/**
* Test script for improved rate limit handling
* Demonstrates the new retry logic, dynamic concurrency adjustment, and smart fallback behavior
*/
const path = require('path');
// Mock OpenAI client for testing rate limit scenarios
class MockOpenAIClient {
constructor(rateLimitMode = false) {
this.rateLimitMode = rateLimitMode;
this.callCount = 0;
this.rateLimitCount = 0;
}
async embeddings() {
this.callCount++;
if (this.rateLimitMode && this.callCount % 3 === 0) { // Every 3rd call hits rate limit
this.rateLimitCount++;
const error = new Error('429 Rate limit reached for text-embedding-3-small in organization org-test on tokens per min (TPM): Limit 1000000, Used 1000000, Requested 3538. Please try again in 212ms. Visit https://platform.openai.com/account/rate-limits to learn more.');
error.status = 429;
throw error;
}
// Simulate successful response
return {
data: [
{ embedding: new Array(1536).fill(0).map(() => Math.random()) },
{ embedding: new Array(1536).fill(0).map(() => Math.random()) }
]
};
}
}
async function testRateLimitHandling() {
console.log('🧪 Testing Improved Rate Limit Handling\n');
// Test scenarios
const scenarios = [
{
name: 'Normal Operation',
rateLimitMode: false,
description: 'No rate limits - should work smoothly'
},
{
name: 'Rate Limit Recovery',
rateLimitMode: true,
description: 'Rate limits with retry logic and recovery'
}
];
for (const scenario of scenarios) {
console.log(`📋 Scenario: ${scenario.name}`);
console.log(` ${scenario.description}`);
console.log('');
// Create mock client
const mockClient = new MockOpenAIClient(scenario.rateLimitMode);
// Simulate the improved behavior
const testTexts = ['Test text 1', 'Test text 2'];
const startTime = Date.now();
try {
let retryCount = 0;
const maxRetries = 3;
const baseDelay = 100;
while (retryCount <= maxRetries) {
try {
const response = await mockClient.embeddings();
const duration = Date.now() - startTime;
console.log(`✅ Success after ${retryCount} retries in ${duration}ms`);
console.log(` - API calls made: ${mockClient.callCount}`);
if (scenario.rateLimitMode) {
console.log(` - Rate limit hits: ${mockClient.rateLimitCount}`);
}
break;
} catch (error) {
if (error.status === 429 && retryCount < maxRetries) {
const retryAfter = baseDelay * Math.pow(2, retryCount);
console.log(`⏳ Rate limit hit (attempt ${retryCount + 1}/${maxRetries}), retrying in ${retryAfter}ms...`);
// Simulate delay
await new Promise(resolve => setTimeout(resolve, retryAfter));
retryCount++;
} else {
throw error;
}
}
}
} catch (error) {
console.log(`❌ Failed after ${maxRetries} retries: ${error.message}`);
}
console.log('');
}
// Demonstrate configuration options
console.log('🔧 Configuration Options:');
console.log('');
console.log('Environment Variables:');
console.log(' EMBEDDING_PARALLEL_MODE=true # Enable parallel processing');
console.log(' EMBEDDING_MAX_CONCURRENCY=5 # Start with conservative concurrency');
console.log(' EMBEDDING_RATE_LIMIT_RETRIES=5 # Max retries for rate limits');
console.log(' EMBEDDING_RATE_LIMIT_BASE_DELAY=1000 # Base delay between retries');
console.log('');
console.log('Behavior:');
console.log(' ✅ Retries rate limits with exponential backoff');
console.log(' ✅ Reduces concurrency when hitting multiple rate limits');
console.log(' ✅ Only falls back to local embeddings for permanent failures');
console.log(' ✅ Recovers automatically after rate limit windows');
console.log('');
console.log('Example with your OpenAI Tier 2 limits:');
console.log(' EMBEDDING_PARALLEL_MODE=true');
console.log(' EMBEDDING_MAX_CONCURRENCY=8 # Conservative for 5,000 RPM limit');
console.log(' EMBEDDING_RATE_LIMIT_RETRIES=3 # Fewer retries, faster failure');
console.log('');
}
// Run the test if this script is executed directly
if (require.main === module) {
testRateLimitHandling().catch(console.error);
}
module.exports = { testRateLimitHandling };
