debug_api.py

"""
Gemini API Key Deep Diagnostic & RCA Tool
Run: python debug_api.py
"""

import os
import sys
import json
import time
from pathlib import Path
from dotenv import load_dotenv

load_dotenv(dotenv_path=Path(__file__).parent / ".env")

api_key = os.getenv("GEMINI_API_KEY")

print("=" * 60)
print("  GEMINI API KEY DIAGNOSTIC TOOL")
print("=" * 60)

# --- STEP 1: Key presence ---
print("\n[STEP 1] Checking .env file...")
if not api_key:
    print("  FAIL: GEMINI_API_KEY not found in .env")
    sys.exit(1)
print(f"  PASS: Key found: {api_key[:12]}...{api_key[-6:]}")
print(f"  Key length: {len(api_key)} chars")
print(f"  Key prefix: {api_key[:5]} (should start with 'AQ.Ab')")

# --- STEP 2: Import library ---
print("\n[STEP 2] Importing google.generativeai...")
try:
    import google.generativeai as genai
    import warnings
    warnings.filterwarnings("ignore")  # suppress FutureWarning noise
    print("  PASS: Library imported successfully")
except ImportError as e:
    print(f"  FAIL: Cannot import google.generativeai: {e}")
    sys.exit(1)

# --- STEP 3: Configure & list models ---
print("\n[STEP 3] Configuring API key and listing available models...")
try:
    genai.configure(api_key=api_key)
    models = [m.name for m in genai.list_models() if "generateContent" in m.supported_generation_methods]
    print(f"  PASS: {len(models)} models available")
    for m in models[:5]:
        print(f"    - {m}")
    if len(models) > 5:
        print(f"    ... and {len(models) - 5} more")
except Exception as e:
    print(f"  FAIL: Cannot list models: {e}")

# --- STEP 4: Test each model with full error breakdown ---
test_models = [
    "gemini-3.5-flash",
    "gemini-flash-latest",
    "gemini-2.0-flash",
    "gemini-2.0-flash-lite",
]

print("\n[STEP 4] Testing content generation across models...")
print("-" * 60)

results = {}

for model_name in test_models:
    print(f"\n  Testing: {model_name}")
    try:
        model = genai.GenerativeModel(model_name)
        start = time.time()
        response = model.generate_content("Say the word: OK")
        elapsed = round(time.time() - start, 2)
        results[model_name] = "WORKING"
        print(f"  RESULT: [SUCCESS] Response='{response.text.strip()}' | Time={elapsed}s")
    except Exception as e:
        error_str = str(e)
        results[model_name] = "FAILED"

        # Parse quota violation details
        if "429" in error_str or "ResourceExhausted" in error_str:
            print(f"  RESULT: [FAIL] 429 Quota Exceeded")

            # Extract exact quota details
            if "GenerateRequestsPerDayPerProjectPerModel-FreeTier" in error_str:
                print(f"         RCA: Daily request limit hit (free tier)")
                # Try to find the quota_value
                if "quota_value: 20" in error_str:
                    print(f"         Limit: 20 requests/day used up")
                elif "limit: 0" in error_str:
                    print(f"         Limit: 0 — This model may be DISABLED for this key's project")

            if "GenerateRequestsPerMinutePerProjectPerModel-FreeTier" in error_str:
                print(f"         RCA: Per-minute rate limit hit")

            if "GenerateContentInputTokensPerModelPerMinute-FreeTier" in error_str:
                print(f"         RCA: Input token per-minute limit hit")

            # Retry delay
            import re
            retry_match = re.search(r"retry_delay \{\s*seconds: (\d+)", error_str)
            if retry_match:
                print(f"         Retry after: {retry_match.group(1)} seconds")

        elif "404" in error_str or "not found" in error_str.lower():
            print(f"  RESULT: [FAIL] Model not found / not available for this key")
        elif "403" in error_str or "API_KEY_INVALID" in error_str:
            print(f"  RESULT: [FAIL] Invalid or revoked API key")
        elif "PermissionDenied" in error_str:
            print(f"  RESULT: [FAIL] Permission denied — billing may be required")
        else:
            print(f"  RESULT: [FAIL] Unknown error:")
            print(f"         {error_str[:300]}")

# --- STEP 5: RCA Summary ---
print("\n" + "=" * 60)
print("  ROOT CAUSE ANALYSIS (RCA) SUMMARY")
print("=" * 60)

working = [m for m, r in results.items() if r == "WORKING"]
failed  = [m for m, r in results.items() if r == "FAILED"]

print(f"\n  Working models : {working if working else 'NONE'}")
print(f"  Failed models  : {failed}")

if not working:
    print("\n  DIAGNOSIS: All models failed.")
    print("  Most likely cause: The Google Cloud project linked to")
    print("  this API key has exhausted its FREE TIER daily quota.")
    print("")
    print("  Even NEW keys from the same Google account share the")
    print("  same project quota. Each Google account's AI Studio")
    print("  project has a hard cap of 20 req/day for gemini-3.5-flash.")
    print("")
    print("  ACTION REQUIRED:")
    print("  1. Sign in to a DIFFERENT Google account at:")
    print("     https://aistudio.google.com/app/apikey")
    print("  2. Create a new API key there")
    print("  3. Paste into .env and re-run: python test_api_key.py")
else:
    print(f"\n  DIAGNOSIS: API key is working on {working[0]}.")
    print("  Update gemini_client.py to use:", working[0])

print("\n" + "=" * 60)