@aituber-onair/chat 0.1.0

package/dist/cjs/services/providers/gemini/GeminiChatService.js

@@ -0,0 +1,653 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.GeminiChatService = void 0;
+const constants_1 = require("../../../constants");
+const chat_1 = require("../../../constants/chat");
+const streamTextAccumulator_1 = require("../../../utils/streamTextAccumulator");
+const chatServiceHttpClient_1 = require("../../../utils/chatServiceHttpClient");
+const mcpSchemaFetcher_1 = require("../../../utils/mcpSchemaFetcher");
+/**
+ * Gemini implementation of ChatService
+ */
+class GeminiChatService {
+    /* ────────────────────────────────── */
+    /*  Utilities                           */
+    /* ────────────────────────────────── */
+    safeJsonParse(str) {
+        try {
+            return JSON.parse(str);
+        }
+        catch {
+            return str; // keep as string
+        }
+    }
+    normalizeToolResult(val) {
+        if (val === null)
+            return { content: null };
+        if (typeof val === 'object')
+            return val;
+        return { content: val }; // wrap primitive
+    }
+    /**
+     * camelCase → snake_case conversion (v1beta)
+     */
+    adaptKeysForApi(obj) {
+        const map = {
+            toolConfig: 'tool_config',
+            functionCallingConfig: 'function_calling_config',
+            functionDeclarations: 'function_declarations',
+            functionCall: 'function_call',
+            functionResponse: 'function_response',
+        };
+        if (Array.isArray(obj))
+            return obj.map((v) => this.adaptKeysForApi(v));
+        if (obj && typeof obj === 'object') {
+            return Object.fromEntries(Object.entries(obj).map(([k, v]) => [
+                map[k] ?? k,
+                this.adaptKeysForApi(v),
+            ]));
+        }
+        return obj;
+    }
+    /**
+     * Constructor
+     * @param apiKey Google API key
+     * @param model Name of the model to use
+     * @param visionModel Name of the vision model
+     * @param tools Array of tool definitions
+     * @param mcpServers Array of MCP server configurations
+     */
+    constructor(apiKey, model = constants_1.MODEL_GEMINI_2_0_FLASH_LITE, visionModel = constants_1.MODEL_GEMINI_2_0_FLASH_LITE, tools = [], mcpServers = [], responseLength) {
+        /** Provider name */
+        this.provider = 'gemini';
+        this.mcpToolSchemas = [];
+        this.mcpSchemasInitialized = false;
+        /** id(OpenAI) → name(Gemini) mapping */
+        this.callIdMap = new Map();
+        this.apiKey = apiKey;
+        this.model = model;
+        this.responseLength = responseLength;
+        // check if the vision model is supported
+        if (!constants_1.GEMINI_VISION_SUPPORTED_MODELS.includes(visionModel)) {
+            throw new Error(`Model ${visionModel} does not support vision capabilities.`);
+        }
+        this.visionModel = visionModel;
+        this.tools = tools;
+        this.mcpServers = mcpServers;
+    }
+    /**
+     * Get the current model name
+     * @returns Model name
+     */
+    getModel() {
+        return this.model;
+    }
+    /**
+     * Get the current vision model name
+     * @returns Vision model name
+     */
+    getVisionModel() {
+        return this.visionModel;
+    }
+    /**
+     * Get configured MCP servers
+     * @returns Array of MCP server configurations
+     */
+    getMCPServers() {
+        return this.mcpServers;
+    }
+    /**
+     * Add MCP server configuration
+     * @param serverConfig MCP server configuration
+     */
+    addMCPServer(serverConfig) {
+        this.mcpServers.push(serverConfig);
+        // Reset initialization flag to re-fetch schemas
+        this.mcpSchemasInitialized = false;
+    }
+    /**
+     * Remove MCP server by name
+     * @param serverName Name of the server to remove
+     */
+    removeMCPServer(serverName) {
+        this.mcpServers = this.mcpServers.filter((server) => server.name !== serverName);
+        // Reset initialization flag to re-fetch schemas
+        this.mcpSchemasInitialized = false;
+    }
+    /**
+     * Check if MCP servers are configured
+     * @returns True if MCP servers are configured
+     */
+    hasMCPServers() {
+        return this.mcpServers.length > 0;
+    }
+    /**
+     * Initialize MCP tool schemas by fetching from servers
+     * @private
+     */
+    async initializeMCPSchemas() {
+        if (this.mcpSchemasInitialized || this.mcpServers.length === 0) {
+            return;
+        }
+        try {
+            // Add timeout to prevent hanging
+            const timeoutPromise = new Promise((_, reject) => setTimeout(() => reject(new Error('MCP schema fetch timeout')), 5000));
+            const schemasPromise = mcpSchemaFetcher_1.MCPSchemaFetcher.fetchAllToolSchemas(this.mcpServers);
+            this.mcpToolSchemas = await Promise.race([
+                schemasPromise,
+                timeoutPromise,
+            ]);
+            this.mcpSchemasInitialized = true;
+        }
+        catch (error) {
+            console.warn('Failed to initialize MCP schemas, using fallback:', error);
