}</span>
- </TooltipTrigger>
- <TooltipContent side="bottom">{label}</TooltipContent>
- </Tooltip>
- </TooltipProvider>
- );
-};
diff --git a/apps/web/src/components/settings/SystemInfoPanel.test.mjs b/apps/web/src/components/settings/SystemInfoPanel.test.mjs
index 1a1230de..374782be 100644
--- a/apps/web/src/components/settings/SystemInfoPanel.test.mjs
+++ b/apps/web/src/components/settings/SystemInfoPanel.test.mjs
@@ -37,19 +37,6 @@ describe("system information diagnostic fields", () => {
expect(feedbackSource).toContain("clientRuntime: clientRuntimeQuery.data");
});
- test("leaves the yt-dlp version and path on the speech settings page", () => {
- const speech = readFileSync(new URL("./SpeechTranscriptionCard.tsx", import.meta.url), "utf8");
- expect(source).not.toContain("ytDlpStatus");
- expect(source).not.toContain('t("systemInfo.ytDlpVersion")');
- expect(speech).toContain('t("systemInfo.ytDlpVersion")');
- expect(speech).toContain('t("systemInfo.ytDlpPath")');
- expect(speech).toContain("ytDlpStatus()");
- expect(speech).toContain("ytDlpVersionValue");
- expect(speech).toContain('t("systemInfo.ytDlpMissing")');
- expect(speech).toContain("refetchInterval: 30_000");
- expect(speech).not.toContain('ytDlpStatus?.state === "downloading"');
- });
-
test("includes current screen resolution", () => {
expect(source).toContain('t("systemInfo.screenResolution")');
expect(source).toContain('t("systemInfo.screenResolutionValue", parts)');
diff --git a/apps/web/src/lib/local-mirror.ts b/apps/web/src/lib/local-mirror.ts
index d3e98408..6bff331c 100644
--- a/apps/web/src/lib/local-mirror.ts
+++ b/apps/web/src/lib/local-mirror.ts
@@ -9,7 +9,6 @@ import {
getDiagramSummary,
getInfographicSummary,
getTableSummary,
- getVideoNoteSummary,
getNotebookDescendantIds,
resolveMemoContentDoc,
resolveMergedMemoTitle,
@@ -82,7 +81,6 @@ const toSummary = (memo: MemoDetail): MemoSummary => ({
...getDiagramSummary(memo.contentMarkdown),
...getInfographicSummary(memo.contentMarkdown),
...getTableSummary(memo.contentMarkdown),
- ...getVideoNoteSummary(memo.contentMarkdown),
tags: memo.tags,
isPinned: memo.isPinned,
isArchived: memo.isArchived,
diff --git a/apps/web/src/vite-env.d.ts b/apps/web/src/vite-env.d.ts
index adbfb2a3..167b9093 100644
--- a/apps/web/src/vite-env.d.ts
+++ b/apps/web/src/vite-env.d.ts
@@ -60,15 +60,6 @@ interface EdgeEverDesktopBridge {
chrome: string;
dataDir: string;
}>;
- ytDlpStatus(): Promise<{
- state: "idle" | "checking" | "downloading" | "ready" | "failed";
- version: string | null;
- path: string;
- errorCode: string | null;
- httpStatus: number | null;
- }>;
- videoCookieBrowser(): Promise<import("@edgeever/shared").VideoCookieBrowser>;
- setVideoCookieBrowser(browser: import("@edgeever/shared").VideoCookieBrowser): Promise<{ browser: import("@edgeever/shared").VideoCookieBrowser }>;
setAccountScope(accountId: string | null): Promise<{ ready: true; scope: string }>;
updateStatus(): Promise<DesktopUpdateStatus>;
checkUpdate(): Promise<DesktopUpdateStatus>;
diff --git a/bun.lock b/bun.lock
index 208b5ef7..f07094f8 100644
--- a/bun.lock
+++ b/bun.lock
@@ -105,9 +105,7 @@
"version": "1.100.0",
"dependencies": {
"@agentclientprotocol/sdk": "^1.5.1",
- "@ai-sdk/openai": "4.0.36",
"@edgeever/public-network": "workspace:*",
- "ai": "7.0.58",
"croner": "^10.0.1",
"electron-updater": "^6.6.2",
"fflate": "0.8.3",
@@ -334,8 +332,6 @@
"@ai-sdk/google": ["@ai-sdk/[email protected]", "", { "dependencies": { "@ai-sdk/provider": "4.0.7", "@ai-sdk/provider-utils": "5.0.25" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-+mRx7UBZn9PkJ4J6YXowaRZKMZYa290cknVqqOw/roZaDg186IUOLn9JHNQkvgaj91/MLW1AZNESy1ZD2yXDCg=="],
- "@ai-sdk/openai": ["@ai-sdk/[email protected]", "", { "dependencies": { "@ai-sdk/provider": "4.0.7", "@ai-sdk/provider-utils": "5.0.25" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-wHJNArBdjJPXb8GXcA+FslbRt+7qIE1KoHSAVi+CeBFidinVrTVBZKFMJz3mNZMj8YILBMl/VIvTD/oGFjX6/g=="],
-
"@ai-sdk/openai-compatible": ["@ai-sdk/[email protected]", "", { "dependencies": { "@ai-sdk/provider": "4.0.7", "@ai-sdk/provider-utils": "5.0.25" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-icGSoBDbFKZotUwehg7h/JjtD7arweeKfeDcsT9UsPY+YdwiT94JciD59bqAOAdNA3wlS+klW2vsaVLxTUUhbA=="],
"@ai-sdk/provider": ["@ai-sdk/[email protected]", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-6or44XprPzKbr8zkmzosowSE0pxkvJcoojBL+mCZvPUt3kvXp3XSNqeVun9golb1acEfSo6yaEBRT18h2VU+1Q=="],
diff --git a/crates/desktop-sidecar/src/note_summary.rs b/crates/desktop-sidecar/src/note_summary.rs
index 832b1906..237a8aa3 100644
--- a/crates/desktop-sidecar/src/note_summary.rs
+++ b/crates/desktop-sidecar/src/note_summary.rs
@@ -49,7 +49,6 @@ pub(crate) fn note_list_metadata(markdown: &str) -> Value {
}
}
metadata.insert("infographic".to_owned(), json!(is_infographic(markdown)));
- metadata.insert("videoNote".to_owned(), json!(is_video_note(markdown)));
match table_preview(markdown) {
Some(preview) => {
metadata.insert("structuredTable".to_owned(), json!(true));
@@ -69,10 +68,6 @@ pub(crate) fn note_list_metadata(markdown: &str) -> Value {
Value::Object(metadata)
}
-fn is_video_note(markdown: &str) -> bool {
- markdown.contains("<!-- edgeever-video-v1:")
-}
-
fn is_infographic(markdown: &str) -> bool {
let Some(payload) = infographic_payload(markdown) else {
return false;
@@ -583,20 +578,10 @@ mod tests {
assert!(metadata.get("diagramPreview").is_none(), "{markdown}");
assert_eq!(metadata["infographic"], false, "{markdown}");
assert_eq!(metadata["structuredTable"], false, "{markdown}");
- assert_eq!(metadata["videoNote"], false, "{markdown}");
assert!(metadata.get("tablePreview").is_none(), "{markdown}");
}
}
- #[test]
- fn video_note_marker_is_recognized_without_reading_the_prose() {
- assert_eq!(
- note_list_metadata("这一集没有可用字幕\n\n<!-- edgeever-video-v1:abc -->")["videoNote"],
- true
- );
- assert_eq!(note_list_metadata("ordinary note")["videoNote"], false);
- }
-
#[test]
fn infographic_marker_from_js_base64_is_recognized_with_history() {
assert_eq!(
diff --git a/docs/video-clip-spec.md b/docs/video-clip-spec.md
new file mode 100644
index 00000000..f4526d13
--- /dev/null
+++ b/docs/video-clip-spec.md
@@ -0,0 +1,24 @@
+# Video notes: save captions already available on the page
+
+On a YouTube or Bilibili watch page, the user chooses **Save video note to EdgeEver**. The extension reads the current video's metadata and any captions already provided by the platform. It saves the title, creator, source link, available cover, and captions with links back to the original timestamps. When captions exist and the workspace has an available chat model, it may also generate a summary, outline, and takeaways; failure does not block saving the note.
+
+Supported pages are ordinary YouTube videos and Shorts, and ordinary Bilibili videos including multipart videos. Live streams, series, courses, and other pages do not create video notes. Without captions, the note contains source information and a **No captions available for this video** message. It does not start speech transcription. Video notes have no command to extract captions again and do not download videos or audio tracks from external platforms.
+
+## Withdrawn speech transcription design: technical notes
+
+The following records the important design choices of an unreleased proposal for possible future evaluation of authorized use. **It does not describe current product behavior or a planned feature.** The withdrawn scope includes external audio retrieval, browser session reuse, automatic yt-dlp installation and updates, video transcription jobs, speech model settings, and desktop polling.
+
+### Jobs and data boundaries
+
+- The proposed design encoded the platform, video ID, source URL, duration, and missing-caption placeholder in a hidden marker in the note body. Saving a note did not create a job; a user action in the note list created one.
+- The API scoped jobs by workspace and note ID and recorded a content hash. Clipper tokens and demo instances could not create or claim jobs. A conditional update let one desktop client claim a job; a job stalled for roughly 20 minutes could be reclaimed.
+- Writing the result back to the same note required the expected revision and content hash. A user edit prevented an overwrite. The placeholder would be replaced when present; otherwise the transcript would be appended. Failure left the original note intact and did not trigger repeated automatic retries.
+
+### Local media processing and model calls
+
+- The proposed design checked an HTTPS source against a platform allowlist, then used yt-dlp locally to select only the best audio track, without saving video. It limited known duration to roughly 30 minutes, audio to roughly 24 MiB, and both download and transcription time. Temporary audio was deleted after the job. Resource limits do not grant copyright permission.
+- The standalone yt-dlp executable came from its releases and was checked against SHA-256; the desktop client checked for updates in the background. Optional `--cookies-from-browser` reused a local browser session, with the browser choice stored only on that device. This widened the range of accessible media; local execution did not establish permission from a platform or rights holder.
+- A workspace could configure speech providers, models, and a default model. API tokens were stored encrypted on the instance and not returned by settings reads. An authenticated desktop client obtained credentials when running a job and sent audio directly to the user's chosen OpenAI-compatible `/audio/transcriptions` endpoint. Audio bypassed the note server but reached the model service, which would require clear disclosure.
