ai-codegen-cli-vrk 2.0.9 → 2.1.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-codegen-cli-vrk",
-  "version": "2.0.9",
+  "version": "2.1.1",
   "description": "Minimalist Terminal-based AI code generator",
   "type": "module",
   "bin": {
@@ -14,5 +14,8 @@
     "@google/generative-ai": "^0.21.0",
     "fs-extra": "^11.2.0",
     "readline-sync": "^1.4.10"
+  },
+  "engines": {
+    "node": ">=18.0.0"
   }
 }

package/src/aiClient.js

@@ -1,52 +1,47 @@
 import { GoogleGenerativeAI } from "@google/generative-ai";
 
 let genAI = null;
-let activeModel = null;
+let modelQueue = [];
+let currentModelIndex = 0;
 
 export function setApiKey(apiKey) {
   genAI = new GoogleGenerativeAI(apiKey.trim());
 }
 
 /**
- * Robust model discovery. Tries newer Flash models first.
+ * Lists all usable models and sorts them by stability.
  */
-async function getWorkingModel() {
-  if (activeModel) return activeModel;
+async function refreshModelQueue() {
+  if (modelQueue.length > 0) return;
 
-  // Added 2.0 Flash to the list
-  const modelNames = ["gemini-1.5-flash", "gemini-2.0-flash-exp", "gemini-1.5-flash-latest", "gemini-pro"];
-  const apiVersions = ["v1beta", "v1"];
-  
-  let lastError = null;
+  try {
+    const response = await fetch(`https://generativelanguage.googleapis.com/v1beta/models?key=${genAI.apiKey}`);
+    const data = await response.json();
+    
+    if (data.models) {
+      // Prioritize 1.5-flash (most stable) then 2.0-flash (experimental)
+      const usable = data.models
+        .filter(m => m.supportedGenerationMethods.includes("generateContent"))
+        .map(m => m.name.split('/').pop());
 
-  for (const ver of apiVersions) {
-    for (const name of modelNames) {
-      try {
-        const m = genAI.getGenerativeModel({ model: name }, { apiVersion: ver });
-        // Verification call
-        await m.generateContent({ 
-          contents: [{ role: "user", parts: [{ text: "hi" }] }], 
-          generationConfig: { maxOutputTokens: 1 } 
-        });
-        activeModel = m;
-        return m;
-      } catch (err) {
-        lastError = err;
-        if (err.message.includes("404") || err.message.includes("not found")) continue;
-        if (err.message.includes("API key not valid") || err.message.includes("401")) {
-           throw new Error("Invalid API Key. Please check it at aistudio.google.com");
-        }
-      }
+      const preferred = ["gemini-1.5-flash", "gemini-1.5-flash-latest", "gemini-2.0-flash-exp", "gemini-pro"];
+      
+      modelQueue = preferred.filter(p => usable.includes(p));
+      // Add any other usable models found as backup
+      usable.forEach(u => { if(!modelQueue.includes(u)) modelQueue.push(u); });
     }
+  } catch (e) {
+    modelQueue = ["gemini-1.5-flash", "gemini-pro"];
   }
-  
-  throw new Error(`Connection Failed. Google's server rejected the model request. Error: ${lastError?.message}`);
 }
 
 export async function generateFullProject(task, tests, retryCount = 0) {
+  await refreshModelQueue();
+  
+  const modelName = modelQueue[currentModelIndex] || "gemini-1.5-flash";
+  const model = genAI.getGenerativeModel({ model: modelName }, { apiVersion: "v1beta" });
+
   try {
-    const model = await getWorkingModel();
-    
     const prompt = `
 Generate the ENTIRE project in a SINGLE response. 
 Strictly pass all test cases.
@@ -75,11 +70,22 @@ ${tests}
     return text.replace(/```[a-z]*\n([\s\S]*?)\n```/gi, "$1").trim();
 
   } catch (error) {
-    // Handle Overloaded (503) or Rate Limit (429) - Critical for Free Tier
-    if ((error.message.includes("503") || error.message.includes("429")) && retryCount < 5) {
-      console.log(`..... (Quota hit or Server busy, waiting to retry ${retryCount + 1}/5)`);
-      // Wait 15 seconds to safely clear the Free Tier 15-RPM limit
-      await new Promise(r => setTimeout(r, 15000));
+    const msg = error.message || "";
+    
+    // Handle Quota Exceeded (429) or Server Busy (503)
+    if ((msg.includes("429") || msg.includes("quota") || msg.includes("503") || msg.includes("overloaded")) && retryCount < 5) {
+      
+      // If a specific model is exhausted, try to switch to the next one in the queue
+      if (currentModelIndex < modelQueue.length - 1) {
+        console.log(`..... (Model ${modelName} exhausted, switching to ${modelQueue[currentModelIndex+1]}...)`);
+        currentModelIndex++;
+        return generateFullProject(task, tests, retryCount + 1);
+      }
+
+      // If all models are exhausted, wait a full 60 seconds
+      console.log(`..... (All models busy/exhausted, waiting 60s to reset quota...)`);
+      await new Promise(r => setTimeout(r, 60000));
+      currentModelIndex = 0; // Reset queue and try again
       return generateFullProject(task, tests, retryCount + 1);
     }
     throw error;