+            // Use fallback schemas - always provide basic functionality
+            this.mcpToolSchemas = this.mcpServers.map((server) => ({
+                name: `mcp_${server.name}_search`,
+                description: `Search using ${server.name} MCP server (fallback)`,
+                parameters: {
+                    type: 'object',
+                    properties: {
+                        query: {
+                            type: 'string',
+                            description: 'Search query',
+                        },
+                    },
+                    required: ['query'],
+                },
+            }));
+            this.mcpSchemasInitialized = true;
+        }
+    }
+    /**
+     * Process chat messages
+     * @param messages Array of messages to send
+     * @param onPartialResponse Callback to receive each part of streaming response
+     * @param onCompleteResponse Callback to execute when response is complete
+     */
+    async processChat(messages, onPartialResponse, onCompleteResponse) {
+        try {
+            // not use tools or MCP servers
+            if (this.tools.length === 0 && this.mcpServers.length === 0) {
+                const res = await this.callGemini(messages, this.model, true);
+                const { blocks } = await this.parseStream(res, onPartialResponse);
+                const full = blocks
+                    .filter((b) => b.type === 'text')
+                    .map((b) => b.text)
+                    .join('');
+                await onCompleteResponse(full);
+                return;
+            }
+            /* with tools (1 turn) */
+            const { blocks, stop_reason } = await this.chatOnce(messages, true, onPartialResponse);
+            if (stop_reason === 'end') {
+                const full = blocks
+                    .filter((b) => b.type === 'text')
+                    .map((b) => b.text)
+                    .join('');
+                await onCompleteResponse(full);
+                return;
+            }
+            throw new Error('Received functionCall. Use chatOnce() loop when tools are enabled.');
+        }
+        catch (err) {
+            console.error('Error in processChat:', err);
+            throw err;
+        }
+    }
+    async processVisionChat(messages, onPartialResponse, onCompleteResponse) {
+        try {
+            if (this.tools.length === 0 && this.mcpServers.length === 0) {
+                const res = await this.callGemini(messages, this.visionModel, true);
+                const { blocks } = await this.parseStream(res, onPartialResponse);
+                const full = blocks
+                    .filter((b) => b.type === 'text')
+                    .map((b) => b.text)
+                    .join('');
+                await onCompleteResponse(full);
+                return;
+            }
+            const { blocks, stop_reason } = await this.visionChatOnce(messages);
+            blocks
+                .filter((b) => b.type === 'text')
+                .forEach((b) => onPartialResponse(b.text));
+            if (stop_reason === 'end') {
+                const full = blocks
+                    .filter((b) => b.type === 'text')
+                    .map((b) => b.text)
+                    .join('');
+                await onCompleteResponse(full);
+                return;
+            }
+            throw new Error('Received functionCall. Use visionChatOnce() loop when tools are enabled.');
+        }
+        catch (err) {
+            console.error('Error in processVisionChat:', err);
+            throw err;
+        }
+    }
+    /* ────────────────────────────────── */
+    /*  OpenAI → Gemini conversion         */
+    /* ────────────────────────────────── */
+    convertMessagesToGeminiFormat(messages) {
+        const gemini = [];
+        let currentRole = null;
+        let currentParts = [];
+        const pushCurrent = () => {
+            if (currentRole && currentParts.length) {
+                gemini.push({ role: currentRole, parts: [...currentParts] });
+                currentParts = [];
+            }
+        };
+        for (const msg of messages) {
+            const role = this.mapRoleToGemini(msg.role);
+            /* assistant: tool_calls -> functionCall */
+            if (msg.tool_calls) {
+                pushCurrent();
+                for (const call of msg.tool_calls) {
+                    this.callIdMap.set(call.id, call.function.name);
+                    gemini.push({
+                        role: 'model',
+                        parts: [
+                            {
+                                functionCall: {
+                                    name: call.function.name,
+                                    args: JSON.parse(call.function.arguments || '{}'),
+                                },
+                            },
+                        ],
+                    });
+                }
+                continue;
+            }
+            /* tool → functionResponse */
+            if (msg.role === 'tool') {
+                pushCurrent();
+                const funcName = msg.name ??
