@henjicc/ai-sdk 0.6.2 → 0.8.0

package/dist/providers/tts-voices.js

@@ -0,0 +1,139 @@
+import { fetchProvider } from './provider-fetch.js';
+const VOICE_PROVIDERS = {
+    'bailian-minimax-speech-2.8': 'bailian',
+    'bailian-cosyvoice-v3.5': 'bailian',
+    'bailian-qwen-audio-3.1-tts-flash': 'bailian',
+    'siliconflow-cosyvoice2-tts': 'siliconflow',
+};
+function object(value) {
+    return value && typeof value === 'object' && !Array.isArray(value)
+        ? value
+        : null;
+}
+function string(value) {
+    return typeof value === 'string' ? value.trim() : '';
+}
+function voiceDescription(value) {
+    if (typeof value === 'string')
+        return value.trim() || undefined;
+    if (Array.isArray(value)) {
+        const joined = value.filter((item) => typeof item === 'string' && item.trim().length > 0).join(' · ');
+        return joined || undefined;
+    }
+    return undefined;
+}
+function uniqueVoices(voices) {
+    return [...new Map(voices.map((voice) => [voice.id, voice])).values()];
+}
+async function requestJson(providerId, url, apiKey, runtime, body) {
+    const response = await fetchProvider(providerId, url, {
+        method: body ? 'POST' : 'GET',
+        headers: {
+            Authorization: `Bearer ${apiKey}`,
+            ...(body ? { 'Content-Type': 'application/json' } : {}),
+        },
+        ...(body ? { body: JSON.stringify(body) } : {}),
+    }, { transport: runtime.transport, retryPreconnectOnce: false });
+    if (!response.ok) {
+        throw new Error(`${providerId} 音色列表请求失败（HTTP ${response.status}）`);
+    }
+    try {
+        return await response.json();
+    }
+    catch {
+        throw new Error(`${providerId} 音色列表响应格式错误`);
+    }
+}
+async function listBailianMinimax(apiKey, runtime) {
+    const payload = object(await requestJson('bailian', 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation', apiKey, runtime, { model: 'MiniMax/speech-2.8-turbo', input: { action: 'get_voice', voice_type: 'all' } }));
+    const output = object(payload?.output);
+    const baseResp = object(output?.base_resp);
+    if (!output || baseResp?.status_code !== 0 || (!Array.isArray(output.system_voice) && !Array.isArray(output.voice_cloning))) {
+        throw new Error('百炼 MiniMax 音色列表返回错误');
+    }
+    const voices = [];
+    for (const source of ['system_voice', 'voice_cloning']) {
+        const entries = output[source];
+        if (!Array.isArray(entries))
+            continue;
+        for (const item of entries) {
+            const entry = object(item);
+            const id = string(entry?.voice_id);
+            if (!id)
+                continue;
+            voices.push({
+                id,
+                name: string(entry?.voice_name) || id,
+                description: voiceDescription(entry?.description),
+                source: source === 'system_voice' ? 'system' : 'clone',
+            });
+        }
+    }
+    return uniqueVoices(voices);
+}
+async function listBailianEnrollment(modelId, apiKey, runtime) {
+    const prefix = modelId === 'bailian-cosyvoice-v3.5' ? 'cosyvoice-v3.5-' : 'qwen-audio-3.1-tts-flash-';
+    const voices = [];
+    const pageSize = 50;
+    for (let pageIndex = 0; pageIndex < 20; pageIndex += 1) {
+        const payload = object(await requestJson('bailian', 'https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization', apiKey, runtime, { model: 'voice-enrollment', input: { action: 'list_voice', page_index: pageIndex, page_size: pageSize } }));
+        const output = object(payload?.output);
+        const entries = output?.voice_list;
+        if (!Array.isArray(entries))
+            throw new Error('百炼复刻音色列表响应格式错误');
+        for (const item of entries) {
+            const entry = object(item);
+            const id = string(entry?.voice_id);
+            if (!id.startsWith(prefix) || entry?.status !== 'OK')
+                continue;
+            voices.push({ id, name: id, source: 'clone' });
+        }
+        if (entries.length < pageSize)
+            return uniqueVoices(voices);
+    }
+    throw new Error('百炼复刻音色超过查询页数限制');
+}
+async function listSiliconflow(apiKey, runtime) {
+    const payload = await requestJson('siliconflow', 'https://api.siliconflow.cn/v1/audio/voice/list', apiKey, runtime);
+    const root = object(payload);
+    const data = object(root?.data);
+    const entries = Array.isArray(payload) ? payload
+        : Array.isArray(root?.data) ? root.data
+            : Array.isArray(data?.voices) ? data.voices
+                : Array.isArray(root?.voices) ? root.voices
+                    : null;
+    if (!entries)
+        throw new Error('硅基流动音色列表响应格式错误');
+    const voices = [];
+    for (const item of entries) {
+        const entry = object(item);
+        const id = string(entry?.uri);
+        if (!id)
+            continue;
+        voices.push({ id, name: string(entry?.customName) || string(entry?.name) || id, source: 'clone' });
+    }
+    return uniqueVoices(voices);
+}
+/** 查询供应商账号中的可用音色；静态系统音色由宿主的展示目录提供。 */
+export async function listTtsVoices(modelId, runtime) {
+    if (!Object.prototype.hasOwnProperty.call(VOICE_PROVIDERS, modelId))
+        throw new Error(`不支持查询 ${modelId} 的音色`);
+    const providerId = VOICE_PROVIDERS[modelId];
+    const apiKey = await runtime.credentials.get('generation', providerId);
+    if (!apiKey)
+        throw new Error(`请先设置${providerId} API Key`);
+    runtime.logger?.info('tts_voice.list.start', { providerId, modelId });
+    try {
+        const voices = modelId === 'bailian-minimax-speech-2.8'
+            ? await listBailianMinimax(apiKey, runtime)
+            : modelId === 'siliconflow-cosyvoice2-tts'
+                ? await listSiliconflow(apiKey, runtime)
+                : await listBailianEnrollment(modelId, apiKey, runtime);
+        runtime.logger?.info('tts_voice.list.completed', { providerId, modelId, context: { count: voices.length } });
+        return voices;
+    }
+    catch (error) {
+        runtime.logger?.warn('tts_voice.list.failed', { providerId, modelId, error });
+        throw error;
+    }
+}

