src/__init__.py

"""
Predictive Maintenance Master Source Package (src)
Organized into modular subpackages:
- src.preprocessing: Preprocessors for AI4I, Pump, CMAPSS, Azure PdM datasets
- src.training: Classification & RUL Regression model trainers
- src.rag: FAISS vector indexing and SmolLM2 explainer
- src.utils: Shared helpers, EDA visualization, and Production Scheduler
"""

import os
import sys

# Ensure all subdirectories of src are in sys.path for direct imports
src_dir = os.path.dirname(os.path.abspath(__file__))
for subfolder in ['preprocessing', 'training', 'rag', 'utils']:
    sub_path = os.path.join(src_dir, subfolder)
    if os.path.isdir(sub_path) and sub_path not in sys.path:
        sys.path.insert(0, sub_path)

from preprocessing import (
    preprocess_ai4i, preprocess_pump,
    preprocess_cmapss, preprocess_azure
)
from training import (
    train_pipeline, run_training_pipeline,
    LSTMRegressor, CNNLSTMRegressor, CMAPSSDataset
)
from rag import (
    build_and_save_index, LogRetriever,
    generate_natural_language_logs, LLMExplainer, clean_report
)
from utils import (
    set_seeds, time_tracker, logger,
    plot_missing_values, plot_class_distributions,
    plot_correlation_heatmap, plot_statistical_feature_importance,
    plot_failure_distribution_timeline, plot_sensor_behavior_overlay,
    ProductionScheduler, generate_mock_schedule
)

__all__ = [
    'preprocess_ai4i', 'preprocess_pump', 'preprocess_cmapss', 'preprocess_azure',
    'train_pipeline', 'run_training_pipeline', 'LSTMRegressor', 'CNNLSTMRegressor', 'CMAPSSDataset',
    'build_and_save_index', 'LogRetriever', 'generate_natural_language_logs', 'LLMExplainer', 'clean_report',
    'set_seeds', 'time_tracker', 'logger',
    'plot_missing_values', 'plot_class_distributions',
    'plot_correlation_heatmap', 'plot_statistical_feature_importance',
    'plot_failure_distribution_timeline', 'plot_sensor_behavior_overlay',
    'ProductionScheduler', 'generate_mock_schedule'
]