deepfish-ai 1.0.26 → 1.0.28

package/src/AgentRobot/BaseAgentRobot/lazy-tools/AliBailian.js

@@ -0,0 +1,997 @@
+const fs = require('fs-extra')
+const path = require('path')
+
+// 阿里百炼官方推荐 Base URL (OpenAI 兼容模式)
+const REQUIRED_BASE_URL = 'https://dashscope.aliyuncs.com/compatible-mode/v1'
+
+// 模型常量
+const MODEL_TEXT_TO_IMAGE = 'wanx2.1-t2i-turbo'
+const MODEL_TEXT_TO_VIDEO = 'wanx2.1-t2v-turbo'
+const MODEL_IMAGE_RECOGNITION = 'qwen-vl-max'
+const MODEL_TEXT_TO_SPEECH = 'qwen-tts'
+const MODEL_SPEECH_TO_TEXT = 'qwen-audio-asr'
+
+// 音频默认参数
+const DEFAULT_TTS_VOICE = 'Cherry'
+const DEFAULT_TTS_FORMAT = 'mp3'
+
+function ok(data = null) {
+	return { success: true, data }
+}
+
+function fail(error, data = null) {
+	return {
+		success: false,
+		error: error?.message || String(error),
+		data,
+	}
+}
+
+function getAiConfig(ctx) {
+	const aiConfig = ctx?.agentRobot?.aiConfig || {}
+	const baseUrl = String(aiConfig.baseUrl || '').trim()
+	const apiKey = String(aiConfig.apiKey || '').trim()
+
+	if (!baseUrl) {
+		throw new Error('aiConfig.baseUrl is empty, please configure AI baseUrl first')
+	}
+	if (!apiKey) {
+		throw new Error('aiConfig.apiKey is empty, please configure AI apiKey first')
+	}
+	return { baseUrl, apiKey }
+}
+
+function checkSkillAvailability(ctx) {
+	const { baseUrl } = getAiConfig(ctx)
+	// 检查是否包含 dashscope.aliyuncs.com
+	const normalized = String(baseUrl || '').toLowerCase()
+	const canUse = normalized.includes('dashscope.aliyuncs.com')
+	return {
+		canUse,
+		baseUrl,
+		requiredBaseUrl: REQUIRED_BASE_URL,
+		reason: canUse
+			? 'AliBailian skill is available.'
+			: `AliBailian skill is unavailable. aiConfig.baseUrl must contain 'dashscope.aliyuncs.com'.`,
+	}
+}
+
+function assertSkillAvailability(ctx) {
+	const checkResult = checkSkillAvailability(ctx)
+	if (!checkResult.canUse) {
+		throw new Error(checkResult.reason)
+	}
+}
+
+function getRootBaseUrl(baseUrl) {
+	const normalized = String(baseUrl || '').replace(/\/+$/, '')
+	const lower = normalized.toLowerCase()
+	// coding 子域用于 OpenAI 兼容调用，AIGC 服务接口需走 dashscope 主域
+	if (lower.includes('coding.dashscope.aliyuncs.com')) {
+		return 'https://dashscope.aliyuncs.com'
+	}
+	if (lower.endsWith('/compatible-mode/v1')) {
+		return normalized.replace(/\/compatible-mode\/v1$/i, '')
+	}
+	if (lower.endsWith('/v1')) {
+		return normalized.replace(/\/v1$/i, '')
+	}
+	return normalized
+}
+
+function joinUrl(baseUrl, endpoint) {
+	return `${String(baseUrl || '').replace(/\/+$/, '')}/${String(endpoint || '').replace(/^\/+/, '')}`
+}
+
+function sleep(ms) {
+	return new Promise((resolve) => setTimeout(resolve, ms))
+}
+
+function buildImageDataUrl(imagePath) {
+	const fullPath = path.resolve(process.cwd(), imagePath)
+	if (!fs.existsSync(fullPath)) {
+		throw new Error(`Image file does not exist: ${fullPath}`)
+	}
+	const ext = path.extname(fullPath).toLowerCase()
+	const mimeTypeMap = {
+		'.png': 'image/png',
+		'.jpg': 'image/jpeg',
+		'.jpeg': 'image/jpeg',
+		'.webp': 'image/webp',
+		'.gif': 'image/gif',
+		'.bmp': 'image/bmp',
+	}
+	const mimeType = mimeTypeMap[ext] || 'application/octet-stream'
+	const base64 = fs.readFileSync(fullPath).toString('base64')
+	return {
+		dataUrl: `data:${mimeType};base64,${base64}`,
+		fullPath,
+	}
+}
+
+function getExtFromUrl(url) {
