from scipy.stats import pearsonr
import numpy as np
import warnings

def analyze_fc_patterns(latents, demographics):
    results = {}
    
    # Suppress the ConstantInputWarning
    with warnings.catch_warnings():
        warnings.filterwarnings('ignore', category=RuntimeWarning)
        
        for demo_name, demo_values in demographics.items():
            # Check if the demographic is categorical or continuous
            if demo_name not in ['sex', 'gender']:  # For continuous variables
                correlations = []
                p_values = []
                
                for latent_dim in range(latents.shape[1]):
                    # Check for constant values that would cause correlation issues
                    if np.all(latents[:, latent_dim] == latents[0, latent_dim]) or np.all(demo_values == demo_values[0]):
                        # If either array is constant, correlation is undefined
                        r, p = np.nan, np.nan
                    else:
                        try:
                            # Convert to numpy array if not already
                            demo_array = np.array(demo_values, dtype=float)
                            # Calculate correlation
                            r, p = pearsonr(latents[:, latent_dim], demo_array)
                        except (ValueError, TypeError) as e:
                            print(f"Error calculating correlation for {demo_name}, dimension {latent_dim}: {e}")
                            r, p = np.nan, np.nan
                    
                    correlations.append(r)
                    p_values.append(p)
                
                results[demo_name] = {'correlations': correlations, 'p_values': p_values}
    
    return results