+- Segments with timestamps became transcript links back to the source; a plain-text-only response became a text block. Failures recorded error codes without writing API tokens to diagnostics.
+
+Any future revival should first establish content rights, platform terms, target markets, and the transcription provider's data handling, then redesign the entry point and verification scope. The retrieval path above should not simply be re-enabled.
diff --git a/docs/video-clip-spec.zh-CN.md b/docs/video-clip-spec.zh-CN.md
index 4282b2b2..3bd65110 100644
--- a/docs/video-clip-spec.zh-CN.md
+++ b/docs/video-clip-spec.zh-CN.md
@@ -1,60 +1,24 @@
-# 视频笔记:从保存到字幕
+# 视频笔记:保存网页已有字幕
-用户在 YouTube 或哔哩哔哩的播放页点一次「保存视频笔记到 EdgeEver」。笔记马上出现。字幕从哪来,取决于这一页当时有没有平台字幕。
+用户在 YouTube 或哔哩哔哩的播放页选择「保存视频笔记到 EdgeEver」。插件读取当前页面的视频信息和平台已经提供的字幕,保存标题、作者、来源链接、可用的封面和带原片时间戳的字幕。已有字幕且工作区配置了可用的聊天模型时,可以生成总结、大纲和要点;失败不阻止笔记保存。
-只认这两类页面:YouTube 的普通视频和 Shorts,哔哩哔哩的普通视频(分 P 以地址里的 `p` 为准,没有就存第 1 P)。直播、番剧、课程和其他站点不创建笔记。
+支持 YouTube 普通视频及 Shorts、哔哩哔哩普通视频及分 P。直播、番剧、课程和其他页面不创建视频笔记。页面没有字幕时,仅保存来源信息和「这一集没有可用字幕」,不启动语音识别。视频笔记不提供再次提取字幕的菜单,也不下载外部平台的视频或音轨。
-## 保存的那一刻
+## 已撤回的语音识别方案:技术备忘
-插件读用户正在看的这一集:标题、作者、来源、封面,以及页面里已经有的字幕。读不到这一集时提示「没有读到这个视频」,不写空笔记。
+以下记录未上线方案的关键设计,供以后评估授权场景时参考,**不是当前产品行为或待实现需求**。决定撤回的范围包括外部音轨抓取、浏览器登录状态复用、yt-dlp 自动安装与更新、视频识别任务、语音模型配置及桌面转录轮询。
-笔记先保存。用户不用等总结,也不用等后面的语音识别。桌面端没开着也一样。
+### 任务与数据边界
-保存下来的视频笔记和普通笔记不一样。在笔记列表里右键它,菜单里有「提取字幕」。普通笔记没有这一项。
+- 原方案把平台、视频 ID、来源 URL、时长和无字幕占位文本编码进笔记正文的隐藏标记。保存笔记本身不建任务;用户在笔记列表主动选择提取时才建任务。
+- API 用工作区和笔记 ID 限定任务,记录正文内容哈希;剪藏令牌和演示实例不能创建或领取任务。条件更新只允许一个桌面端领取,约 20 分钟未完成的任务可重新领取,避免永久卡住。
+- 桌面端写回同一条笔记时带修订号和内容哈希。正文被用户修改时不覆盖;原占位仍在则替换,否则追加实录。失败保留原笔记,不自动反复尝试。
-## 这一页有字幕
+### 本地媒体处理与模型调用
-字幕直接来自 YouTube 或哔哩哔哩,带可点回原片的时间戳,收在笔记底部的「字幕实录」里。
+- 原方案按平台白名单验证 HTTPS 来源,在本机调用 yt-dlp 只选最优音轨,不保存视频;限制已知时长约 30 分钟、音频约 24 MiB、下载和识别超时。任务结束后清理临时音频。大小和时长限制只减少资源占用,不构成版权授权。
+- yt-dlp 独立可执行文件从其发布页取得,并核对 SHA-256;桌面端后台检查版本。可选的 `--cookies-from-browser` 会复用本机浏览器登录状态,浏览器选择仅存在本机。这会扩大可获取的媒体范围,因此不能把「本地执行」等同于获得平台或权利人许可。
+- 语音模型服务、模型和默认模型原本按工作区配置。API Token 加密存放在实例,设置读取不返回明文;已登录桌面端在执行时领取凭据,直接将音频发送到用户配置的 OpenAI 兼容 `/audio/transcriptions` 接口。音频不经过笔记服务器,但会到达该模型服务,必须如实告知用户。
+- 模型返回分段时间戳时生成可点回原片的实录;仅返回全文时写入普通文本。任务失败记录错误码,不把 API Token 写入诊断信息。
-工作区已经配好默认聊天模型,并且剪藏用的 Token 允许调用它时,再在实录前面补三块:一句话总结、分段大纲、几条要点。有章节就用章节当大纲。模型没配、没有权限、超时或结果不可用时,总结那几块空着,实录仍然在。
-
-保存时不下载音频。列表里仍可以再点「提取字幕」,那是再听一遍音频,新的实录接在笔记后面。
-
-## 这一页没有字幕
-
-笔记仍然保存:标题、作者、来源,以及一句「这一集没有可用字幕」。不调用聊天模型,也不会自己开始识别。插件到这里就结束。
-
-要字幕时,在笔记列表右键这条笔记,选「提取字幕」。之后由已登录的桌面端接着做,仍然是同一条笔记:
-
-1. 语音识别没开启,或本机下载工具还没准备好:笔记保持原样。这次请求留着,等以后条件齐了再试。
-2. 桌面端没开:请求留到下次打开。只使用浏览器、没有桌面端的人,只能停在这句「没有可用字幕」。
-3. 语音识别已开启,下载工具也就绪:桌面端下载这一集的音频,交给用户配置的语音识别接口。下载默认带上这台电脑上 Chrome 的登录状态,所以会员或需要登录的视频也能下,只要那个浏览器里已经登录。成功后,用字幕实录换掉那句「这一集没有可用字幕」。识别结果带时间戳时,时间戳可以点回原片;接口只返回全文时,笔记里就是一段文字。
-4. 下载失败、识别失败、视频已知超过大约 30 分钟、音频过大,或用户已经改过这条笔记:保持保存时的内容,不会自动再试。想再试,再点一次「提取字幕」。所选浏览器没有登录,浏览器没安装,或系统不允许读取登录状态时,也是这样,笔记留在来源信息。
-
-点过之后不再弹窗。失败不删笔记。
-
-## 模型怎么配置
-
-聊天模型在设置 → 模型与代理。字幕识别是左侧单独的一项,两套配置互不替代。本机 Agent 那一排不参与视频笔记。
-
-**AI 模型服务**是原来的聊天模型。它只在「这一页有字幕」时使用,读的是字幕文本,用来写总结、大纲和要点。它不听音频。没配默认模型时,有字幕的笔记照常保存,只是没有总结。
-
-**字幕识别**是左侧单独的一项,里面配置语音识别模型,管法和 AI 模型服务一样。可以添加多套服务。每一套有名称、兼容标准、接口地址、API Token、一个或多个模型,以及自己的启用开关。目前的兼容标准是 OpenAI,以后可以再加其他厂商。上面的默认语音模型决定提取字幕用哪一个。新添加的服务不会换掉当前默认模型。没有默认模型,或默认模型那套服务停用了,桌面端不会去下载音频。
-
-请求发到所选服务地址下的 `/audio/transcriptions`。Token 加密存在实例上,设置页读不回原文。真正调用时,由已登录的桌面端取用,音频也从这台电脑直接发给该接口,不经过笔记服务器。
-
-设置里可以选下载时用哪个浏览器的登录状态,网页和桌面都能看到。默认是 Chrome,也可以改成其他浏览器,或改成不使用。真正记住这个选择的是桌面端,选择只留在这台电脑上,不会上传。网页上改一项只用来查看界面。同一页也显示本机下载工具的版本和路径;浏览器里没有这个工具时,版本显示尚未下载。
-
-接口要兼容 OpenAI 的语音转写:用 Bearer Token,提交音频文件和模型 ID。聊天模型的地址不能填在这里。
-
-## 技术路径
-
-插件把这一集的身份写成笔记正文里的隐藏标记,随笔记同步,不另建字段。标记含平台、视频编号、地址、时长,以及没有字幕时的那句占位。保存只写笔记,不创建识别任务,音频也不上传。
-
-列表读到有效标记,右键才有「提取字幕」。已登录用户向实例登记一笔任务,并记下当时的正文校验。剪藏令牌和演示实例不能登记。任务已在进行时返回进行中,不重复下载。
-
-桌面端用条件更新领取,同一笔只有一处在做。停在识别中约 20 分钟仍未完成的可以再领。领取后在本机用 yt-dlp 取音频,默认复用这台电脑上 Chrome 的登录状态;选哪个浏览器只存在本机,不上传。已知长于约 30 分钟,或音频过大,则失败且不下载。
-
-接着向实例领取当前默认且已启用的一套语音凭据。设置页读不回 Token。桌面进程把音频直接发到该地址下的 `/audio/transcriptions`,使用 Bearer 和音频文件。音频不经过笔记服务器,也不交给本机 Agent。目前只走 OpenAI 兼容转写,其他标准先失败,不调用接口。
-
-结果仍写回同一条笔记。正文校验未变、且仍有那句占位时,用字幕实录替换它;占位已不在时,实录接在后面。用户改过正文则整次失败,不写入。失败只结束任务,笔记留着,不弹窗。
+若未来重新考虑,应先明确内容授权、平台条款、目标市场和识别服务的数据处理方式,再重新设计入口与验证范围;不能直接恢复上述抓取链路。
diff --git a/migrations/0064_retire_video_transcription.sql b/migrations/0064_retire_video_transcription.sql
new file mode 100644
index 00000000..3ce96e95
--- /dev/null
+++ b/migrations/0064_retire_video_transcription.sql
@@ -0,0 +1,8 @@
+-- Retire the unreleased video transcription feature without rewriting applied migrations.