+	try {
+		const u = new URL(url)
+		const ext = path.extname(u.pathname || '').toLowerCase()
+		if (['.png', '.jpg', '.jpeg', '.webp', '.gif', '.bmp'].includes(ext)) {
+			return ext
+		}
+		return '.png'
+	} catch {
+		return '.png'
+	}
+}
+
+function getMediaExtFromUrl(url, fallbackExt = '.bin') {
+	try {
+		const u = new URL(url)
+		const ext = path.extname(u.pathname || '').toLowerCase()
+		return ext || fallbackExt
+	} catch {
+		return fallbackExt
+	}
+}
+
+function collectImageUrls(payload) {
+	const urls = new Set()
+
+	function visit(node) {
+		if (!node) return
+		if (typeof node === 'string') {
+			if (/^https?:\/\//i.test(node) && /\.(png|jpe?g|webp|gif|bmp)(\?|$)/i.test(node)) {
+				urls.add(node)
+			}
+			return
+		}
+		if (Array.isArray(node)) {
+			node.forEach((item) => visit(item))
+			return
+		}
+		if (typeof node === 'object') {
+			const directCandidates = [
+				node.url,
+				node.image_url,
+				node.imageUrl,
+				node.result_url,
+				node.resultUrl,
+			]
+			directCandidates.forEach((item) => visit(item))
+			Object.values(node).forEach((value) => visit(value))
+		}
+	}
+
+	visit(payload)
+	return Array.from(urls)
+}
+
+function collectMediaUrls(payload, type = 'video') {
+	const urls = new Set()
+	const regexByType = {
+		video: /\.(mp4|mov|m4v|webm|avi|mkv)(\?|$)/i,
+		audio: /\.(mp3|wav|pcm|opus|aac|flac|m4a)(\?|$)/i,
+	}
+	const matcher = regexByType[type] || regexByType.video
+
+	function visit(node) {
+		if (!node) return
+		if (typeof node === 'string') {
+			if (/^https?:\/\//i.test(node) && matcher.test(node)) {
+				urls.add(node)
+			}
+			return
+		}
+		if (Array.isArray(node)) {
+			node.forEach((item) => visit(item))
+			return
+		}
+		if (typeof node === 'object') {
+			Object.values(node).forEach((value) => visit(value))
+		}
+	}
+
+	visit(payload)
+	return Array.from(urls)
+}
+
+async function downloadImagesToCwd(ctx, imageUrls = []) {
+	const cwd = process.cwd()
+	const downloadedFiles = []
+
+	for (let i = 0; i < imageUrls.length; i += 1) {
+		const url = imageUrls[i]
+		const ext = getExtFromUrl(url)
+		const fileName = `bailian_image_${Date.now()}_${i + 1}${ext}`
+		const outputPath = path.join(cwd, fileName)
+		const response = await ctx.axios.get(url, {
+			responseType: 'arraybuffer',
+			timeout: 180000,
+		})
+		fs.writeFileSync(outputPath, response.data)
+		downloadedFiles.push(outputPath)
+	}
+
+	return downloadedFiles
+}
+
+async function downloadMediaToCwd(ctx, mediaUrls = [], prefix = 'bailian_media', fallbackExt = '.bin') {
+	const cwd = process.cwd()
+	const downloadedFiles = []
+
+	for (let i = 0; i < mediaUrls.length; i += 1) {
+		const url = mediaUrls[i]
+		const ext = getMediaExtFromUrl(url, fallbackExt)
+		const fileName = `${prefix}_${Date.now()}_${i + 1}${ext}`
+		const outputPath = path.join(cwd, fileName)
+		const response = await ctx.axios.get(url, {
+			responseType: 'arraybuffer',
+			timeout: 240000,
+		})
+		fs.writeFileSync(outputPath, response.data)
+		downloadedFiles.push(outputPath)
+	}
+
+	return downloadedFiles
+}
+
+async function postJson(ctx, url, body, timeout = 180000, headers = {}) {
+	const { apiKey } = getAiConfig(ctx)
+	const defaultHeaders = {
+		Authorization: `Bearer ${apiKey}`,
+		'Content-Type': 'application/json',