+                    this.callIdMap.get(msg.tool_call_id) ??
+                    'result';
+                gemini.push({
+                    role: 'user',
+                    parts: [
+                        {
+                            functionResponse: {
+                                name: funcName,
+                                response: this.normalizeToolResult(this.safeJsonParse(msg.content)),
+                            },
+                        },
+                    ],
+                });
+                continue;
+            }
+            /* normal text */
+            if (role !== currentRole)
+                pushCurrent();
+            currentRole = role;
+            currentParts.push({ text: msg.content });
+        }
+        pushCurrent();
+        return gemini;
+    }
+    /* ────────────────────────────────── */
+    /*  HTTP call                           */
+    /* ────────────────────────────────── */
+    async callGemini(messages, model, stream = false, maxTokens) {
+        const hasVision = messages.some((m) => Array.isArray(m.content) &&
+            m.content.some((b) => b?.type === 'image_url' || b?.inlineData));
+        const contents = hasVision
+            ? await this.convertVisionMessagesToGeminiFormat(messages)
+            : this.convertMessagesToGeminiFormat(messages);
+        const body = {
+            contents,
+            generationConfig: {
+                maxOutputTokens: maxTokens !== undefined
+                    ? maxTokens
+                    : (0, chat_1.getMaxTokensForResponseLength)(this.responseLength),
+            },
+        };
+        // Add tools configuration (regular tools + MCP tools as functionDeclarations)
+        const allToolDeclarations = [];
+        // Add regular function tools
+        if (this.tools.length > 0) {
+            allToolDeclarations.push(...this.tools.map((t) => ({
+                name: t.name,
+                description: t.description,
+                parameters: t.parameters,
+            })));
+        }
+        // Add MCP tools as functionDeclarations
+        if (this.mcpServers.length > 0) {
+            try {
+                // Initialize MCP schemas if not already done
+                await this.initializeMCPSchemas();
+                // Add MCP tool schemas as regular function declarations
+                // Gemini will call these as normal functions, and ToolExecutor will handle the MCP routing
+                allToolDeclarations.push(...this.mcpToolSchemas.map((t) => ({
+                    name: t.name,
+                    description: t.description,
+                    parameters: t.parameters,
+                })));
+            }
+            catch (error) {
+                console.warn('MCP initialization failed, skipping MCP tools:', error);
+                // Continue without MCP tools if initialization fails
+            }
+        }
+        if (allToolDeclarations.length > 0) {
+            body.tools = [
+                {
+                    functionDeclarations: allToolDeclarations,
+                },
+            ];
+            body.toolConfig = { functionCallingConfig: { mode: 'AUTO' } };
+        }
+        const fetchOnce = async (ver, payload) => {
+            const fn = stream ? 'streamGenerateContent' : 'generateContent';
+            const alt = stream ? '?alt=sse' : '';
+            const url = `${constants_1.ENDPOINT_GEMINI_API}/${ver}/models/${model}:${fn}${alt}${alt ? '&' : '?'}key=${this.apiKey}`;
+            return chatServiceHttpClient_1.ChatServiceHttpClient.post(url, payload);
+        };
+        const isLite = /flash[-_]lite/.test(model);
+        const isGemini25 = /gemini-2\.5/.test(model);
+        const firstVer = isLite || isGemini25 ? 'v1beta' : 'v1';
+        const tryApi = async () => {
+            try {
+                const payload = firstVer === 'v1' ? body : this.adaptKeysForApi(body); // snake_case conversion
+                return await fetchOnce(firstVer, payload);
+            }
+            catch (e) {
+                // Only retry v1beta if camel/snake case mismatch error occurs in models that don't require v1beta
+                if (!(isLite || isGemini25) &&
+                    /Unknown name|Cannot find field|404/.test(e.message)) {
+                    return await fetchOnce('v1beta', this.adaptKeysForApi(body));
+                }
+                throw e; // otherwise, throw to upper layer
+            }
+        };
+        try {
+            const res = await tryApi();
+            return res;
+        }
+        catch (error) {
+            // Enhanced error logging for debugging
+            if (error.body) {
+                console.error('Gemini API Error Details:', error.body);
+                console.error('Request Body:', JSON.stringify(body, null, 2));
+            }
+            throw error;
+        }
+    }
+    /**
+     * Convert AITuber OnAir vision messages to Gemini format
+     * @param messages Array of vision messages
+     * @returns Gemini formatted vision messages
+     */
+    async convertVisionMessagesToGeminiFormat(messages) {
+        const geminiMessages = [];
+        let currentRole = null;
+        let currentParts = [];
+        for (const msg of messages) {
+            // Map AITuber OnAir roles to Gemini roles
+            const role = this.mapRoleToGemini(msg.role);
+            /* ----------- OpenAI compatible tool metadata ----------- */
+            // assistant: { tool_calls:[{id,name,function:{arguments}}] }
+            if (msg.tool_calls) {
+                for (const call of msg.tool_calls) {
+                    // Gemini does not need id. Insert functionCall into parts
+                    geminiMessages.push({
+                        role: 'model',
+                        parts: [
+                            {
+                                functionCall: {
+                                    name: call.function.name,
+                                    args: JSON.parse(call.function.arguments || '{}'),
+                                },
+                            },
+                        ],
+                    });
+                }
+                continue;
+            }
+            // tool role → user role + functionResponse
+            if (msg.role === 'tool') {
+                const funcName = msg.name ??