+PRAGMA foreign_keys = ON;
+
+DROP TABLE IF EXISTS ai_transcription_workspace_settings;
+DROP TABLE IF EXISTS ai_transcription_models;
+DROP TABLE IF EXISTS ai_transcription_providers;
+DROP TABLE IF EXISTS ai_transcription_settings;
+DROP TABLE IF EXISTS video_transcript_jobs;
diff --git a/package.json b/package.json
index 45d073ce..299289b7 100644
--- a/package.json
+++ b/package.json
@@ -64,7 +64,7 @@
"publish:stores": "bun scripts/store-delivery.mjs",
"publish:extension": "bun scripts/publish-web-clipper.mjs",
"test:desktop:sidecar": "bun scripts/test-desktop-sidecar.mjs",
- "test:desktop": "bun test apps/desktop/src/main/account-scope.test.mjs apps/desktop/src/main/app-protocol.test.mjs apps/desktop/src/main/child-process-start.test.mjs apps/desktop/src/main/clipboard-write.test.mjs apps/desktop/src/main/desktop-diagnostics.test.mjs apps/desktop/src/main/desktop-menu.test.mjs apps/desktop/src/main/desktop-update-flow.test.mjs apps/desktop/src/main/diagnostic-log.test.mjs apps/desktop/src/main/file-association.test.mjs apps/desktop/src/main/wechat-chat-archive.test.mjs apps/desktop/src/main/wechat-share-import.test.mjs apps/desktop/src/main/share-extension-registration.test.mjs apps/desktop/src/main/windows-share-menu.test.mjs apps/desktop/src/main/file-permissions.test.mjs apps/desktop/src/main/installation-location.test.mjs apps/desktop/src/main/local-data-reset.test.mjs apps/desktop/src/main/ai-direct.test.mjs apps/desktop/src/main/plugin-public-network.test.mjs apps/desktop/src/main/renderer-recovery.test.mjs apps/desktop/src/main/renderer-hibernate.test.mjs apps/desktop/src/main/renderer-startup-guard.test.mjs apps/desktop/src/main/renderer-storage-migration.test.mjs apps/desktop/src/main/resource-request.test.mjs apps/desktop/src/main/resource-url.test.mjs apps/desktop/src/main/scheduled-task-scheduler.test.mjs apps/desktop/src/main/screenshot-capture.test.mjs apps/desktop/src/main/staged-resource.test.mjs apps/desktop/src/main/staged-resource-alias.test.mjs apps/desktop/src/main/tray-icon.test.mjs apps/desktop/src/main/user-data-directory.test.mjs apps/desktop/src/main/yt-dlp-tool.test.mjs apps/desktop/src/main/speech-transcription.test.mjs apps/desktop/src/main/window-open-policy.test.mjs apps/desktop/src/main/window-visibility.test.mjs apps/desktop/src/main/windows-update-trust.test.mjs apps/web/src/lib/api.test.mjs apps/web/src/lib/app-page-path.test.mjs apps/web/src/lib/clipboard.test.mjs apps/web/src/lib/desktop-renderer-ready.test.mjs apps/web/src/lib/desktop-workspace-restore.test.mjs apps/web/src/lib/desktop-repository.test.mjs apps/web/src/lib/desktop-resources.test.mjs apps/web/src/lib/desktop-acp.test.mjs apps/web/src/lib/desktop-sync.test.mjs apps/web/src/lib/local-mirror.test.mjs apps/web/src/lib/repository.test.mjs apps/web/src/lib/runtime.test.mjs apps/web/src/lib/screenshot-import.test.mjs apps/web/src/lib/wechat-chat-import.test.mjs apps/web/src/lib/shared-file-import.test.mjs apps/web/src/lib/sync-queue.test.mjs && bun test --cwd apps/desktop src/main/acp-adapter-manager.test.mjs src/main/acp-host.test.mjs src/main/antigravity-proxy.test.mjs src/main/acp-mcp-bridge.test.mjs && bun run test:desktop:sidecar",
+ "test:desktop": "bun test apps/desktop/src/main/account-scope.test.mjs apps/desktop/src/main/app-protocol.test.mjs apps/desktop/src/main/child-process-start.test.mjs apps/desktop/src/main/clipboard-write.test.mjs apps/desktop/src/main/desktop-diagnostics.test.mjs apps/desktop/src/main/desktop-menu.test.mjs apps/desktop/src/main/desktop-update-flow.test.mjs apps/desktop/src/main/diagnostic-log.test.mjs apps/desktop/src/main/file-association.test.mjs apps/desktop/src/main/wechat-chat-archive.test.mjs apps/desktop/src/main/wechat-share-import.test.mjs apps/desktop/src/main/share-extension-registration.test.mjs apps/desktop/src/main/windows-share-menu.test.mjs apps/desktop/src/main/file-permissions.test.mjs apps/desktop/src/main/installation-location.test.mjs apps/desktop/src/main/local-data-reset.test.mjs apps/desktop/src/main/ai-direct.test.mjs apps/desktop/src/main/plugin-public-network.test.mjs apps/desktop/src/main/renderer-recovery.test.mjs apps/desktop/src/main/renderer-hibernate.test.mjs apps/desktop/src/main/renderer-startup-guard.test.mjs apps/desktop/src/main/renderer-storage-migration.test.mjs apps/desktop/src/main/resource-request.test.mjs apps/desktop/src/main/resource-url.test.mjs apps/desktop/src/main/scheduled-task-scheduler.test.mjs apps/desktop/src/main/screenshot-capture.test.mjs apps/desktop/src/main/staged-resource.test.mjs apps/desktop/src/main/staged-resource-alias.test.mjs apps/desktop/src/main/tray-icon.test.mjs apps/desktop/src/main/user-data-directory.test.mjs apps/desktop/src/main/window-open-policy.test.mjs apps/desktop/src/main/window-visibility.test.mjs apps/desktop/src/main/windows-update-trust.test.mjs apps/web/src/lib/api.test.mjs apps/web/src/lib/app-page-path.test.mjs apps/web/src/lib/clipboard.test.mjs apps/web/src/lib/desktop-renderer-ready.test.mjs apps/web/src/lib/desktop-workspace-restore.test.mjs apps/web/src/lib/desktop-repository.test.mjs apps/web/src/lib/desktop-resources.test.mjs apps/web/src/lib/desktop-acp.test.mjs apps/web/src/lib/desktop-sync.test.mjs apps/web/src/lib/local-mirror.test.mjs apps/web/src/lib/repository.test.mjs apps/web/src/lib/runtime.test.mjs apps/web/src/lib/screenshot-import.test.mjs apps/web/src/lib/wechat-chat-import.test.mjs apps/web/src/lib/shared-file-import.test.mjs apps/web/src/lib/sync-queue.test.mjs && bun test --cwd apps/desktop src/main/acp-adapter-manager.test.mjs src/main/acp-host.test.mjs src/main/antigravity-proxy.test.mjs src/main/acp-mcp-bridge.test.mjs && bun run test:desktop:sidecar",
"verify:desktop-renderer": "bun scripts/verify-desktop-renderer.mjs",
"verify:renderer-origin-migration": "bun run --cwd apps/desktop verify-origin-migration",
"verify:desktop-cross-version-startup": "bun scripts/verify-desktop-cross-version-startup.mjs",
diff --git a/packages/client/src/index.ts b/packages/client/src/index.ts
index 95a10141..7bbb28f0 100644
--- a/packages/client/src/index.ts
+++ b/packages/client/src/index.ts
@@ -53,10 +53,8 @@ import type {
TagSummary,
TiptapDoc,
AiSettings,
- AiTranscriptionSettings,
AiDiscoveredModel,
AiProvider,
- AiTranscriptionStandard,
AiPromptTemplate,
AiPromptTemplateCreateInput,
AiPromptTemplateUpdateInput,
@@ -304,23 +302,6 @@ export type AiProviderUpdatePayload = {
isEnabled: boolean;
};
-export type AiTranscriptionProviderCreatePayload = {
- provider: AiTranscriptionStandard;
- displayName: string;
- baseUrl: string;
- apiKey: string;
- isEnabled: boolean;
- initialModelId?: string;
-};
-
-export type AiTranscriptionProviderUpdatePayload = {
- provider: AiTranscriptionStandard;
- displayName: string;
- baseUrl: string;
- apiKey?: string;
- isEnabled: boolean;
-};
-
/** @deprecated Use SyncBootstrapResponse; kept for mobile source compatibility. */
export type MobileSyncBootstrapPage = SyncBootstrapResponse;
@@ -914,50 +895,6 @@ export const createEdgeEverClient = (options: EdgeEverClientOptions = {}) => {
return request<AiSettings>(`/api/v1/ai/settings${search}`);
},
- getAiTranscriptionSettings: () =>
- request<AiTranscriptionSettings>("/api/v1/ai/transcription-settings"),
-
- createAiTranscriptionProvider: (payload: AiTranscriptionProviderCreatePayload) =>
- request<AiTranscriptionSettings>("/api/v1/ai/transcription-providers", {
- method: "POST",
- body: JSON.stringify(payload),
- }),
-
- updateAiTranscriptionProvider: (providerId: string, payload: AiTranscriptionProviderUpdatePayload) =>
- request<AiTranscriptionSettings>(`/api/v1/ai/transcription-providers/${encodeURIComponent(providerId)}`, {
- method: "PUT",
- body: JSON.stringify(payload),
- }),
-
- deleteAiTranscriptionProvider: (providerId: string) =>
- request<AiTranscriptionSettings>(`/api/v1/ai/transcription-providers/${encodeURIComponent(providerId)}`, {
- method: "DELETE",
- }),
-
- addAiTranscriptionModel: (providerId: string, payload: { modelId: string; displayName?: string }) =>