+		'X-DashScope-Async': 'enable', // 默认开启异步任务（适用于生图/生视频）
+		...headers,
+	}
+	const response = await ctx.axios.post(url, body, {
+		headers: defaultHeaders,
+		timeout,
+	})
+	return response.data
+}
+
+function getAudioExt(format = DEFAULT_TTS_FORMAT) {
+	const fmt = String(format || '').toLowerCase()
+	if (['mp3', 'wav', 'pcm', 'opus', 'aac', 'flac'].includes(fmt)) {
+		return fmt === 'pcm' ? 'pcm' : fmt
+	}
+	return DEFAULT_TTS_FORMAT
+}
+
+async function saveAudioToCwd(buffer, format = DEFAULT_TTS_FORMAT, outputFileName = '') {
+	const cwd = process.cwd()
+	const ext = getAudioExt(format)
+	const fileName = outputFileName
+		? outputFileName
+		: `bailian_audio_${Date.now()}.${ext}`
+	const outputPath = path.join(cwd, fileName)
+	fs.writeFileSync(outputPath, buffer)
+	return outputPath
+}
+
+async function saveTextToCwd(text, prefix = 'bailian_text', ext = '.txt') {
+	const outputPath = path.join(process.cwd(), `${prefix}_${Date.now()}${ext}`)
+	fs.writeFileSync(outputPath, String(text || ''), 'utf8')
+	return outputPath
+}
+
+async function saveJsonToCwd(data, prefix = 'bailian_result') {
+	const outputPath = path.join(process.cwd(), `${prefix}_${Date.now()}.json`)
+	fs.writeJsonSync(outputPath, data, { spaces: 2 })
+	return outputPath
+}
+
+async function getJson(ctx, url, timeout = 120000) {
+	const { apiKey } = getAiConfig(ctx)
+	const headers = {
+		Authorization: `Bearer ${apiKey}`,
+	}
+	const response = await ctx.axios.get(url, { headers, timeout })
+	return response.data
+}
+
+function parseTaskId(data) {
+	return (
+		data?.output?.task_id ||
+		data?.output?.taskId ||
+		data?.task_id ||
+		data?.taskId ||
+		data?.request_id ||
+		''
+	)
+}
+
+function parseTaskStatus(data) {
+	return (
+		data?.output?.task_status ||
+		data?.output?.taskStatus ||
+		data?.task_status ||
+		data?.taskStatus ||
+		''
+	)
+}
+
+async function pollTaskResult(ctx, taskId, intervalMs = 3000, maxPoll = 40) {
+	const { baseUrl } = getAiConfig(ctx)
+	const rootBaseUrl = getRootBaseUrl(baseUrl)
+	const taskUrl = joinUrl(rootBaseUrl, `api/v1/tasks/${taskId}`)
+
+	let lastData = null
+	for (let i = 0; i < maxPoll; i += 1) {
+		const data = await getJson(ctx, taskUrl)
+		lastData = data
+		const status = String(parseTaskStatus(data)).toUpperCase()
+		if (!status) {
+			await sleep(intervalMs)
+			continue
+		}
+		if (['SUCCEEDED', 'SUCCESS', 'COMPLETED'].includes(status)) {
+			return ok({
+				taskId,
+				status,
+				task: data,
+			})
+		}
+		if (['FAILED', 'CANCELED', 'CANCELLED', 'ERROR'].includes(status)) {
+			return fail('Task failed', {
+				taskId,
+				status,
+				task: data,
+			})
+		}
+		await sleep(intervalMs)
+	}
+
+	return fail('Task polling timeout', {
+		taskId,
+		task: lastData,
+	})
+}
+
+/**
+ * 阿里百炼文生图
+ * 接口: POST /api/v1/services/aigc/text2image/image-synthesis
+ */
+async function aliBailianTextToImage(
+	prompt,
+	model = MODEL_TEXT_TO_IMAGE,
+	size = '1024*1024',
+	n = 1,
+	autoPoll = true,
+	autoDownload = true,
+) {
+	try {
+		assertSkillAvailability(this)
+		if (!prompt) {
+			return fail('prompt is required')
+		}
+		const { baseUrl } = getAiConfig(this)
+		const rootBaseUrl = getRootBaseUrl(baseUrl)
+		const url = joinUrl(rootBaseUrl, 'api/v1/services/aigc/text2image/image-synthesis')
+		const body = {
+			model,
