import os
from collections import Counter

def count_instances(label_dir):
    counts = Counter()
    for fname in os.listdir(label_dir):
        if fname.endswith(".txt"):
            with open(os.path.join(label_dir, fname)) as f:
                for line in f:
                    if line.strip():
                        cls_id = line.strip().split()[0]
                        counts[cls_id] += 1
    return counts

# ------------------------- CONFIG (edit these) -------------------------

TRAIN_LABELS = r"C:\Users\shrey\Downloads\dataset\labels\train"
VAL_LABELS = r"C:\Users\shrey\Downloads\dataset\labels\val"
TEST_LABELS = r"C:\Users\shrey\Downloads\dataset\labels\test"

# Map class IDs to names for readable output -- confirm against your data.yaml
CLASS_NAMES = {
    "0": "plastic_bottle",
    "1": "drink_can",
    "2": "paper",
}

# -------------------------------------------------------------------------

def print_counts(label, counts):
    print(f"{label}:")
    total = sum(counts.values())
    for cls_id in sorted(counts.keys()):
        name = CLASS_NAMES.get(cls_id, f"class_{cls_id}")
        pct = (counts[cls_id] / total * 100) if total else 0
        print(f"  {name:15s} (id {cls_id}): {counts[cls_id]:5d}  ({pct:.1f}%)")
    print(f"  {'TOTAL':15s}      : {total:5d}\n")


if __name__ == "__main__":
    for label_name, label_dir in [
        ("Train", TRAIN_LABELS),
        ("Val", VAL_LABELS),
        ("Test", TEST_LABELS),
    ]:
        if os.path.isdir(label_dir):
            counts = count_instances(label_dir)
            print_counts(label_name, counts)
        else:
            print(f"Skipping {label_name} -- folder not found: {label_dir}\n")