Snaseem2026/code-comment-classifier

1"""2Validation utilities for model and data validation3"""4import os5import yaml6from typing import Dict, List, Optional7import logging8from pathlib import Path9 10 11def validate_config(config: Dict) -> List[str]:12    """13    Validate configuration file for common issues.14    15    Args:16        config: Configuration dictionary17        18    Returns:19        List of validation error messages (empty if valid)20    """21    errors = []22    23    # Check required sections24    required_sections = ['model', 'training', 'data', 'labels']25    for section in required_sections:26        if section not in config:27            errors.append(f"Missing required section: {section}")28    29    if errors:30        return errors31    32    # Validate model section33    if 'name' not in config['model']:34        errors.append("model.name is required")35    if 'num_labels' not in config['model']:36        errors.append("model.num_labels is required")37    elif config['model']['num_labels'] != len(config.get('labels', [])):38        errors.append(f"model.num_labels ({config['model']['num_labels']}) doesn't match number of labels ({len(config['labels'])})")39    40    # Validate training section41    training = config['training']42    if 'num_train_epochs' in training and training['num_train_epochs'] <= 0:43        errors.append("training.num_train_epochs must be positive")44    if 'learning_rate' in training and training['learning_rate'] <= 0:45        errors.append("training.learning_rate must be positive")46    if 'per_device_train_batch_size' in training and training['per_device_train_batch_size'] <= 0:47        errors.append("training.per_device_train_batch_size must be positive")48    49    # Validate data section50    data = config['data']51    if 'data_path' in data and not os.path.exists(data['data_path']):52        errors.append(f"Data file not found: {data['data_path']}")53    54    train_size = data.get('train_size', 0)55    val_size = data.get('val_size', 0)56    test_size = data.get('test_size', 0)57    total = train_size + val_size + test_size58    if abs(total - 1.0) > 1e-6:59        errors.append(f"Data split sizes must sum to 1.0, got {total}")60    61    # Validate labels62    if 'labels' not in config or not config['labels']:63        errors.append("labels section is required and cannot be empty")64    elif len(set(config['labels'])) != len(config['labels']):65        errors.append("labels must be unique")66    67    return errors68 69 70def validate_model_path(model_path: str) -> bool:71    """72    Validate that model path exists and contains required files.73    74    Args:75        model_path: Path to model directory76        77    Returns:78        True if valid, False otherwise79    """80    if not os.path.exists(model_path):81        logging.error(f"Model path does not exist: {model_path}")82        return False83    84    required_files = ['config.json']85    for file in required_files:86        file_path = os.path.join(model_path, file)87        if not os.path.exists(file_path):88            logging.error(f"Required file missing: {file_path}")89            return False90    91    return True92 93 94def validate_data_file(data_path: str, required_columns: List[str] = None) -> List[str]:95    """96    Validate data file format and content.97    98    Args:99        data_path: Path to data file100        required_columns: List of required column names101        102    Returns:103        List of validation error messages (empty if valid)104    """105    errors = []106    107    if required_columns is None:108        required_columns = ['comment', 'label']109    110    if not os.path.exists(data_path):111        errors.append(f"Data file not found: {data_path}")112        return errors113    114    try:115        import pandas as pd116        df = pd.read_csv(data_path)117        118        # Check required columns119        missing_columns = [col for col in required_columns if col not in df.columns]120        if missing_columns:121            errors.append(f"Missing required columns: {missing_columns}")122        123        # Check for empty dataframe124        if len(df) == 0:125            errors.append("Data file is empty")126        127        # Check for missing values in required columns128        if 'comment' in df.columns:129            empty_comments = df['comment'].isna().sum() + (df['comment'].str.strip().str.len() == 0).sum()130            if empty_comments > 0:131                errors.append(f"Found {empty_comments} empty comments")132        133        if 'label' in df.columns:134            missing_labels = df['label'].isna().sum()135            if missing_labels > 0:136                errors.append(f"Found {missing_labels} missing labels")137        138    except Exception as e:139        errors.append(f"Error reading data file: {str(e)}")140    141    return errors142 143 144def validate_config_file(config_path: str) -> bool:145    """146    Validate configuration file.147    148    Args:149        config_path: Path to configuration file150        151    Returns:152        True if valid, False otherwise153    """154    if not os.path.exists(config_path):155        logging.error(f"Config file not found: {config_path}")156        return False157    158    try:159        with open(config_path, 'r') as f:160            config = yaml.safe_load(f)161        162        errors = validate_config(config)163        if errors:164            logging.error("Configuration validation errors:")165            for error in errors:166                logging.error(f"  - {error}")167            return False168        169        logging.info("Configuration file is valid")170        return True171        172    except Exception as e:173        logging.error(f"Error reading config file: {str(e)}")174        return False175