+			input: { prompt },
+			parameters: {
+				size,
+				n: Number(n) > 0 ? Number(n) : 1,
+			},
+		}
+		const data = await postJson(this, url, body)
+		const taskId = parseTaskId(data)
+
+		if (!autoPoll || !taskId) {
+			const imageUrls = collectImageUrls(data)
+			let downloadedFiles = []
+			const savedResultFile = await saveJsonToCwd(data, 'bailian_image_result')
+			if (autoDownload && imageUrls.length > 0) {
+				downloadedFiles = await downloadImagesToCwd(this, imageUrls)
+			}
+			return ok({
+				taskId,
+				status: parseTaskStatus(data) || 'PENDING',
+				imageUrls,
+				downloadedFiles,
+				savedResultFile,
+				response: data,
+			})
+		}
+		const pollResult = await pollTaskResult(this, taskId)
+		if (!pollResult.success) {
+			return pollResult
+		}
+		const imageUrls = collectImageUrls(pollResult.data?.task)
+		let downloadedFiles = []
+		const savedResultFile = await saveJsonToCwd(
+			pollResult.data?.task || {},
+			'bailian_image_result',
+		)
+		if (autoDownload && imageUrls.length > 0) {
+			downloadedFiles = await downloadImagesToCwd(this, imageUrls)
+		}
+		return ok({
+			...pollResult.data,
+			imageUrls,
+			downloadedFiles,
+			savedResultFile,
+		})
+	} catch (error) {
+		return fail(error, { prompt, model, size, n, autoPoll, autoDownload })
+	}
+}
+
+/**
+ * 阿里百炼文生视频
+ * 接口: POST /api/v1/services/aigc/video-generation/video-synthesis
+ */
+async function aliBailianTextToVideo(
+	prompt,
+	model = MODEL_TEXT_TO_VIDEO,
+	size = '1280*720',
+	duration = 5,
+	autoPoll = true,
+	autoDownload = true,
+) {
+	try {
+		assertSkillAvailability(this)
+		if (!prompt) {
+			return fail('prompt is required')
+		}
+		const { baseUrl } = getAiConfig(this)
+		const rootBaseUrl = getRootBaseUrl(baseUrl)
+		const url = joinUrl(rootBaseUrl, 'api/v1/services/aigc/video-generation/video-synthesis')
+		const body = {
+			model,
+			input: { prompt },
+			parameters: {
+				size,
+				duration: Number(duration) > 0 ? Number(duration) : 5,
+			},
+		}
+		const data = await postJson(this, url, body, 240000)
+		const taskId = parseTaskId(data)
+
+		if (!autoPoll || !taskId) {
+			const videoUrls = collectMediaUrls(data, 'video')
+			let downloadedFiles = []
+			const savedResultFile = await saveJsonToCwd(data, 'bailian_video_result')
+			if (autoDownload && videoUrls.length > 0) {
+				downloadedFiles = await downloadMediaToCwd(this, videoUrls, 'bailian_video', '.mp4')
+			}
+			return ok({
+				taskId,
+				status: parseTaskStatus(data) || 'PENDING',
+				videoUrls,
+				downloadedFiles,
+				savedResultFile,
+				response: data,
+			})
+		}
+		const pollResult = await pollTaskResult(this, taskId, 5000, 60)
+		if (!pollResult.success) {
+			return pollResult
+		}
+		const videoUrls = collectMediaUrls(pollResult.data?.task, 'video')
+		let downloadedFiles = []
+		const savedResultFile = await saveJsonToCwd(
+			pollResult.data?.task || {},
+			'bailian_video_result',
+		)
+		if (autoDownload && videoUrls.length > 0) {
+			downloadedFiles = await downloadMediaToCwd(this, videoUrls, 'bailian_video', '.mp4')
+		}
+		return ok({
+			...pollResult.data,
+			videoUrls,
+			downloadedFiles,
+			savedResultFile,
+		})
+	} catch (error) {
+		return fail(error, { prompt, model, size, duration, autoPoll, autoDownload })
+	}
+}
+
+/**
