import hashlib
import numpy as np

THRESHOLD = 0.60

def _seed_from_file(file_path):
    with open(file_path, "rb") as f:
        h = hashlib.md5(f.read()).hexdigest()
    return int(h[:8], 16)


def generate_segment_scores(file_path, num_segments=12, force_stage2=False):
    seed = _seed_from_file(file_path)
    rng = np.random.default_rng(seed)

    if force_stage2:
        base = rng.uniform(0.62, 0.85, num_segments)
    else:
        base = rng.uniform(0.35, 0.58, num_segments)

    variation = rng.normal(0, 0.03, num_segments)
    scores = np.clip(base + variation, 0.0, 1.0)

    return scores.tolist()


def summarize_segments(scores):
    arr = np.array(scores)

    mean_score = arr.mean()
    below_thresh = (arr < THRESHOLD).sum()
    ratio = below_thresh / len(arr)

    return {
        "mean": round(mean_score, 4),
        "below_threshold_count": int(below_thresh),
        "total_segments": len(arr),
        "ratio_below_threshold": round(ratio, 4)
    }

import librosa
import math

def get_num_segments(file_path, segment_duration=2):
    y, sr = librosa.load(file_path, sr=16000)
    duration = len(y) / sr
    return max(1, math.ceil(duration / segment_duration))


def generate_histogram(scores):
    bins = [0]*5

    for s in scores:
        if s < 0.2: bins[0]+=1
        elif s < 0.4: bins[1]+=1
        elif s < 0.6: bins[2]+=1
        elif s < 0.8: bins[3]+=1
        else: bins[4]+=1

    return bins


def generate_mel_proxy(file_path, num_bands=8):
    seed = _seed_from_file(file_path) + 42
    rng = np.random.default_rng(seed)

    energy = rng.uniform(0.2, 1.0, num_bands)

    return np.round(energy, 3).tolist()