+                    this.callIdMap.get(msg.tool_call_id) ??
+                    'result';
+                geminiMessages.push({
+                    role: 'user',
+                    parts: [
+                        {
+                            functionResponse: {
+                                name: funcName,
+                                response: this.normalizeToolResult(this.safeJsonParse(msg.content)),
+                            },
+                        },
+                    ],
+                });
+                continue;
+            }
+            // If role changes, start a new message
+            if (role !== currentRole && currentParts.length > 0) {
+                geminiMessages.push({
+                    role: currentRole,
+                    parts: [...currentParts],
+                });
+                currentParts = [];
+            }
+            currentRole = role;
+            // If the message has content blocks, process them
+            if (typeof msg.content === 'string') {
+                currentParts.push({ text: msg.content });
+            }
+            else if (Array.isArray(msg.content)) {
+                // Process each content block (text or image)
+                for (const block of msg.content) {
+                    if (block.type === 'text') {
+                        currentParts.push({ text: block.text });
+                    }
+                    else if (block.type === 'image_url') {
+                        try {
+                            // Fetch the image data from URL
+                            const imageResponse = await chatServiceHttpClient_1.ChatServiceHttpClient.get(block.image_url.url);
+                            // Convert image to blob and then to base64
+                            const imageBlob = await imageResponse.blob();
+                            const base64Data = await this.blobToBase64(imageBlob);
+                            // Add image data in Gemini format
+                            currentParts.push({
+                                inlineData: {
+                                    mimeType: imageBlob.type || 'image/jpeg',
+                                    data: base64Data.split(',')[1], // Remove the "data:image/jpeg;base64," prefix
+                                },
+                            });
+                        }
+                        catch (error) {
+                            console.error('Error processing image:', error);
+                            throw new Error(`Failed to process image: ${error.message}`);
+                        }
+                    }
+                }
+            }
+        }
+        // Add the last message
+        if (currentRole && currentParts.length > 0) {
+            geminiMessages.push({
+                role: currentRole,
+                parts: [...currentParts],
+            });
+        }
+        return geminiMessages;
+    }
+    /**
+     * Convert Blob to Base64 string
+     * @param blob Image blob
+     * @returns Promise with base64 encoded string
+     */
+    blobToBase64(blob) {
+        return new Promise((resolve, reject) => {
+            const reader = new FileReader();
+            reader.onloadend = () => resolve(reader.result);
+            reader.onerror = reject;
+            reader.readAsDataURL(blob);
+        });
+    }
+    /**
+     * Map AITuber OnAir roles to Gemini roles
+     * @param role AITuber OnAir role
+     * @returns Gemini role
+     */
+    mapRoleToGemini(role) {
+        switch (role) {
+            case 'system':
+                return 'model'; // Gemini uses 'model' for system messages
+            case 'user':
+                return 'user';
+            case 'assistant':
+                return 'model';
+            default:
+                return 'user';
+        }
+    }
+    /* ────────────────────────────────────────────────────────── */
+    /*  Convert NDJSON stream to common format             */
+    /* ────────────────────────────────────────────────────────── */
+    async parseStream(res, onPartial) {
+        const reader = res.body.getReader();
+        const dec = new TextDecoder();
+        const textBlocks = [];
+        const toolBlocks = [];
+        let buf = '';
+        const flush = (payload) => {
+            if (!payload || payload === '[DONE]')
+                return;
+            let obj;
+            try {
+                obj = JSON.parse(payload);
+            }
+            catch {
+                return;
+            }
+            for (const cand of obj.candidates ?? []) {
+                for (const part of cand.content?.parts ?? []) {
+                    if (part.text) {
+                        onPartial(part.text);
+                        streamTextAccumulator_1.StreamTextAccumulator.addTextBlock(textBlocks, part.text);
+                    }
+                    if (part.functionCall) {
+                        toolBlocks.push({
