Delete startup.py
Browse files- startup.py +0 -225
startup.py
DELETED
|
@@ -1,225 +0,0 @@
|
|
| 1 |
-
#!/usr/bin/env python3
|
| 2 |
-
"""
|
| 3 |
-
Startup script for Medical RAG HuggingFace Space.
|
| 4 |
-
Validates setup and initializes the application.
|
| 5 |
-
"""
|
| 6 |
-
|
| 7 |
-
import os
|
| 8 |
-
import sys
|
| 9 |
-
import logging
|
| 10 |
-
from pathlib import Path
|
| 11 |
-
|
| 12 |
-
# Configure logging
|
| 13 |
-
logging.basicConfig(
|
| 14 |
-
level=logging.INFO,
|
| 15 |
-
format='%(asctime)s - %(levelname)s - %(message)s'
|
| 16 |
-
)
|
| 17 |
-
logger = logging.getLogger(__name__)
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
def check_python_version():
|
| 21 |
-
"""Verify Python version is 3.9 or higher."""
|
| 22 |
-
version = sys.version_info
|
| 23 |
-
if version.major < 3 or (version.major == 3 and version.minor < 9):
|
| 24 |
-
logger.error(f"Python 3.9+ required, found {version.major}.{version.minor}")
|
| 25 |
-
return False
|
| 26 |
-
logger.info(f"✓ Python {version.major}.{version.minor}.{version.micro}")
|
| 27 |
-
return True
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
def check_dependencies():
|
| 31 |
-
"""Verify all required packages are installed."""
|
| 32 |
-
required_packages = {
|
| 33 |
-
'gradio': 'Gradio (web UI)',
|
| 34 |
-
'torch': 'PyTorch (deep learning)',
|
| 35 |
-
'transformers': 'Transformers (models)',
|
| 36 |
-
'sentence_transformers': 'Sentence Transformers (embeddings)',
|
| 37 |
-
'chromadb': 'ChromaDB (vector database)',
|
| 38 |
-
'numpy': 'NumPy (numerical)',
|
| 39 |
-
'pandas': 'Pandas (data)',
|
| 40 |
-
}
|
| 41 |
-
|
| 42 |
-
logger.info("Checking dependencies...")
|
| 43 |
-
missing = []
|
| 44 |
-
|
| 45 |
-
for package, name in required_packages.items():
|
| 46 |
-
try:
|
| 47 |
-
__import__(package)
|
| 48 |
-
logger.info(f" ✓ {name}")
|
| 49 |
-
except ImportError:
|
| 50 |
-
logger.warning(f" ✗ {name} (not installed)")
|
| 51 |
-
missing.append(package)
|
| 52 |
-
|
| 53 |
-
if missing:
|
| 54 |
-
logger.error(f"\nMissing packages: {', '.join(missing)}")
|
| 55 |
-
logger.error("Install with: pip install -r requirements.txt")
|
| 56 |
-
return False
|
| 57 |
-
|
| 58 |
-
return True
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
def check_vector_database():
|
| 62 |
-
"""Verify vector database exists."""
|
| 63 |
-
from config import VECTOR_DB_PATH
|
| 64 |
-
|
| 65 |
-
logger.info(f"Checking vector database at: {VECTOR_DB_PATH}")
|
| 66 |
-
|
| 67 |
-
if not os.path.exists(VECTOR_DB_PATH):
|
| 68 |
-
logger.warning(f" ⚠ Vector database not found: {VECTOR_DB_PATH}")
|
| 69 |
-
logger.info(" To setup:")
|
| 70 |
-
logger.info(" 1. Download from HuggingFace Hub, OR")
|
| 71 |
-
logger.info(" 2. Build using notebook's chunking code")
|
| 72 |
-
return False
|
| 73 |
-
|
| 74 |
-
# Check for actual DB files
|
| 75 |
-
db_files = list(Path(VECTOR_DB_PATH).glob("*.db*"))
|
| 76 |
-
if db_files:
|
| 77 |
-
logger.info(f" ✓ Found {len(db_files)} database file(s)")
|
| 78 |
-
return True
|
| 79 |
-
else:
|
| 80 |
-
logger.warning(" ⚠ No .db files found in directory")
|
| 81 |
-
return False
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
def check_model_accessibility():
|
| 85 |
-
"""Check if fine-tuned model is accessible."""
|
| 86 |
-
from config import FINETUNED_MODEL_ID
|
| 87 |
-
|
| 88 |
-
logger.info(f"Model ID configured: {FINETUNED_MODEL_ID}")
|
| 89 |
-
|
| 90 |
-
if not FINETUNED_MODEL_ID:
|
| 91 |
-
logger.error(" ✗ FINETUNED_MODEL_ID not set in config.py")
|
| 92 |
-
return False
|
| 93 |
-
|
| 94 |
-
logger.info(" ℹ Model will be downloaded on first use (requires internet)")
|
| 95 |
-
return True
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
def check_disk_space():
|
| 99 |
-
"""Check available disk space."""
|
| 100 |
-
import shutil
|
| 101 |
-
|
| 102 |
-
logger.info("Checking disk space...")
|
| 103 |
-
|
| 104 |
-
# Check root directory
|
| 105 |
-
stat = shutil.disk_usage("/")
|
| 106 |
-
free_gb = stat.free / (1024**3)
|
| 107 |
-
|
| 108 |
-
if free_gb < 20:
|
| 109 |
-
logger.warning(f" ⚠ Only {free_gb:.1f}GB free (20GB+ recommended)")
|
| 110 |
-
logger.warning(" Models may not download properly")
|
| 111 |
-
return False
|
| 112 |
-
|
| 113 |
-
logger.info(f" ✓ {free_gb:.1f}GB free")
|
| 114 |
-
return True
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
def check_memory():
|
| 118 |
-
"""Check available RAM."""