- request<AiTranscriptionSettings>(`/api/v1/ai/transcription-providers/${encodeURIComponent(providerId)}/models`, {
- method: "POST",
- body: JSON.stringify(payload),
- }),
-
- deleteAiTranscriptionModel: (providerId: string, modelConfigId: string) =>
- request<AiTranscriptionSettings>(
- `/api/v1/ai/transcription-providers/${encodeURIComponent(providerId)}/models/${encodeURIComponent(modelConfigId)}`,
- { method: "DELETE" },
- ),
-
- updateDefaultAiTranscriptionModel: (modelConfigId: string | null) =>
- request<AiTranscriptionSettings>("/api/v1/ai/transcription-default-model", {
- method: "PUT",
- body: JSON.stringify({ modelConfigId }),
- }),
-
- extractVideoTranscript: (memoId: string) =>
- request<{ status: "queued" | "running" }>(
- `/api/v1/video-transcript-jobs/${encodeURIComponent(memoId)}/extract`,
- { method: "POST" },
- ),
-
createAiProvider: (payload: AiProviderCreatePayload) =>
request<AiSettings>("/api/v1/ai/providers", {
method: "POST",
diff --git a/packages/shared/src/document-extensions.ts b/packages/shared/src/document-extensions.ts
index 8b9154ed..e1f60a17 100644
--- a/packages/shared/src/document-extensions.ts
+++ b/packages/shared/src/document-extensions.ts
@@ -8,7 +8,6 @@ import { createEdgeEverDetailsExtensions } from "./details";
import { FileAttachment } from "./file-attachment";
import { ImageGallery } from "./image-gallery";
import { MergeDivider } from "./merge-divider";
-import { VideoNoteMarker } from "./video-note";
import { PdfAttachment } from "./pdf-attachment";
import { PluginEmbed } from "./plugin-embed";
import { EmptyExternalLink } from "./empty-external-link";
@@ -53,7 +52,6 @@ export const createEdgeEverDocumentExtensions = (
...withOptional(options.pdf, PdfAttachment),
...withOptional(options.file, FileAttachment),
MergeDivider,
- VideoNoteMarker,
...createEdgeEverDetailsExtensions(),
...withOptional(options.pluginEmbed, PluginEmbed),
...options.mathematics,
diff --git a/packages/shared/src/i18n/en-US.ts b/packages/shared/src/i18n/en-US.ts
index 6a0ad2c1..94d7557e 100644
--- a/packages/shared/src/i18n/en-US.ts
+++ b/packages/shared/src/i18n/en-US.ts
@@ -942,7 +942,6 @@ export const enUS = {
general: "General",
shortcuts: "Shortcuts",
ai: "Models and agents",
- speech: "Subtitle recognition",
mcp: "API / MCP",
data: "Import & Export",
advanced: "Advanced",
@@ -1169,67 +1168,6 @@ export const enUS = {
savedCredentialsUnavailable: "Saved model credentials cannot be decrypted. Open the service and enter the API key again.",
savedCredentialsUnavailableBadge: "Re-enter API key",
},
- speechTranscription: {
- title: "Speech recognition model settings",
- baseUrl: "Endpoint URL",
- baseUrlPlaceholder: "https://api.example.com/v1",
- baseUrlHint: "Requests go to /audio/transcriptions under this URL.",
- modelId: "Model ID",
- modelIdPlaceholder: "whisper-1",
- apiToken: "API token",
- apiTokenHint: "The credential is encrypted on the EdgeEver server.",
- apiTokenSavedHint: "The token is encrypted. Leave this blank to keep it unchanged.",
- apiTokenStoredPlaceholder: "Saved; enter a new token to replace it",
- apiTokenRequired: "An API token is required to turn speech transcription on.",
- saved: "Speech transcription settings saved.",
- failed: "Could not save speech transcription settings.",
- demoDisabled: "Speech transcription cannot be changed on the demo instance.",
- defaultModel: "Default speech model",
- noDefaultModel: "No model selected",
- defaultUnavailable: "The service for the current default model is disabled. Enable it or choose another model.",
- defaultBadge: "Default",
- addProvider: "Add a model service",
- noProviders: "No speech services yet. Add one, enable it with the switch, and choose a default model above.",
- displayName: "Name",
- standard: "Compatibility standard",
- standards: {
- "openai-compatible": "OpenAI-compatible",
- },
- serviceEnabled: "Enable service",
- serviceEnabledStatus: "Enabled",
- serviceDisabledStatus: "Disabled",
- serviceActions: "Service actions",
- addModel: "Add a model",
- editConnection: "Edit connection",
- removeModel: "Remove model",
- noModels: "No models yet",
- deleteProviderConfirm: "Delete the speech service “{{name}}”? Its saved token is deleted too.",
- deleteModelConfirm: "Remove the model “{{name}}”?",
- savedCredentialsUnavailable: "The saved speech token cannot be decrypted. Open this service and enter it again.",
- statusOn: "On",
- statusOff: "Off",
- cookieBrowser: "Browser sign-in reused by yt-dlp",
- cookieBrowserNone: "None",
- cookieBrowserApplied: "Later downloads will use the {{browser}} sign-in.",
- cookieBrowserOff: "Later downloads will not use a browser sign-in.",
- cookieBrowserFailed: "Could not save the browser choice.",
- cookieBrowsers: {
- chrome: "Chrome",
- chromium: "Chromium",
- edge: "Edge",
- firefox: "Firefox",
- safari: "Safari",
- brave: "Brave",
- opera: "Opera",
- vivaldi: "Vivaldi",
- whale: "Whale",
- },
- complianceNoticeTitle: "Usage Guidelines & Copyright Notice",
- complianceNoticePersonalUse: "Subtitle extraction and audio transcription features are intended solely for personal study, offline research, and assistive note-taking. Do not publicly distribute, broadcast, or use extracted content for commercial purposes.",
- complianceNoticeCopyright: "Intellectual property rights for all media assets, original scripts, and subtitles belong to the original creators and respective hosting platforms.",
- complianceNoticeLocalSecurity: "Parsing and download operations run entirely on your local device. Reused browser sign-in states are only used locally to access media streams you are authorized to view and are never uploaded or stored in the cloud.",
- cookieBrowserNotice: "Used only locally to access media streams in your active browser session for audio and subtitle tracks.",
- },
aiPrompts: {
title: "Prompt library",
description: "Manage reusable note-processing prompts. New workspaces start with editable starters you can change or delete.",
@@ -1580,25 +1518,6 @@ export const enUS = {
existingAttachments: "Existing attachments",
existingAttachmentsOriginalStorage: "Continue reading from their original storage",
clientVersion: "Client version",
- ytDlpVersion: "yt-dlp version",
- ytDlpPath: "yt-dlp path",
- ytDlpDownloading: "Downloading",
- ytDlpUpdating: "{{version}} · updating",
- ytDlpUpdateFailed: "{{version}} · {{reason}}",
- ytDlpFailed: "Download failed",
- ytDlpMissing: "Not downloaded",
- ytDlpErrors: {
- checksum_mismatch: "Checksum does not match",
- checksum_missing: "Release has no checksum",
- version_mismatch: "Version check failed",
- not_executable: "File cannot run",
- timeout: "Download timed out",
- network: "Network failed",
- unexpected_size: "Unexpected download size",
- unrecognized_release: "Latest release was not recognized",
- storage: "Could not write the local file",
- http: "HTTP {{status}}",
- },
releaseTime: "Release time",
build: "Build",
client: "Client",
@@ -1911,10 +1830,6 @@ export const enUS = {
importSharedFileInProgress: "Saving the file as a note…",
loadingMore: "Loading more notes",
openMemo: "Open note",
- extractTranscript: "Extract subtitles",
- extractTranscriptStarted: "Subtitle extraction started",
- extractTranscriptRunning: "Subtitles are already being extracted",
- extractTranscriptFailed: "Couldn't start subtitle extraction",
unpin: "Unpin",
pinMemo: "Pin note",
restoreMemo: "Restore note",
diff --git a/packages/shared/src/i18n/ja.ts b/packages/shared/src/i18n/ja.ts
index 7b67858f..bc2ac029 100644
--- a/packages/shared/src/i18n/ja.ts
+++ b/packages/shared/src/i18n/ja.ts
@@ -942,7 +942,6 @@ export const ja = {
general: "一般",
shortcuts: "ショートカット",
ai: "モデルとエージェント",
- speech: "字幕認識",
mcp: "API / MCP",
data: "インポートとエクスポート",
advanced: "詳細",
@@ -1169,67 +1168,6 @@ export const ja = {
savedCredentialsUnavailable: "保存済みのモデル認証情報を復号できません。サービスを開いて API Key を再入力してください。",
savedCredentialsUnavailableBadge: "API Key を再入力",
},
- speechTranscription: {