package/dist/providers/volcengine-speech.d.ts

@@ -0,0 +1,3 @@
+import type { ProviderContinuePollingInput, ProviderExecutionInput, ProviderExecutionResult } from '../types/runtime.js';
+export declare function execute(input: ProviderExecutionInput): Promise<ProviderExecutionResult>;
+export declare function continuePolling(_input: ProviderContinuePollingInput): Promise<ProviderExecutionResult>;

package/dist/providers/volcengine-speech.js

@@ -0,0 +1,143 @@
+import { AiRuntimeError } from '../runtime/AiRuntimeError.js';
+import { volcengineRequestId } from '../capabilities/speech-recognition/volcengine/request-id.js';
+import { createUtf8StreamDecoder } from '../protocols/utf8-stream-decoder.js';
+import { fromBase64 } from '../upload/base64.js';
+import { audioBytesToDataUri } from './audio-data.js';
+import { isJsonObject, normalizeEndpoint } from './helpers.js';
+import { fetchProvider } from './provider-fetch.js';
+const BASE_URL = 'https://openspeech.bytedance.com';
+const ENDPOINT = '/api/v3/tts/unidirectional';
+function decodeChunk(encoded) {
+    try {
+        return fromBase64(encoded);
+    }
+    catch {
+        throw new AiRuntimeError('invalid_response', 'Volcengine speech returned invalid base64 audio');
+    }
+}
+function parseFrame(payload, audio, usage) {
+    if (!payload)
+        return;
+    let value;
+    try {
+        value = JSON.parse(payload);
+    }
+    catch {
+        throw new AiRuntimeError('invalid_response', 'Volcengine speech returned invalid JSON frame');
+    }
+    if (!isJsonObject(value) || value.code !== 0) {
+        throw new AiRuntimeError('provider_task_failed', isJsonObject(value) && typeof value.message === 'string' ? value.message : 'Volcengine speech failed');
+    }
+    if (typeof value.data === 'string' && value.data.length > 0)
+        audio.push(decodeChunk(value.data));
+    if (isJsonObject(value.usage) && typeof value.usage.text_words === 'number')
+        usage.textWords = value.usage.text_words;
+}
+function consumeFrames(buffer, audio, usage) {
+    let start = -1;
+    let depth = 0;
+    let quoted = false;
+    let escaped = false;
+    let consumed = 0;
+    for (let index = 0; index < buffer.length; index += 1) {
+        const character = buffer[index];
+        if (start < 0) {
+            if (/\s/.test(character)) {
+                consumed = index + 1;
+                continue;
+            }
+            if (character !== '{')
+                throw new AiRuntimeError('invalid_response', 'Volcengine speech returned invalid JSON frame');
+            start = index;
+            depth = 1;
+            continue;
+        }
+        if (quoted) {
+            if (escaped)
+                escaped = false;
+            else if (character === '\\')
+                escaped = true;
+            else if (character === '"')
+                quoted = false;
+        }
+        else if (character === '"')
+            quoted = true;
+        else if (character === '{')
+            depth += 1;
+        else if (character === '}') {
+            depth -= 1;
+            if (depth === 0) {
+                parseFrame(buffer.slice(start, index + 1), audio, usage);
+                start = -1;
+                consumed = index + 1;
+            }
+        }
+    }
+    return buffer.slice(start < 0 ? consumed : start);
+}
+export async function execute(input) {
+    if (!isJsonObject(input.body) || !isJsonObject(input.body.req_params)) {
+        throw new AiRuntimeError('invalid_request', 'Volcengine speech requires req_params');
+    }
+    const resourceId = input.body.resource_id;
+    if (resourceId !== 'seed-tts-2.0' && resourceId !== 'seed-icl-2.0') {
+        throw new AiRuntimeError('invalid_request', 'Volcengine speech resource ID is invalid');
+    }
+    const response = await fetchProvider('Volcengine Speech', normalizeEndpoint(BASE_URL, ENDPOINT), {
+        method: 'POST',
+        headers: {
+            'X-Api-Key': input.apiKey,
+            'X-Api-Resource-Id': resourceId,
+            'X-Api-Request-Id': volcengineRequestId(input.requestId, undefined),
+            'X-Control-Require-Usage-Tokens-Return': '*',
+            'Content-Type': 'application/json',
+        },
+        body: JSON.stringify({ req_params: input.body.req_params }),
+        signal: input.signal,
+    }, { transport: input.runtime.transport, retryPreconnectOnce: false });
+    if (!response.ok) {
+        await response.body?.cancel().catch(() => undefined);
+        throw new AiRuntimeError('provider_task_failed', `Volcengine speech request failed (HTTP ${response.status})`);
+    }
+    if (!response.body)
+        throw new AiRuntimeError('empty_result', 'Volcengine speech returned no body');
+    const reader = response.body.getReader();
+    const decoder = createUtf8StreamDecoder();
+    const audio = [];
+    const usage = {};
+    let pending = '';
+    const cancelReader = () => { void reader.cancel(input.signal?.reason).catch(() => undefined); };
+    input.signal?.addEventListener('abort', cancelReader, { once: true });
+    try {
+        while (true) {
+            input.signal?.throwIfAborted();
+            const { done, value } = await reader.read();
+            input.signal?.throwIfAborted();
+            pending += done ? decoder.decode() : decoder.decode(value, { stream: true });
+            pending = consumeFrames(pending, audio, usage);
+            if (done)
+                break;
+        }
+        if (pending.trim())
+            throw new AiRuntimeError('invalid_response', 'Volcengine speech returned incomplete JSON frame');
+    }
+    finally {
+        input.signal?.removeEventListener('abort', cancelReader);
+        reader.releaseLock();
+    }
+    const total = audio.reduce((sum, chunk) => sum + chunk.byteLength, 0);
+    const joined = new Uint8Array(total);
+    let offset = 0;
+    for (const chunk of audio) {
+        joined.set(chunk, offset);
+        offset += chunk.byteLength;
+    }
+    return {
+        status: 'completed',
+        url: audioBytesToDataUri(joined, 'mp3'),
+        metadata: { byteLength: total, ...(usage.textWords === undefined ? {} : { textWords: usage.textWords }) },
+    };
+}
+export async function continuePolling(_input) {
+    throw new AiRuntimeError('unsupported_provider', 'Volcengine speech uses one HTTP stream');
+}