|
| 119 |
-
import psutil
|
| 120 |
-
|
| 121 |
-
logger.info("Checking system memory...")
|
| 122 |
-
|
| 123 |
-
# Get memory info
|
| 124 |
-
memory = psutil.virtual_memory()
|
| 125 |
-
available_gb = memory.available / (1024**3)
|
| 126 |
-
total_gb = memory.total / (1024**3)
|
| 127 |
-
|
| 128 |
-
logger.info(f" Total: {total_gb:.1f}GB, Available: {available_gb:.1f}GB")
|
| 129 |
-
|
| 130 |
-
if available_gb < 8:
|
| 131 |
-
logger.warning(f" ⚠ Low available RAM ({available_gb:.1f}GB)")
|
| 132 |
-
logger.warning(" May cause performance issues")
|
| 133 |
-
elif available_gb < 4:
|
| 134 |
-
logger.error(f" ✗ Insufficient RAM ({available_gb:.1f}GB < 4GB required)")
|
| 135 |
-
return False
|
| 136 |
-
else:
|
| 137 |
-
logger.info(f" ✓ Sufficient RAM")
|
| 138 |
-
return True
|
| 139 |
-
|
| 140 |
-
return True
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
def check_gpu():
|
| 144 |
-
"""Check GPU availability."""
|
| 145 |
-
try:
|
| 146 |
-
import torch
|
| 147 |
-
if torch.cuda.is_available():
|
| 148 |
-
logger.info(f"✓ GPU available: {torch.cuda.get_device_name(0)}")
|
| 149 |
-
logger.info(f" VRAM: {torch.cuda.get_device_properties(0).total_memory / 1024**3:.1f}GB")
|
| 150 |
-
return True
|
| 151 |
-
else:
|
| 152 |
-
logger.info("⚠ GPU not available (running on CPU)")
|
| 153 |
-
return False
|
| 154 |
-
except Exception as e:
|
| 155 |
-
logger.warning(f"⚠ Could not check GPU: {e}")
|
| 156 |
-
return False
|
| 157 |
-
|
| 158 |
-
|
| 159 |
-
def validate_configuration():
|
| 160 |
-
"""Validate configuration file."""
|
| 161 |
-
from config import validate_config
|
| 162 |
-
|
| 163 |
-
logger.info("Validating configuration...")
|
| 164 |
-
return validate_config()
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
def main():
|
| 168 |
-
"""Run all startup checks."""
|
| 169 |
-
logger.info("=" * 60)
|
| 170 |
-
logger.info("MEDICAL RAG SYSTEM - STARTUP VALIDATION")
|
| 171 |
-
logger.info("=" * 60)
|
| 172 |
-
|
| 173 |
-
checks = [
|
| 174 |
-
("Python Version", check_python_version),
|
| 175 |
-
("Dependencies", check_dependencies),
|
| 176 |
-
("Vector Database", check_vector_database),
|
| 177 |
-
("Model Configuration", check_model_accessibility),
|
| 178 |
-
("Disk Space", check_disk_space),
|
| 179 |
-
("RAM", check_memory),
|
| 180 |
-
("GPU", check_gpu),
|
| 181 |
-
("Configuration", validate_configuration),
|
| 182 |
-
]
|
| 183 |
-
|
| 184 |
-
results = {}
|
| 185 |
-
critical_failed = False
|
| 186 |
-
|
| 187 |
-
for name, check_func in checks:
|
| 188 |
-
logger.info(f"\n[{name}]")
|
| 189 |
-
try:
|
| 190 |
-
result = check_func()
|
| 191 |
-
results[name] = result
|
| 192 |
-
|
| 193 |
-
# Determine if critical
|
| 194 |
-
critical = name in ["Python Version", "Dependencies", "Configuration"]
|
| 195 |
-
if not result and critical:
|
| 196 |
-
critical_failed = True
|
| 197 |
-
except Exception as e:
|
| 198 |
-
logger.error(f"Error during check: {e}")
|
| 199 |
-
results[name] = False
|
| 200 |
-
critical_failed = True
|
| 201 |
-
|
| 202 |
-
# Summary
|
| 203 |
-
logger.info("\n" + "=" * 60)
|
| 204 |
-
logger.info("SUMMARY")
|
| 205 |
-
logger.info("=" * 60)
|
| 206 |
-
|
| 207 |
-
for name, result in results.items():
|
| 208 |
-
status = "✓ PASS" if result else "⚠ WARN"
|
| 209 |
-
logger.info(f"{status:8} - {name}")
|
| 210 |
-
|
| 211 |
-
logger.info("=" * 60)
|
| 212 |
-
|
| 213 |
-
if critical_failed:
|
| 214 |
-
logger.error("\n✗ CRITICAL ISSUES FOUND - CANNOT START")
|
| 215 |
-
logger.error("Please fix the issues above before running the application.")
|
| 216 |
-
return False
|
| 217 |
-
else:
|
| 218 |
-
logger.info("\n✓ All critical checks passed!")
|
| 219 |
-
logger.info("You can now run: python app.py")
|
| 220 |
-
return True
|
| 221 |
-
|
| 222 |
-
|
| 223 |
-
if __name__ == "__main__":
|
| 224 |
-
success = main()
|
| 225 |
-
sys.exit(0 if success else 1)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|