- title: "音声認識モデル設定",
- baseUrl: "エンドポイント URL",
- baseUrlPlaceholder: "https://api.example.com/v1",
- baseUrlHint: "リクエストはこの URL 配下の /audio/transcriptions に送られます。",
- modelId: "モデル ID",
- modelIdPlaceholder: "whisper-1",
- apiToken: "API トークン",
- apiTokenHint: "認証情報は EdgeEver サーバーで暗号化されます。",
- apiTokenSavedHint: "トークンは暗号化されています。空欄のままなら変更しません。",
- apiTokenStoredPlaceholder: "保存済み。新しいトークンを入力すると置き換えます",
- apiTokenRequired: "音声認識を有効にするには API トークンが必要です。",
- saved: "音声認識の設定を保存しました。",
- failed: "音声認識の設定を保存できませんでした。",
- demoDisabled: "デモインスタンスでは音声認識の設定を変更できません。",
- defaultModel: "デフォルト音声モデル",
- noDefaultModel: "モデル未選択",
- defaultUnavailable: "現在のデフォルトモデルのサービスは無効です。有効にするか、別のモデルを選んでください。",
- defaultBadge: "デフォルト",
- addProvider: "モデルサービスを追加",
- noProviders: "音声認識サービスはまだありません。追加してスイッチで有効にし、上でデフォルトモデルを選んでください。",
- displayName: "名前",
- standard: "互換規格",
- standards: {
- "openai-compatible": "OpenAI 互換",
- },
- serviceEnabled: "サービスを有効にする",
- serviceEnabledStatus: "有効",
- serviceDisabledStatus: "無効",
- serviceActions: "サービスの操作",
- addModel: "モデルを追加",
- editConnection: "接続を編集",
- removeModel: "モデルを削除",
- noModels: "モデルはまだありません",
- deleteProviderConfirm: "音声サービス「{{name}}」を削除しますか?保存したトークンも削除されます。",
- deleteModelConfirm: "モデル「{{name}}」を削除しますか?",
- savedCredentialsUnavailable: "保存済みの音声認識トークンを復号できません。このサービスを開いて、トークンを入力し直してください。",
- statusOn: "有効",
- statusOff: "無効",
- cookieBrowser: "yt-dlp が再利用するブラウザのログイン状態",
- cookieBrowserNone: "使わない",
- cookieBrowserApplied: "これ以降のダウンロードは {{browser}} のログイン状態を使います。",
- cookieBrowserOff: "これ以降のダウンロードはブラウザのログイン状態を使いません。",
- cookieBrowserFailed: "ブラウザの選択を保存できませんでした。",
- cookieBrowsers: {
- chrome: "Chrome",
- chromium: "Chromium",
- edge: "Edge",
- firefox: "Firefox",
- safari: "Safari",
- brave: "Brave",
- opera: "Opera",
- vivaldi: "Vivaldi",
- whale: "Whale",
- },
- complianceNoticeTitle: "利用規約と著作権に関する注意事項",
- complianceNoticePersonalUse: "字幕抽出および音声文字起こし機能は、個人の学習、オフライン研究、およびメモ作成の補助のみを目的としています。抽出したコンテンツを公開、配布、または商用利用しないでください。",
- complianceNoticeCopyright: "動画・音声素材およびその台詞、字幕等の知的財産権は、原作者および配信プラットフォームに帰属します。",
- complianceNoticeLocalSecurity: "関連する処理やダウンロードはすべてお使いのローカル端末上で実行されます。ブラウザのログイン状態はアクセス権のあるメディアの読み取りのみに利用され、クラウドに送信または保存されることはありません。",
- cookieBrowserNotice: "音声や字幕を取得するために、現在ログイン中のブラウザセッションをローカルでのみ利用します。",
- },
aiPrompts: {
title: "プロンプトライブラリ",
description: "再利用できるノート処理プロンプトを管理します。新しいワークスペースには、編集や削除できるスターターが入ります。",
@@ -1580,25 +1518,6 @@ export const ja = {
existingAttachments: "既存の添付",
existingAttachmentsOriginalStorage: "元のストレージから読み続けます",
clientVersion: "クライアントバージョン",
- ytDlpVersion: "yt-dlp のバージョン",
- ytDlpPath: "yt-dlp のパス",
- ytDlpDownloading: "ダウンロード中",
- ytDlpUpdating: "{{version}} · 更新中",
- ytDlpUpdateFailed: "{{version}} · {{reason}}",
- ytDlpFailed: "ダウンロード失敗",
- ytDlpMissing: "未ダウンロード",
- ytDlpErrors: {
- checksum_mismatch: "チェックサムが一致しません",
- checksum_missing: "リリースにチェックサムがありません",
- version_mismatch: "バージョン確認に失敗しました",
- not_executable: "ファイルを実行できません",
- timeout: "ダウンロードがタイムアウトしました",
- network: "ネットワークに失敗しました",
- unexpected_size: "ダウンロードサイズが異常です",
- unrecognized_release: "最新リリースを認識できません",
- storage: "ローカルファイルに書き込めません",
- http: "HTTP {{status}}",
- },
releaseTime: "リリース日時",
build: "ビルド",
client: "クライアント",
@@ -1911,10 +1830,6 @@ export const ja = {
importSharedFileInProgress: "ファイルをノートとして保存しています…",
loadingMore: "さらにノートを読み込み中",
openMemo: "ノートを開く",
- extractTranscript: "字幕を抽出",
- extractTranscriptStarted: "字幕の抽出を開始しました",
- extractTranscriptRunning: "このノートは字幕を抽出しています",
- extractTranscriptFailed: "字幕の抽出を開始できませんでした",
unpin: "ピン留めを外す",
pinMemo: "ノートをピン留め",
restoreMemo: "ノートを復元",
diff --git a/packages/shared/src/i18n/pl.ts b/packages/shared/src/i18n/pl.ts
index 0233c5ab..2fcf4624 100644
--- a/packages/shared/src/i18n/pl.ts
+++ b/packages/shared/src/i18n/pl.ts
@@ -942,7 +942,6 @@ export const pl = {
general: "Ogólne",
shortcuts: "Skróty",
ai: "Modele i agenci",
- speech: "Rozpoznawanie napisów",
mcp: "API / MCP",
data: "Import i eksport",
advanced: "Zaawansowane",
@@ -1169,67 +1168,6 @@ export const pl = {
savedCredentialsUnavailable: "Nie można odszyfrować zapisanych danych uwierzytelniających modelu. Otwórz usługę i ponownie wpisz klucz API.",
savedCredentialsUnavailableBadge: "Wpisz ponownie klucz API",
},
- speechTranscription: {
- title: "Ustawienia modeli rozpoznawania mowy",
- baseUrl: "Adres interfejsu",
- baseUrlPlaceholder: "https://api.example.com/v1",
- baseUrlHint: "Żądanie trafia na /audio/transcriptions pod tym adresem.",
- modelId: "ID modelu",
- modelIdPlaceholder: "whisper-1",
- apiToken: "Token API",
- apiTokenHint: "Dane uwierzytelniające są szyfrowane na serwerze EdgeEver.",
- apiTokenSavedHint: "Token jest zaszyfrowany. Pozostaw puste, aby go nie zmieniać.",
- apiTokenStoredPlaceholder: "Zapisano; wpisz nowy token, aby go zastąpić",
- apiTokenRequired: "Aby włączyć rozpoznawanie mowy, wpisz token API.",
- saved: "Zapisano ustawienia rozpoznawania mowy.",
- failed: "Nie udało się zapisać ustawień rozpoznawania mowy.",
- demoDisabled: "Na instancji demonstracyjnej nie można zmieniać rozpoznawania mowy.",
- defaultModel: "Domyślny model mowy",
- noDefaultModel: "Nie wybrano modelu",
- defaultUnavailable: "Usługa obecnego modelu domyślnego jest wyłączona. Włącz ją lub wybierz inny model.",
- defaultBadge: "Domyślny",
- addProvider: "Dodaj usługę modelu",
- noProviders: "Nie ma jeszcze usług rozpoznawania mowy. Dodaj jedną, włącz ją przełącznikiem i wybierz powyżej model domyślny.",
- displayName: "Nazwa",
- standard: "Standard zgodności",
- standards: {
- "openai-compatible": "Zgodny z OpenAI",
- },
- serviceEnabled: "Włącz usługę",
- serviceEnabledStatus: "Włączona",
- serviceDisabledStatus: "Wyłączona",
- serviceActions: "Działania usługi",
- addModel: "Dodaj model",
- editConnection: "Edytuj połączenie",
- removeModel: "Usuń model",
- noModels: "Nie ma jeszcze modeli",
- deleteProviderConfirm: "Usunąć usługę mowy „{{name}}”? Zapisany token też zostanie usunięty.",
- deleteModelConfirm: "Usunąć model „{{name}}”?",
- savedCredentialsUnavailable: "Nie można odszyfrować zapisanego tokenu rozpoznawania mowy. Otwórz tę usługę i wpisz token ponownie.",
- statusOn: "Włączone",
- statusOff: "Wyłączone",
- cookieBrowser: "Logowanie przeglądarki używane przez yt-dlp",
- cookieBrowserNone: "Nie używaj",
- cookieBrowserApplied: "Kolejne pobrania użyją logowania {{browser}}.",
- cookieBrowserOff: "Kolejne pobrania nie użyją logowania w przeglądarce.",
- cookieBrowserFailed: "Nie udało się zapisać wyboru przeglądarki.",
- cookieBrowsers: {
- chrome: "Chrome",
- chromium: "Chromium",
- edge: "Edge",
- firefox: "Firefox",
- safari: "Safari",
- brave: "Brave",
- opera: "Opera",
- vivaldi: "Vivaldi",
- whale: "Whale",
- },
- complianceNoticeTitle: "Zasady użytkowania i prawa autorskie",
- complianceNoticePersonalUse: "Funkcje wyodrębniania napisów i transkrypcji mowy są przeznaczone wyłącznie do nauki osobistej, badań offline i pomocy w sporządzaniu notatek. Nie należy publicznie rozpowszechniać ani wykorzystywać wyodrębnionych treści w celach komercyjnych.",
- complianceNoticeCopyright: "Prawa autorskie i własność intelektualna do materiałów wideo, oryginalnych dialogów i napisów należą do ich twórców oraz platform źródłowych.",
- complianceNoticeLocalSecurity: "Wszystkie operacje przetwarzania i pobierania działają wyłącznie na Twoim urządzeniu lokalnym. Zalogowane sesje przeglądarki są używane tylko lokalnie i nigdy nie są wysyłane do chmury.",
- cookieBrowserNotice: "Używane tylko lokalnie do odczytu strumieni multimediów w Twojej aktywnej przeglądarce.",
- },
aiPrompts: {
title: "Biblioteka promptów",
description: "Zarządzaj promptami wielokrotnego użytku do przetwarzania notatek. Nowe obszary robocze zaczynają z przykładowymi promptami, które możesz zmienić lub usunąć.",
@@ -1580,25 +1518,6 @@ export const pl = {
existingAttachments: "Istniejące załączniki",
existingAttachmentsOriginalStorage: "Nadal odczytywane z pierwotnego magazynu",
clientVersion: "Wersja klienta",
- ytDlpVersion: "Wersja yt-dlp",
- ytDlpPath: "Ścieżka yt-dlp",