+ * 阿里百炼图片识别 (多模态对话)
+ * 接口: POST /compatible-mode/v1/chat/completions
+ */
+async function aliBailianImageRecognition(
+	imagePath,
+	question = '请详细描述这张图片内容。',
+	model = MODEL_IMAGE_RECOGNITION,
+	autoSave = true,
+	outputFileName = '',
+) {
+	try {
+		assertSkillAvailability(this)
+		if (!imagePath) {
+			return fail('imagePath is required')
+		}
+
+		const { dataUrl, fullPath } = buildImageDataUrl(imagePath)
+		const { baseUrl } = getAiConfig(this)
+		// 直接使用 baseUrl，因为 baseUrl 通常是 compatible-mode/v1
+		const chatUrl = joinUrl(baseUrl, 'chat/completions')
+
+		const body = {
+			model,
+			messages: [
+				{
+					role: 'user',
+					content: [
+						{
+							type: 'text',
+							text: question,
+						},
+						{
+							type: 'image_url',
+							image_url: {
+								url: dataUrl,
+							},
+						},
+					],
+				},
+			],
+			stream: false,
+		}
+
+		// 多模态对话通常不需要 X-DashScope-Async 头，直接同步返回
+		const { apiKey } = getAiConfig(this)
+		const headers = {
+			Authorization: `Bearer ${apiKey}`,
+			'Content-Type': 'application/json',
+		}
+		const response = await this.axios.post(chatUrl, body, { headers, timeout: 180000 })
+		const data = response.data
+		const content = data?.choices?.[0]?.message?.content || ''
+		let savedTextFile = ''
+		let savedResultFile = ''
+		if (autoSave) {
+			savedTextFile = outputFileName
+				? path.join(process.cwd(), outputFileName)
+				: await saveTextToCwd(content, 'bailian_image_recognition', '.txt')
+			if (outputFileName) {
+				fs.writeFileSync(savedTextFile, String(content || ''), 'utf8')
+			}
+			savedResultFile = await saveJsonToCwd(data, 'bailian_image_recognition_result')
+		}
+		return ok({
+			imagePath: fullPath,
+			question,
+			model,
+			content,
+			savedTextFile,
+			savedResultFile,
+			response: data,
+		})
+	} catch (error) {
+		return fail(error, { imagePath, question, model, autoSave, outputFileName })
+	}
+}
+
+/**
+ * 阿里百炼文字转语音 (OpenAI 兼容音频接口)
+ * 接口: POST /audio/speech
+ */
+async function aliBailianTextToSpeech(
+	text,
+	model = MODEL_TEXT_TO_SPEECH,
+	voice = DEFAULT_TTS_VOICE,
+	format = DEFAULT_TTS_FORMAT,
+	autoDownload = true,
+	outputFileName = '',
+) {
+	try {
+		assertSkillAvailability(this)
+		if (!text) {
+			return fail('text is required')
+		}
+		const { baseUrl, apiKey } = getAiConfig(this)
+		const url = joinUrl(baseUrl, 'audio/speech')
+		const response = await this.axios.post(
+			url,
+			{
+				model,
+				input: text,
+				voice,
+				format,
+			},
+			{
+				headers: {
+					Authorization: `Bearer ${apiKey}`,
+					'Content-Type': 'application/json',
+				},
+				responseType: 'arraybuffer',
+				timeout: 180000,
+			},
+		)
+
+		const audioBuffer = Buffer.from(response.data)
+		let downloadedFile = ''
+		if (autoDownload) {
+			downloadedFile = await saveAudioToCwd(audioBuffer, format, outputFileName)
+		}
+
+		return ok({
+			model,
+			voice,
+			format: getAudioExt(format),
+			audioSize: audioBuffer.length,
+			downloadedFile,
+		})
+	} catch (error) {
+		return fail(error, {
+			text,
+			model,
+			voice,
+			format,
+			autoDownload,
+			outputFileName,
+		})
+	}
+}
+
+/**
+ * 阿里百炼语音转文字 (OpenAI 兼容音频接口)
+ * 接口: POST /audio/transcriptions
+ */
+async function aliBailianSpeechToText(
+	audioPath,
+	model = MODEL_SPEECH_TO_TEXT,
+	language = 'zh',