+                            type: 'tool_use',
+                            id: this.genUUID(),
+                            name: part.functionCall.name,
+                            input: part.functionCall.args ?? {},
+                        });
+                    }
+                    if (part.functionResponse) {
+                        toolBlocks.push({
+                            type: 'tool_result',
+                            tool_use_id: part.functionResponse.name,
+                            content: JSON.stringify(part.functionResponse.response),
+                        });
+                    }
+                }
+            }
+        };
+        while (true) {
+            const { done, value } = await reader.read();
+            if (done)
+                break;
+            buf += dec.decode(value, { stream: true });
+            let nl;
+            while ((nl = buf.indexOf('\n')) !== -1) {
+                let line = buf.slice(0, nl);
+                buf = buf.slice(nl + 1);
+                if (line.endsWith('\r'))
+                    line = line.slice(0, -1); // CRLF support
+                if (!line.trim()) {
+                    flush('');
+                    continue;
+                } // keep-alive empty line
+                if (line.startsWith('data:'))
+                    line = line.slice(5).trim();
+                if (!line)
+                    continue;
+                flush(line);
+            }
+        }
+        if (buf)
+            flush(buf);
+        const blocks = [...textBlocks, ...toolBlocks];
+        return {
+            blocks,
+            stop_reason: toolBlocks.some((b) => b.type === 'tool_use')
+                ? 'tool_use'
+                : 'end',
+        };
+    }
+    /* ────────────────────────────────────────────────────────── */
+    /*  Convert JSON of non-stream (= generateContent)        */
+    /* ────────────────────────────────────────────────────────── */
+    parseOneShot(data) {
+        const textBlocks = [];
+        const toolBlocks = [];
+        for (const cand of data.candidates ?? []) {
+            for (const part of cand.content?.parts ?? []) {
+                if (part.text) {
+                    textBlocks.push({ type: 'text', text: part.text });
+                }
+                if (part.functionCall) {
+                    toolBlocks.push({
+                        type: 'tool_use',
+                        id: this.genUUID(),
+                        name: part.functionCall.name,
+                        input: part.functionCall.args ?? {},
+                    });
+                }
+                if (part.functionResponse) {
+                    toolBlocks.push({
+                        type: 'tool_result',
+                        tool_use_id: part.functionResponse.name,
+                        content: JSON.stringify(part.functionResponse.response),
+                    });
+                }
+            }
+        }
+        const blocks = [...textBlocks, ...toolBlocks];
+        return {
+            blocks,
+            stop_reason: toolBlocks.some((b) => b.type === 'tool_use')
+                ? 'tool_use'
+                : 'end',
+        };
+    }
+    /* ────────────────────────────────────────────────────────── */
+    /*  chatOnce (text)                                       */
+    /* ────────────────────────────────────────────────────────── */
+    async chatOnce(messages, stream = true, onPartialResponse = () => { }, maxTokens) {
+        const res = await this.callGemini(messages, this.model, stream, maxTokens);
+        return stream
+            ? this.parseStream(res, onPartialResponse)
+            : this.parseOneShot(await res.json());
+    }
+    /* ────────────────────────────────────────────────────────── */
+    /*  visionChatOnce (images)                               */
+    /* ────────────────────────────────────────────────────────── */
+    async visionChatOnce(messages, stream = false, onPartialResponse = () => { }, maxTokens) {
+        const res = await this.callGemini(messages, this.visionModel, stream, maxTokens);
+        return stream
+            ? this.parseStream(res, onPartialResponse)
+            : this.parseOneShot(await res.json());
+    }
+    /* ────────────────────────────────────────────────────────── */
+    /*  UUID helper                                           */
+    /* ────────────────────────────────────────────────────────── */
+    genUUID() {
+        return typeof crypto !== 'undefined' && crypto.randomUUID
+            ? crypto.randomUUID()
+            : 'xxxxxxxx-xxxx-4xxx-yxxx-xxxxxxxxxxxx'.replace(/[xy]/g, (c) => {
+                const r = (Math.random() * 16) | 0;
+                const v = c === 'x' ? r : (r & 0x3) | 0x8;
+                return v.toString(16);
+            });
+    }
+}
+exports.GeminiChatService = GeminiChatService;
+//# sourceMappingURL=GeminiChatService.js.map