package/dist/types/model.d.ts

@@ -17,11 +17,11 @@ export type BuiltinModelType = 'image' | 'video' | 'audio';
  * 它描述模型目录中的产出能力，不等同于上传预处理层的 `MediaKind`。
  */
 export type ModelType = BuiltinModelType | (string & {});
-/** SDK 随包提供的 8 个供应商；同时作为初始化与测试的单一清单。 */
-export declare const BUILTIN_PROVIDER_IDS: readonly ["apimart", "bailian", "volcengine", "ppio", "kie", "modelscope", "fal", "grsai"];
-/** 开放 ProviderId 仍保留这 8 个字面量的编辑器补全。 */
+/** SDK 随包提供的生成供应商；同时作为初始化与测试的单一清单。 */
+export declare const BUILTIN_PROVIDER_IDS: readonly ["apimart", "bailian", "volcengine", "volcengine-speech", "ppio", "kie", "modelscope", "fal", "grsai", "siliconflow"];
+/** 开放 ProviderId 仍保留内置供应商字面量的编辑器补全。 */
 export type BuiltinProviderId = (typeof BUILTIN_PROVIDER_IDS)[number];
-/** Provider ID：8 个内置供应商 + 第三方扩展字符串。 */
+/** Provider ID：内置供应商 + 第三方扩展字符串。 */
 export type ProviderId = BuiltinProviderId | (string & {});
 /**
  * 模型标签。

package/dist/types/model.js

@@ -8,14 +8,16 @@
  * 这里保留 selector/builder 的真实函数，是 catalog 与所有宿主运行时共同消费的唯一模型
  * 契约；JSON DTO 等跨进程数据类型仍位于 `./runtime.ts`。
  */
-/** SDK 随包提供的 8 个供应商；同时作为初始化与测试的单一清单。 */
+/** SDK 随包提供的生成供应商；同时作为初始化与测试的单一清单。 */
 export const BUILTIN_PROVIDER_IDS = [
     'apimart',
     'bailian',
     'volcengine',
+    'volcengine-speech',
     'ppio',
     'kie',
     'modelscope',
     'fal',
     'grsai',
+    'siliconflow',
 ];