+	prompt = '',
+	autoSave = true,
+	outputFileName = '',
+) {
+	try {
+		assertSkillAvailability(this)
+		if (!audioPath) {
+			return fail('audioPath is required')
+		}
+		const fullPath = path.resolve(process.cwd(), audioPath)
+		if (!fs.existsSync(fullPath)) {
+			return fail(`Audio file does not exist: ${fullPath}`)
+		}
+		if (typeof FormData === 'undefined' || typeof Blob === 'undefined') {
+			return fail('Current Node.js runtime does not support FormData/Blob for audio upload')
+		}
+
+		const { baseUrl, apiKey } = getAiConfig(this)
+		const url = joinUrl(baseUrl, 'audio/transcriptions')
+		const fileBuffer = fs.readFileSync(fullPath)
+		const form = new FormData()
+		form.append('file', new Blob([fileBuffer]), path.basename(fullPath))
+		form.append('model', model)
+		if (language) {
+			form.append('language', language)
+		}
+		if (prompt) {
+			form.append('prompt', prompt)
+		}
+
+		const extraHeaders =
+			typeof form.getHeaders === 'function' ? form.getHeaders() : {}
+		const response = await this.axios.post(url, form, {
+			headers: {
+				Authorization: `Bearer ${apiKey}`,
+				...extraHeaders,
+			},
+			timeout: 180000,
+		})
+		const data = response.data || {}
+		let savedTextFile = ''
+		let savedResultFile = ''
+		if (autoSave) {
+			savedTextFile = outputFileName
+				? path.join(process.cwd(), outputFileName)
+				: await saveTextToCwd(data.text || '', 'bailian_stt', '.txt')
+			if (outputFileName) {
+				fs.writeFileSync(savedTextFile, String(data.text || ''), 'utf8')
+			}
+			savedResultFile = await saveJsonToCwd(data, 'bailian_stt_result')
+		}
+
+		return ok({
+			audioPath: fullPath,
+			model,
+			language,
+			text: data.text || '',
+			savedTextFile,
+			savedResultFile,
+			response: data,
+		})
+	} catch (error) {
+		return fail(error, { audioPath, model, language, prompt, autoSave, outputFileName })
+	}
+}
+
+async function aliBailianGetTaskResult(taskId, autoDownload = true) {
+	try {
+		assertSkillAvailability(this)
+		if (!taskId) {
+			return fail('taskId is required')
+		}
+		const result = await pollTaskResult(this, taskId, 0, 1)
+		if (!result.success) {
+			return result
+		}
+		const taskPayload = result.data?.task || {}
+		const imageUrls = collectImageUrls(taskPayload)
+		const videoUrls = collectMediaUrls(taskPayload, 'video')
+		const savedResultFile = await saveJsonToCwd(taskPayload, 'bailian_task_result')
+		let downloadedImageFiles = []
+		let downloadedVideoFiles = []
+		if (autoDownload && imageUrls.length > 0) {
+			downloadedImageFiles = await downloadImagesToCwd(this, imageUrls)
+		}
+		if (autoDownload && videoUrls.length > 0) {
+			downloadedVideoFiles = await downloadMediaToCwd(this, videoUrls, 'bailian_video', '.mp4')
+		}
+		return ok({
+			...result.data,
+			imageUrls,
+			videoUrls,
+			downloadedImageFiles,
+			downloadedVideoFiles,
+			savedResultFile,
+		})
+	} catch (error) {
+		return fail(error, { taskId, autoDownload })
+	}
+}
+
+function aliBailianReadme() {
+	try {
+		const checkResult = checkSkillAvailability(this)
+		const statusText = checkResult.canUse ? 'Available' : 'Unavailable'
+		const usageNote = checkResult.canUse
+			? 'You can call aliBailianTextToImage, aliBailianTextToVideo, aliBailianImageRecognition and aliBailianGetTaskResult directly.'