- ytDlpDownloading: "Pobieranie",
- ytDlpUpdating: "{{version}} · aktualizowanie",
- ytDlpUpdateFailed: "{{version}} · {{reason}}",
- ytDlpFailed: "Pobieranie nie powiodło się",
- ytDlpMissing: "Nie pobrano",
- ytDlpErrors: {
- checksum_mismatch: "Suma kontrolna nie pasuje",
- checksum_missing: "Wydanie nie ma sumy kontrolnej",
- version_mismatch: "Sprawdzenie wersji nie powiodło się",
- not_executable: "Nie można uruchomić pliku",
- timeout: "Przekroczono czas pobierania",
- network: "Błąd sieci",
- unexpected_size: "Nieoczekiwany rozmiar pobierania",
- unrecognized_release: "Nie rozpoznano najnowszego wydania",
- storage: "Nie można zapisać pliku lokalnego",
- http: "HTTP {{status}}",
- },
releaseTime: "Data wydania",
build: "Kompilacja",
client: "Klient",
@@ -1911,10 +1830,6 @@ export const pl = {
importSharedFileInProgress: "Zapisywanie pliku jako notatki…",
loadingMore: "Wczytywanie kolejnych notatek",
openMemo: "Otwórz notatkę",
- extractTranscript: "Wyodrębnij napisy",
- extractTranscriptStarted: "Rozpoczęto wyodrębnianie napisów",
- extractTranscriptRunning: "Napisy są już wyodrębniane",
- extractTranscriptFailed: "Nie udało się rozpocząć wyodrębniania napisów",
unpin: "Odepnij",
pinMemo: "Przypnij notatkę",
restoreMemo: "Przywróć notatkę",
diff --git a/packages/shared/src/i18n/zh-CN.ts b/packages/shared/src/i18n/zh-CN.ts
index 87b9bad5..b125ed22 100644
--- a/packages/shared/src/i18n/zh-CN.ts
+++ b/packages/shared/src/i18n/zh-CN.ts
@@ -942,7 +942,6 @@ export const zhCN = {
general: "常规设置",
shortcuts: "快捷键",
ai: "模型与代理",
- speech: "字幕识别",
mcp: "API / MCP",
data: "导入导出",
advanced: "高级设置",
@@ -1169,67 +1168,6 @@ export const zhCN = {
savedCredentialsUnavailable: "已保存的模型凭据无法解密,请打开该服务并重新填写 API Key。",
savedCredentialsUnavailableBadge: "需重新填写密钥",
},
- speechTranscription: {
- title: "语音识别模型配置",
- baseUrl: "接口地址",
- baseUrlPlaceholder: "https://api.example.com/v1",
- baseUrlHint: "请求会发到该地址下的 /audio/transcriptions。",
- modelId: "模型 ID",
- modelIdPlaceholder: "whisper-1",
- apiToken: "API Token",
- apiTokenHint: "凭据将由 EdgeEver 服务端加密保存。",
- apiTokenSavedHint: "密钥已加密保存;留空表示保持不变。",
- apiTokenStoredPlaceholder: "已保存;输入新密钥可替换",
- apiTokenRequired: "开启语音识别需要填写 API Token。",
- saved: "语音识别设置已保存。",
- failed: "语音识别设置保存失败。",
- demoDisabled: "演示实例不允许修改语音识别设置。",
- defaultModel: "默认语音模型",
- noDefaultModel: "未选择模型",
- defaultUnavailable: "当前默认模型所在的服务已停用,请启用该服务或另选一个模型。",
- defaultBadge: "默认",
- addProvider: "添加模型服务",
- noProviders: "还没有语音识别服务。添加后用开关启用,并在上面选择默认模型。",
- displayName: "名称",
- standard: "兼容标准",
- standards: {
- "openai-compatible": "OpenAI 兼容标准",
- },
- serviceEnabled: "启用服务",
- serviceEnabledStatus: "已启用",
- serviceDisabledStatus: "已停用",
- serviceActions: "服务操作",
- addModel: "添加模型",
- editConnection: "编辑连接",
- removeModel: "移除模型",
- noModels: "还没有模型",
- deleteProviderConfirm: "删除语音服务「{{name}}」?保存的 Token 会一起删除。",
- deleteModelConfirm: "移除模型「{{name}}」?",
- savedCredentialsUnavailable: "已保存的语音识别 Token 无法解密,请打开该服务并重新填写。",
- statusOn: "已开启",
- statusOff: "未开启",
- cookieBrowser: "yt-dlp 复用的浏览器登录状态",
- cookieBrowserNone: "不使用",
- cookieBrowserApplied: "之后的下载会使用 {{browser}} 的登录状态。",
- cookieBrowserOff: "之后的下载不使用浏览器登录状态。",
- cookieBrowserFailed: "没能保存浏览器选择。",
- cookieBrowsers: {
- chrome: "Chrome",
- chromium: "Chromium",
- edge: "Edge",
- firefox: "Firefox",
- safari: "Safari",
- brave: "Brave",
- opera: "Opera",
- vivaldi: "Vivaldi",
- whale: "Whale",
- },
- complianceNoticeTitle: "使用规范与版权说明",
- complianceNoticePersonalUse: "字幕提取与音视频转录功能仅供个人学习、离线研读及辅助记录使用,请勿将提取内容公开发布、传播或用于商业营利。",
- complianceNoticeCopyright: "音视频素材及其原始台词、字幕等衍生内容的知识产权均归原作者及发布平台所有。",
- complianceNoticeLocalSecurity: "相关解析与下载逻辑完全在您的本地设备运行,浏览器登录状态仅用于读取您有权访问的公开或已购媒体流,绝不上报或云端存储。",
- cookieBrowserNotice: "仅在本地用于读取当前浏览器已登录的媒体流以获取音轨与字幕。",
- },
aiPrompts: {
title: "指令库",
description: "管理可复用的笔记处理指令。新工作区会预置常用示例,你可随意编辑或删除。",
@@ -1580,25 +1518,6 @@ export const zhCN = {
existingAttachments: "已有附件",
existingAttachmentsOriginalStorage: "继续从原存储读取",
clientVersion: "客户端版本",
- ytDlpVersion: "yt-dlp 版本",
- ytDlpPath: "yt-dlp 路径",
- ytDlpDownloading: "正在下载",
- ytDlpUpdating: "{{version}} · 正在更新",
- ytDlpUpdateFailed: "{{version}} · {{reason}}",
- ytDlpFailed: "下载失败",
- ytDlpMissing: "尚未下载",
- ytDlpErrors: {
- checksum_mismatch: "校验和不符",
- checksum_missing: "发布页没有校验和",
- version_mismatch: "版本校验失败",
- not_executable: "文件无法运行",
- timeout: "下载超时",
- network: "网络失败",
- unexpected_size: "下载大小异常",
- unrecognized_release: "无法识别最新版本",
- storage: "无法写入本地文件",
- http: "HTTP {{status}}",
- },
releaseTime: "发版时间",
build: "构建",
client: "客户端",
@@ -1911,10 +1830,6 @@ export const zhCN = {
importSharedFileInProgress: "正在把文件保存为笔记…",
loadingMore: "正在加载更多笔记",
openMemo: "打开笔记",
- extractTranscript: "提取字幕",
- extractTranscriptStarted: "已开始提取字幕",
- extractTranscriptRunning: "这条笔记正在提取字幕",
- extractTranscriptFailed: "没能开始提取字幕",
unpin: "取消置顶",
pinMemo: "置顶笔记",
restoreMemo: "恢复笔记",
diff --git a/packages/shared/src/index.ts b/packages/shared/src/index.ts
index 1c00893f..2d4abbf6 100644
--- a/packages/shared/src/index.ts
+++ b/packages/shared/src/index.ts
@@ -49,8 +49,6 @@ export * from "./file-attachment";
export * from "./note-image-card";
export * from "./pdf-attachment";
export * from "./types";
-export * from "./video-note";
-export * from "./video-cookie-browser";
export * from "./version";
export * from './plugin-capabilities';
diff --git a/packages/shared/src/schemas.ts b/packages/shared/src/schemas.ts
index 2cc7ed0d..4203397d 100644
--- a/packages/shared/src/schemas.ts
+++ b/packages/shared/src/schemas.ts
@@ -29,70 +29,6 @@ export const NotebookUpdateSchema = z.object({
sortOrder: z.number().int().optional(),
});
-const PUBLIC_VIDEO_HOSTS = new Set([
- "youtube.com",
- "www.youtube.com",
- "m.youtube.com",
- "youtu.be",
- "bilibili.com",
- "www.bilibili.com",
- "m.bilibili.com",
-]);
-
-export const isPublicVideoSourceUrl = (value: string) => {
- try {
- const url = new URL(value);
- return url.protocol === "https:"
- && !url.username
- && !url.password
- && PUBLIC_VIDEO_HOSTS.has(url.hostname);
- } catch {
- return false;
- }
-};
-
-export const VIDEO_TRANSCRIPT_ERROR_CODES = [
- "duration_limit",
- "unsupported_source",
- "audio_too_large",
- "download_failed",
- "transcribe_failed",
- "empty_transcript",
- "note_changed",
- "note_write_failed",
- "credentials_unavailable",
- "disabled",
-] as const;
-
-export const VideoTranscriptJobInputSchema = z.object({
- platform: z.enum(["youtube", "bilibili"]),
- videoId: z.string().trim().min(1).max(64),
- sourceUrl: z.string().trim().max(2048).refine(isPublicVideoSourceUrl, {
- message: "sourceUrl must be an https YouTube or Bilibili address.",
- }),
- durationSeconds: z.number().int().min(0).max(24 * 60 * 60).default(0),
- placeholderText: z.string().trim().min(1).max(200),
- transcriptLabel: z.string().trim().min(1).max(80),
-});
-
-export const parseVideoTranscriptJob = (value: unknown) => {
- const parsed = VideoTranscriptJobInputSchema.safeParse(value);
- return parsed.success ? parsed.data : undefined;
-};
-
-export const VideoTranscriptFinishSchema = z.object({
- status: z.enum(["ready", "failed"]),
- errorCode: z.enum(VIDEO_TRANSCRIPT_ERROR_CODES).nullable().optional(),
- claimedAt: z.string().datetime(),
-}).superRefine((input, context) => {
- if (input.status === "failed" && !input.errorCode) {
- context.addIssue({ code: "custom", path: ["errorCode"], message: "A failure needs an error code." });
- }
- if (input.status === "ready" && input.errorCode) {
- context.addIssue({ code: "custom", path: ["errorCode"], message: "A finished transcript has no error code." });
- }
-});
-
export const MemoCreateSchema = z.object({
notebookId: z.string().trim().min(1),
title: z.string().trim().max(160).optional(),
@@ -101,8 +37,6 @@ export const MemoCreateSchema = z.object({
tags: z.array(z.string()).optional(),
createdAt: z.string().datetime().optional(),
updatedAt: z.string().datetime().optional(),
- // An unusable job must not reject the note. The desktop simply has nothing to claim.
