Fabric-Analytics-MCP

import fetch from 'node-fetch'; import fs from 'fs'; import path from 'path'; interface AzureOpenAIConfig { azureOpenAI: { apiKey: string; endpoint: string; apiVersion: string; deploymentName: string; model: string; maxTokens: number; temperature: number; }; analysis: { enableLLMAnalysis: boolean; analysisTypes: string[]; }; } interface OpenAIResponse { choices: Array<{ message: { content: string; }; }>; } export class AzureOpenAIAnalyzer { private config: AzureOpenAIConfig; constructor(configPath?: string) { const configFile = configPath || path.join(process.cwd(), 'azure-openai-config.json'); this.config = JSON.parse(fs.readFileSync(configFile, 'utf8')); } async analyzeSparkLogs(logs: string[], analysisType: string = 'comprehensive'): Promise<string> { if (!this.config.analysis.enableLLMAnalysis) { return "LLM analysis is disabled in configuration"; } const prompt = this.buildSparkAnalysisPrompt(logs, analysisType); return await this.callAzureOpenAI(prompt); } async analyzeStatementPerformance(statementData: any, analysisType: string = 'optimization'): Promise<string> { if (!this.config.analysis.enableLLMAnalysis) { return "LLM analysis is disabled in configuration"; } const prompt = this.buildStatementAnalysisPrompt(statementData, analysisType); return await this.callAzureOpenAI(prompt); } async analyzeExecutionHistory(sessions: any[], timeRange: string, analysisType: string = 'trends'): Promise<string> { if (!this.config.analysis.enableLLMAnalysis) { return "LLM analysis is disabled in configuration"; } const prompt = this.buildHistoryAnalysisPrompt(sessions, timeRange, analysisType); return await this.callAzureOpenAI(prompt); } private buildSparkAnalysisPrompt(logs: string[], analysisType: string): string { const logSample = logs.slice(0, 50).join('\n'); // First 50 lines return `You are a Spark performance expert. Analyze the following Spark session logs and provide insights based on the analysis type: ${analysisType}. Spark Session Logs: ${logSample} Please provide: 1. Performance bottlenecks identified 2. Memory usage patterns 3. Error analysis (if any) 4. Optimization recommendations 5. Resource utilization insights Focus on actionable recommendations for improving Spark job performance.`; } private buildStatementAnalysisPrompt(statementData: any, analysisType: string): string { return `You are a Spark SQL optimization expert. Analyze the following Spark statement execution and provide ${analysisType} insights. Statement Details: - Code: ${statementData.code} - State: ${statementData.state} - Execution Time: ${statementData.executionTime || 'Unknown'} - Output Type: ${statementData.outputType || 'Unknown'} Please provide: 1. Query optimization opportunities 2. Performance analysis 3. Best practices recommendations 4. Potential issues and solutions 5. Resource optimization suggestions Focus on specific, actionable improvements for this Spark statement.`; } private buildHistoryAnalysisPrompt(sessions: any[], timeRange: string, analysisType: string): string { const sessionSummary = sessions.map(s => ({ id: s.id, state: s.state, kind: s.kind, duration: s.duration || 'Unknown' })); return `You are a Spark operations analyst. Analyze the following session execution history over ${timeRange} and provide ${analysisType} insights. Session History Summary: ${JSON.stringify(sessionSummary, null, 2)} Please provide: 1. Performance trends analysis 2. Failure pattern identification 3. Resource utilization trends 4. Operational recommendations 5. Capacity planning insights Focus on trends and patterns that can help optimize overall Spark operations.`; } private async callAzureOpenAI(prompt: string): Promise<string> { const { azureOpenAI } = this.config; const url = `${azureOpenAI.endpoint}openai/deployments/${azureOpenAI.deploymentName}/chat/completions?api-version=${azureOpenAI.apiVersion}`; const requestBody = { messages: [ { role: "system", content: "You are an expert Apache Spark performance analyst and optimization consultant. Provide detailed, actionable insights and recommendations." }, { role: "user", content: prompt } ], max_tokens: azureOpenAI.maxTokens, temperature: azureOpenAI.temperature, top_p: 0.95, frequency_penalty: 0, presence_penalty: 0 }; try { const response = await fetch(url, { method: 'POST', headers: { 'Content-Type': 'application/json', 'api-key': azureOpenAI.apiKey }, body: JSON.stringify(requestBody) }); if (!response.ok) { const errorText = await response.text(); throw new Error(`Azure OpenAI API error: ${response.status} - ${errorText}`); } const data = await response.json() as OpenAIResponse; return data.choices[0]?.message?.content || "No response from Azure OpenAI"; } catch (error) { console.error('Azure OpenAI API call failed:', error); return `Error calling Azure OpenAI: ${error instanceof Error ? error.message : String(error)}`; } } async testConnection(): Promise<{ success: boolean; message: string }> { try { const testPrompt = "Hello, this is a test connection to Azure OpenAI. Please respond with 'Connection successful' and today's date."; const response = await this.callAzureOpenAI(testPrompt); if (response.includes('Error calling Azure OpenAI')) { return { success: false, message: response }; } return { success: true, message: `✅ Azure OpenAI connection successful! Response: ${response}` }; } catch (error) { return { success: false, message: `❌ Connection test failed: ${error instanceof Error ? error.message : String(error)}` }; } } }