+			: `This skill is disabled because aiConfig.baseUrl must contain 'dashscope.aliyuncs.com'. Please update your AI config first.`
+
+		return `# AliBailian Skill Guide\n\n- Status: ${statusText}\n- Current baseUrl: ${checkResult.baseUrl || '(empty)'}\n- Required baseUrl: ${checkResult.requiredBaseUrl}\n\n${usageNote}`
+	} catch (error) {
+		return `# AliBailian Skill Guide\n\n- Status: Unavailable\n- Reason: ${error?.message || String(error)}`
+	}
+}
+
+const descriptions = [
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianReadme',
+			description:
+				'阿里百炼说明与可用性检测：先判断当前aiConfig.baseUrl是否配置正确，再返回工具说明。调用本工具前建议先调用此函数确认可用性。',
+			parameters: {
+				type: 'object',
+				properties: {},
+				required: [],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianTextToSpeech',
+			description:
+				'阿里百炼文字转语音：将文本合成为语音。默认自动下载到当前工作目录。',
+			parameters: {
+				type: 'object',
+				properties: {
+					text: {
+						type: 'string',
+						description: '要合成语音的文本内容。',
+					},
+					model: {
+						type: 'string',
+						description: `文字转语音模型名，默认 ${MODEL_TEXT_TO_SPEECH}。`,
+					},
+					voice: {
+						type: 'string',
+						description: `音色名称，默认 ${DEFAULT_TTS_VOICE}。`,
+					},
+					format: {
+						type: 'string',
+						description: `音频格式（mp3/wav/opus 等），默认 ${DEFAULT_TTS_FORMAT}。`,
+					},
+					autoDownload: {
+						type: 'boolean',
+						description: '是否自动下载到当前工作目录，默认 true。',
+					},
+					outputFileName: {
+						type: 'string',
+						description: '可选，自定义输出文件名（如 my_tts.mp3）。',
+					},
+				},
+				required: ['text'],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianSpeechToText',
+			description:
+				'阿里百炼语音转文字：上传本地音频文件并返回识别文本。',
+			parameters: {
+				type: 'object',
+				properties: {
+					audioPath: {
+						type: 'string',
+						description: '本地音频路径（相对路径或绝对路径）。',
+					},
+					model: {
+						type: 'string',
+						description: `语音转文字模型名，默认 ${MODEL_SPEECH_TO_TEXT}。`,
+					},
+					language: {
+						type: 'string',
+						description: '识别语言，默认 zh。',
+					},
+					prompt: {
+						type: 'string',
+						description: '可选，识别提示词。',
+					},
+					autoSave: {
+						type: 'boolean',
+						description: '是否自动保存识别文本和原始结果到当前工作目录，默认 true。',
+					},
+					outputFileName: {
+						type: 'string',
+						description: '可选，自定义转写文本文件名。',
+					},
+				},
+				required: ['audioPath'],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianTextToImage',
+			description:
+				'阿里百炼文生图：根据文本提示词生成图片。自动读取 this.agentRobot.aiConfig 中的 baseUrl 和 apiKey 进行调用。',
+			parameters: {
+				type: 'object',
+				properties: {
+					prompt: {
+						type: 'string',
+						description: '图片提示词。',
+					},
+					model: {
+						type: 'string',
+						description: `文生图模型名，默认 ${MODEL_TEXT_TO_IMAGE}。`,
+					},
+					size: {
+						type: 'string',
+						description: '图片尺寸，默认 1024*1024。',