- videoTranscript: z.unknown().optional().transform((value) => parseVideoTranscriptJob(value)),
});
export const MemoUpdateSchema = z.object({
@@ -316,8 +250,6 @@ export const ObjectStorageConnectionTestSchema = z.object({
export const AiProviderSchema = z.enum(["openai-compatible", "anthropic", "google"]);
-export const AiTranscriptionStandardSchema = z.enum(["openai-compatible"]);
-
const rejectUnsafeAiBaseUrl = (value: string, context: z.RefinementCtx) => {
let url: URL;
try {
@@ -336,40 +268,6 @@ const rejectUnsafeAiBaseUrl = (value: string, context: z.RefinementCtx) => {
const AiBaseUrlSchema = z.string().trim().url().max(500).superRefine(rejectUnsafeAiBaseUrl);
-const AiTranscriptionProviderBaseUrlSchema = z.string().trim().min(1).max(500).superRefine((value, context) => {
- if (!z.string().url().safeParse(value).success) {
- context.addIssue({ code: "custom", message: "AI Base URL must be a valid URL." });
- return;
- }
- rejectUnsafeAiBaseUrl(value, context);
-});
-
-export const AiTranscriptionProviderCreateSchema = z.object({
- provider: AiTranscriptionStandardSchema.default("openai-compatible"),
- displayName: z.string().trim().min(1).max(80),
- baseUrl: AiTranscriptionProviderBaseUrlSchema,
- apiKey: z.string().trim().min(1).max(4096),
- isEnabled: z.boolean().default(true),
- initialModelId: z.string().trim().min(1).max(200).optional(),
-});
-
-export const AiTranscriptionProviderUpdateSchema = z.object({
- provider: AiTranscriptionStandardSchema.default("openai-compatible"),
- displayName: z.string().trim().min(1).max(80),
- baseUrl: AiTranscriptionProviderBaseUrlSchema,
- apiKey: z.string().trim().min(1).max(4096).optional(),
- isEnabled: z.boolean().default(true),
-});
-
-export const AiTranscriptionModelCreateSchema = z.object({
- modelId: z.string().trim().min(1).max(200),
- displayName: z.string().trim().min(1).max(80).optional(),
-});
-
-export const AiTranscriptionDefaultModelUpdateSchema = z.object({
- modelConfigId: z.string().trim().min(1).nullable(),
-});
-
const AiProviderConfigFieldsSchema = z.object({
provider: AiProviderSchema,
displayName: z.string().trim().min(1).max(80),
@@ -575,7 +473,6 @@ export type PublicTableForm = {
export type NotebookCreateInput = z.infer<typeof NotebookCreateSchema>;
export type NotebookUpdateInput = z.infer<typeof NotebookUpdateSchema>;
-export type VideoTranscriptJobInput = z.infer<typeof VideoTranscriptJobInputSchema>;
export type MemoCreateInput = z.infer<typeof MemoCreateSchema>;
export type MemoUpdateInput = z.infer<typeof MemoUpdateSchema>;
export type TemplateCreateInput = z.infer<typeof TemplateCreateSchema>;
diff --git a/packages/shared/src/types.ts b/packages/shared/src/types.ts
index 3a39cb5d..7d88dc84 100644
--- a/packages/shared/src/types.ts
+++ b/packages/shared/src/types.ts
@@ -29,7 +29,6 @@ export type MemoSummary = {
structuredTable?: boolean;
tablePreview?: TableSummaryPreview;
infographic?: boolean;
- videoNote?: boolean;
tags: string[];
isPinned: boolean;
isArchived: boolean;
@@ -182,8 +181,6 @@ export type ObjectStorageSettings = {
export type AiProvider = "openai-compatible" | "anthropic" | "google";
-export type AiTranscriptionStandard = "openai-compatible";
-
export type AiModelConfig = {
id: string;
providerConfigId: string;
@@ -209,32 +206,6 @@ export type AiSettings = {
readOnly: boolean;
};
-export type AiTranscriptionModel = {
- id: string;
- providerId: string;
- modelId: string;
- displayName: string;
-};
-
-export type AiTranscriptionProvider = {
- id: string;
- provider: AiTranscriptionStandard;
- displayName: string;
- baseUrl: string;
- isEnabled: boolean;
- hasApiKey: boolean;
- models: AiTranscriptionModel[];
- credentialsUnavailable?: boolean;
-};
-
-export type AiTranscriptionSettings = {
- providers: AiTranscriptionProvider[];
- defaultModelId: string | null;
- enabled: boolean;
- encryptionConfigured: boolean;
- readOnly: boolean;
-};
-
export type AiPromptTemplate = {
id: string;
origin: "default" | "custom";
diff --git a/packages/shared/src/video-cookie-browser.ts b/packages/shared/src/video-cookie-browser.ts
deleted file mode 100644
index 3abe6eda..00000000
--- a/packages/shared/src/video-cookie-browser.ts
+++ /dev/null
@@ -1,24 +0,0 @@
-// Browser names accepted by yt-dlp --cookies-from-browser. "none" turns the
-// reuse off. The desktop app keeps its own copy of this list so Electron does
-// not have to load this TypeScript module.
-export const VIDEO_COOKIE_BROWSERS = [
- "chrome",
- "chromium",
- "edge",
- "firefox",
- "safari",
- "brave",
- "opera",
- "vivaldi",
- "whale",
-] as const;
-
-export const DEFAULT_VIDEO_COOKIE_BROWSER = "chrome";
-
-export type VideoCookieBrowser = (typeof VIDEO_COOKIE_BROWSERS)[number] | "none";
-
-export const videoCookieBrowser = (value: unknown): VideoCookieBrowser => {
- const name = String(value ?? "").trim().toLowerCase();
- if (name === "none") return "none";
- return VIDEO_COOKIE_BROWSERS.find((browser) => browser === name) ?? DEFAULT_VIDEO_COOKIE_BROWSER;
-};
diff --git a/packages/shared/src/video-note.test.mjs b/packages/shared/src/video-note.test.mjs
deleted file mode 100644
index 97daf91e..00000000
--- a/packages/shared/src/video-note.test.mjs
+++ /dev/null
@@ -1,33 +0,0 @@
-import { expect, test } from "bun:test";
-import { docToMarkdown, markdownToDoc } from "./content.ts";
-import { getVideoNoteSummary, parseVideoNoteMarker, serializeVideoNoteMarker } from "./video-note.ts";
-
-const identity = {
- platform: "youtube",
- videoId: "abcdefghijk",
- sourceUrl: "https://www.youtube.com/watch?v=abcdefghijk",
- durationSeconds: 42,
- placeholderText: "这一集没有可用字幕",
- transcriptLabel: "字幕实录",
-};
-
-test("a video-note marker round-trips and stays out of the visible text", () => {
- const marker = serializeVideoNoteMarker(identity);
- const markdown = `# 标题\n\n这一集没有可用字幕\n\n${marker}\n`;
- expect(parseVideoNoteMarker(markdown)).toEqual(identity);
- expect(getVideoNoteSummary(markdown)).toEqual({ videoNote: true });
- expect(getVideoNoteSummary("普通笔记")).toEqual({ videoNote: false });
- expect(parseVideoNoteMarker("<!-- edgeever-video-v1:not-json -->")).toBeUndefined();
-
- const roundTrip = docToMarkdown(markdownToDoc(markdown));
- expect(parseVideoNoteMarker(roundTrip)).toEqual(identity);
- expect(roundTrip).not.toContain("&lt;!--");
- const doc = markdownToDoc(markdown);
- const texts = [];
- const walk = (node) => {
- if (node.text) texts.push(node.text);
- for (const child of node.content ?? []) walk(child);
- };
- walk(doc);
- expect(texts.join("\n")).not.toContain("edgeever-video");
-});
diff --git a/packages/shared/src/video-note.ts b/packages/shared/src/video-note.ts
deleted file mode 100644
index e20593ee..00000000
--- a/packages/shared/src/video-note.ts
+++ /dev/null
@@ -1,91 +0,0 @@
-import { mergeAttributes, Node } from "@tiptap/core";
-import { Base64 } from "js-base64";
-import { parseVideoTranscriptJob, type VideoTranscriptJobInput } from "./schemas";
-
-/** Hidden block that marks a saved YouTube or Bilibili note. */
-export const VIDEO_NOTE_NODE_TYPE = "edgeeverVideoNote" as const;
-
-const MARKER_PREFIX = "edgeever-video-v1";
-const MARKER = new RegExp(`<!--\\s*${MARKER_PREFIX}:([A-Za-z0-9_=-]+)\\s*-->`);
-const MARKER_TOKEN = new RegExp(`^<!--\\s*${MARKER_PREFIX}:([A-Za-z0-9_=-]+)\\s*-->[ \\t]*(?:\\n+|$)`);
-
-export const serializeVideoNoteMarker = (input: VideoTranscriptJobInput) =>
- `<!-- ${MARKER_PREFIX}:${Base64.encodeURI(JSON.stringify(input))} -->`;
-
-export const parseVideoNoteMarker = (markdown: string | null | undefined) => {
- const encoded = markdown?.match(MARKER)?.[1];
- if (!encoded || encoded.length > 8000) return undefined;
- try {
- return parseVideoTranscriptJob(JSON.parse(Base64.decode(encoded)));
- } catch {
- return undefined;
- }
-};
-
-export const getVideoNoteSummary = (markdown: string | null | undefined) => ({
- videoNote: Boolean(parseVideoNoteMarker(markdown)),
-});
-
-export const VideoNoteMarker = Node.create({
- name: VIDEO_NOTE_NODE_TYPE,
- group: "block",
- atom: true,
- selectable: false,
- draggable: false,
-
- addAttributes() {
- return {
- encoded: {
- default: "",
- parseHTML: (element) => element.getAttribute("data-edgeever-video-note") ?? "",
- renderHTML: (attributes) => (
- typeof attributes.encoded === "string" && attributes.encoded
- ? { "data-edgeever-video-note": attributes.encoded }
- : {}
- ),
- },
- };
- },
-
- parseHTML() {
- return [{ tag: "div[data-edgeever-video-note]" }];
- },
-
- renderHTML({ HTMLAttributes }) {
- return ["div", mergeAttributes(HTMLAttributes, {
- class: "edgeever-video-note",
- contenteditable: "false",
- style: "display:none",
- })];
- },
-
- renderMarkdown(node) {
- const encoded = String(node.attrs?.encoded ?? "");
- if (!/^[A-Za-z0-9_=-]+$/.test(encoded)) return "";
- return `<!-- ${MARKER_PREFIX}:${encoded} -->\n`;
- },
-
- parseMarkdown(token) {
- return {
- type: VIDEO_NOTE_NODE_TYPE,
- attrs: { encoded: String(token.encoded ?? "") },
- };
- },
-
- markdownTokenizer: {
- name: VIDEO_NOTE_NODE_TYPE,
- level: "block",
- start(src: string) {
- return src.indexOf(`<!-- ${MARKER_PREFIX}:`);
- },
- tokenize(src: string) {
- const match = MARKER_TOKEN.exec(src);
- if (!match) return undefined;
- return {
- type: VIDEO_NOTE_NODE_TYPE,
- raw: match[0],
- encoded: match[1],
- };
- },
- },
-});
diff --git a/scripts/test-desktop-sidecar.mjs b/scripts/test-desktop-sidecar.mjs
index 3f18de52..730fef82 100644
--- a/scripts/test-desktop-sidecar.mjs