+					},
+					n: {
+						type: 'number',
+						description: '生成图片数量，默认 1。',
+					},
+					autoPoll: {
+						type: 'boolean',
+						description: '是否自动轮询任务结果，默认 true。',
+					},
+					autoDownload: {
+						type: 'boolean',
+						description: '是否自动下载生成图片到当前工作目录，默认 true。',
+					},
+				},
+				required: ['prompt'],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianTextToVideo',
+			description:
+				'阿里百炼文生视频：根据文本提示词生成视频。自动读取 this.agentRobot.aiConfig 中的 baseUrl 和 apiKey 进行调用。',
+			parameters: {
+				type: 'object',
+				properties: {
+					prompt: {
+						type: 'string',
+						description: '视频提示词。',
+					},
+					model: {
+						type: 'string',
+						description: `文生视频模型名，默认 ${MODEL_TEXT_TO_VIDEO}。`,
+					},
+					size: {
+						type: 'string',
+						description: '视频尺寸，默认 1280*720。',
+					},
+					duration: {
+						type: 'number',
+						description: '视频时长（秒），默认 5。',
+					},
+					autoPoll: {
+						type: 'boolean',
+						description: '是否自动轮询任务结果，默认 true。',
+					},
+					autoDownload: {
+						type: 'boolean',
+						description: '是否自动下载生成视频到当前工作目录，默认 true。',
+					},
+				},
+				required: ['prompt'],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianImageRecognition',
+			description:
+				'阿里百炼图片识别：输入本地图片路径和问题，返回识别与理解结果。自动读取 this.agentRobot.aiConfig 中的 baseUrl 和 apiKey 进行调用。',
+			parameters: {
+				type: 'object',
+				properties: {
+					imagePath: {
+						type: 'string',
+						description: '本地图片路径（相对路径或绝对路径）。',
+					},
+					question: {
+						type: 'string',
+						description: '关于图片的问题，默认“请详细描述这张图片内容。”。',
+					},
+					model: {
+						type: 'string',
+						description: `多模态模型名，默认 ${MODEL_IMAGE_RECOGNITION}。`,
+					},
+					autoSave: {
+						type: 'boolean',
+						description: '是否自动保存识别文本和原始结果到当前工作目录，默认 true。',
+					},
+					outputFileName: {
+						type: 'string',
+						description: '可选，自定义识别文本文件名。',
+					},
+				},
+				required: ['imagePath'],
+			},
+		},
+	},
+	{
+		type: 'function',
+		function: {
+			name: 'aliBailianGetTaskResult',
+			description: '阿里百炼任务查询：根据 taskId 查询一次任务当前状态与结果。',
+			parameters: {
+				type: 'object',
+				properties: {
+					taskId: {
+						type: 'string',
+						description: '阿里百炼异步任务ID。',
+					},
+					autoDownload: {
+						type: 'boolean',
+						description: '是否自动下载任务中的图片/视频并保存结果JSON到当前工作目录，默认 true。',
+					},
+				},
+				required: ['taskId'],
+			},
+		},
+	},
+]
+
+const functions = {
+	aliBailianReadme,
+	aliBailianTextToSpeech,
+	aliBailianSpeechToText,
+	aliBailianTextToImage,
+	aliBailianTextToVideo,
+	aliBailianImageRecognition,
+	aliBailianGetTaskResult,
+}
+
+const AliBailianTool = {
+	name: 'AliBailianTool',
+	description:
+		'提供阿里百炼多模态能力，包括文生图、文生视频、图片识别与任务查询，统一使用当前AI配置的baseUrl/apiKey。',
+	platform: 'all',
+	descriptions,
+	functions,
+	isSystem: true,
+}
+
+module.exports = AliBailianTool