+++ b/scripts/test-desktop-sidecar.mjs
@@ -7,7 +7,6 @@ import { join } from "node:path";
import { createDefaultDiagramDocument, getDiagramSummary, serializeDiagramDocument } from "../packages/shared/src/diagram.ts";
import { createDefaultInfographicDocument, getInfographicSummary, serializeInfographicDocument } from "../packages/shared/src/infographic.ts";
import { createDefaultTableDocument, getTableSummary, serializeTableDocument } from "../packages/shared/src/table.ts";
-import { getVideoNoteSummary, serializeVideoNoteMarker } from "../packages/shared/src/video-note.ts";
const sidecarPath = process.env.EDGE_EVER_SIDECAR_PATH ?? join(process.cwd(), "crates/desktop-sidecar/target/debug/edgeever-sidecar");
const migrationsPath = process.env.EDGE_EVER_MIGRATIONS_PATH ?? join(process.cwd(), "migrations");
@@ -187,7 +186,6 @@ assert.deepEqual(search.memos.map((memo) => memo.id), [first.memo.id]);
assert.equal(search.memos[0].diagramKind, null, "an ordinary note has no diagram kind");
assert.equal(search.memos[0].infographic, false, "an ordinary note is not an infographic");
assert.equal(search.memos[0].structuredTable, false, "an ordinary note is not a table");
-assert.equal(search.memos[0].videoNote, false, "an ordinary note is not a video note");
assert.equal(Object.hasOwn(search.memos[0], "diagramPreview"), false);
assert.equal(Object.hasOwn(search.memos[0], "tablePreview"), false);
assert.equal(Object.hasOwn(search.memos[0], "contentMarkdown"), false, "list rows must not include the document payload");
@@ -195,12 +193,11 @@ const assertListMetadata = async (title, markdown) => {
const created = await request("memo.create", { notebookId: inbox.id, title, contentMarkdown: markdown, tags: [] });
const listed = (await request("memo.list", { q: title, limit: 20 })).memos.find((memo) => memo.id === created.memo.id);
assert.ok(listed, `${title} should appear in the desktop list`);
- const expected = { ...getDiagramSummary(markdown), ...getInfographicSummary(markdown), ...getTableSummary(markdown), ...getVideoNoteSummary(markdown) };
+ const expected = { ...getDiagramSummary(markdown), ...getInfographicSummary(markdown), ...getTableSummary(markdown) };
assert.equal(listed.diagramKind, expected.diagramKind, title);
assert.deepEqual(listed.diagramPreview, expected.diagramPreview, title);
assert.equal(listed.infographic, expected.infographic, title);
assert.equal(listed.structuredTable, expected.structuredTable, title);
- assert.equal(listed.videoNote, expected.videoNote, title);
assert.deepEqual(listed.tablePreview, expected.tablePreview, title);
assert.equal(Object.hasOwn(listed, "contentMarkdown"), false, `${title} list row must not include the document payload`);
};
@@ -215,14 +212,6 @@ await assertListMetadata("Flowchart list icon", serializeDiagramDocument(createD
await assertListMetadata("Architecture list icon", serializeDiagramDocument(createDefaultDiagramDocument("architecture")));
await assertListMetadata("Table list icon", serializeTableDocument(createDefaultTableDocument()));
await assertListMetadata("Broken infographic list icon", "<!-- edgeever-infographic-v1:broken -->");
-await assertListMetadata("Video note list flag", `标题\n\n${serializeVideoNoteMarker({
- platform: "youtube",
- videoId: "abcdefghijk",
- sourceUrl: "https://www.youtube.com/watch?v=abcdefghijk",
- durationSeconds: 12,
- placeholderText: "这一集没有可用字幕",
- transcriptLabel: "字幕实录",
-})}\n`);
await request("memo.create", { notebookId: inbox.id, title: "Local daily", contentMarkdown: "prefix overlap", tags: ["local-daily"] });
const tagged = await request("memo.list", { tag: "local", limit: 20 });
assert.deepEqual(tagged.memos.map((memo) => memo.id), [first.memo.id], "tag filter should match an exact tag, not a prefix");
diff --git a/scripts/verify-deployment.mjs b/scripts/verify-deployment.mjs
index 91158d69..df590a97 100644
--- a/scripts/verify-deployment.mjs
+++ b/scripts/verify-deployment.mjs
@@ -20,11 +20,6 @@ export const REQUIRED_TABLES = [
"ai_provider_configs",
"ai_models",
"ai_workspace_settings",
- "ai_transcription_settings",
- "ai_transcription_providers",
- "ai_transcription_models",
- "ai_transcription_workspace_settings",
- "video_transcript_jobs",
];
export const buildSchemaVerificationSql = () =>
diff --git a/tests/ai-transcription-migration.test.ts b/tests/ai-transcription-migration.test.ts
deleted file mode 100644
index 529d02b0..00000000
--- a/tests/ai-transcription-migration.test.ts
+++ /dev/null
@@ -1,70 +0,0 @@
-import { globSync, readFileSync } from "node:fs";
-import { Database } from "bun:sqlite";
-import { describe, expect, test } from "bun:test";
-
-describe("speech transcription provider migration", () => {
- test("copies a filled-in endpoint into one selected service and leaves a cleared endpoint behind", () => {
- const db = new Database(":memory:");
- const migrations = globSync("migrations/*.sql").sort();
- const migrationIndex = migrations.findIndex((path) => path.endsWith("0062_ai_transcription_providers.sql"));
-
- for (const migration of migrations.slice(0, migrationIndex)) {
- db.exec(readFileSync(migration, "utf8"));
- }
- db.query("INSERT INTO workspaces (id, name, is_personal) VALUES (?, ?, 0)").run("ws_speech_off", "Off");
- db.query(
- `INSERT INTO ai_transcription_settings (
- workspace_id, base_url, model_id, api_key_encrypted
- ) VALUES (?, ?, ?, ?)`,
- ).run(
- "ws_default",
- "https://api.groq.com/openai/v1/",
- "whisper-large-v3-turbo",
- "v1.encrypted-speech-key",
- );
- db.query(
- `INSERT INTO ai_transcription_settings (
- workspace_id, base_url, model_id, api_key_encrypted
- ) VALUES (?, ?, ?, ?)`,
- ).run("ws_speech_off", "", "whisper-1", "v1.kept-but-hidden");
-
- for (const migration of migrations.slice(migrationIndex)) {
- db.exec(readFileSync(migration, "utf8"));
- }
-
- expect(db.query(
- `SELECT id, provider, display_name, base_url, api_key_encrypted, is_enabled
- FROM ai_transcription_providers WHERE workspace_id = ?`,
- ).get("ws_default")).toEqual({
- id: "atp_ws_default",
- provider: "openai-compatible",
- display_name: "api.groq.com",
- base_url: "https://api.groq.com/openai/v1",
- api_key_encrypted: "v1.encrypted-speech-key",
- is_enabled: 1,
- });
- expect(db.query(
- `SELECT id, provider_id, model_id, display_name
- FROM ai_transcription_models WHERE provider_id = ?`,
- ).get("atp_ws_default")).toEqual({
- id: "atm_ws_default",
- provider_id: "atp_ws_default",
- model_id: "whisper-large-v3-turbo",
- display_name: "whisper-large-v3-turbo",
- });
- expect(db.query(
- `SELECT default_model_id FROM ai_transcription_workspace_settings WHERE workspace_id = ?`,
- ).get("ws_default")).toEqual({ default_model_id: "atm_ws_default" });
- expect(db.query(
- `SELECT COUNT(*) AS count FROM ai_transcription_providers WHERE workspace_id = ?`,
- ).get("ws_speech_off")).toEqual({ count: 0 });
- expect(db.query(
- `SELECT api_key_encrypted FROM ai_transcription_settings WHERE workspace_id = ?`,
- ).get("ws_speech_off")).toEqual({ api_key_encrypted: "v1.kept-but-hidden" });
-
- db.query("DELETE FROM ai_transcription_providers WHERE id = ?").run("atp_ws_default");
- expect(db.query(
- `SELECT default_model_id FROM ai_transcription_workspace_settings WHERE workspace_id = ?`,
- ).get("ws_default")).toEqual({ default_model_id: null });
- });
-});
diff --git a/tests/video-transcription-retirement-migration.test.ts b/tests/video-transcription-retirement-migration.test.ts
new file mode 100644
index 00000000..de8207f0
--- /dev/null
+++ b/tests/video-transcription-retirement-migration.test.ts
@@ -0,0 +1,46 @@
+import { globSync, readFileSync } from "node:fs";
+import { Database } from "bun:sqlite";
+import { expect, test } from "bun:test";
+
+test("an existing speech configuration is removed without changing note and AI tables", () => {
+ const db = new Database(":memory:");
+ try {
+ const migrations = globSync("migrations/*.sql").sort();
+ const providerIndex = migrations.findIndex((path) => path.endsWith("0062_ai_transcription_providers.sql"));
+ const retirementIndex = migrations.findIndex((path) => path.endsWith("0064_retire_video_transcription.sql"));
+ expect(providerIndex).toBeGreaterThan(0);
+ expect(retirementIndex).toBeGreaterThan(providerIndex);
+
+ for (const migration of migrations.slice(0, providerIndex)) {
+ db.exec(readFileSync(migration, "utf8"));
+ }
+ db.query(
+ `INSERT INTO ai_transcription_settings
+ (workspace_id, base_url, model_id, api_key_encrypted)
+ VALUES (?, ?, ?, ?)`,
+ ).run("ws_default", "https://asr.example/v1", "whisper-1", "encrypted-test-key");
+
+ for (const migration of migrations.slice(providerIndex, retirementIndex)) {
+ db.exec(readFileSync(migration, "utf8"));
+ }
+ expect(db.query("SELECT COUNT(*) AS count FROM ai_transcription_providers").get()).toEqual({ count: 1 });
+
+ db.exec(readFileSync(migrations[retirementIndex], "utf8"));
+ const tables = db.query("SELECT name FROM sqlite_master WHERE type = 'table'").all().map((row) => row.name);
+ for (const retired of [
+ "ai_transcription_settings",
+ "ai_transcription_providers",
+ "ai_transcription_models",
+ "ai_transcription_workspace_settings",
+ "video_transcript_jobs",
+ ]) {
+ expect(tables).not.toContain(retired);
+ }
+ for (const kept of ["workspaces", "memos", "ai_provider_configs", "ai_models"]) {
+ expect(tables).toContain(kept);
+ }
+ expect(db.query("PRAGMA foreign_key_check").all()).toEqual([]);
+ } finally {
+ db.close();
+ }
+});