Files
SequenceCompressor/check_sequences.py

102 lines
3.6 KiB
Python
Raw Permalink Normal View History

2025-12-11 12:12:10 -07:00
#!/usr/bin/env python3
"""
Diagnostic script to check which sequences are in input vs output.
"""
from pathlib import Path
from collections import defaultdict
def get_sequences(directory):
"""Get all sequence folders (directories containing PNG files)."""
sequences = defaultdict(list)
root = Path(directory)
if not root.exists():
return sequences
for png_file in root.rglob('*.png'):
# Get the sequence folder (parent directory)
seq_folder = png_file.parent
relative_seq = seq_folder.relative_to(root)
sequences[str(relative_seq)].append(png_file)
return sequences
def main():
input_dir = Path('input')
output_dir = Path('output')
print("=" * 80)
print("SEQUENCE DIAGNOSTIC REPORT")
print("=" * 80)
input_sequences = get_sequences(input_dir)
output_sequences = get_sequences(output_dir)
print(f"\nInput sequences: {len(input_sequences)}")
print(f"Output sequences: {len(output_sequences)}")
# Find missing sequences
missing = set(input_sequences.keys()) - set(output_sequences.keys())
if missing:
print(f"\n{'=' * 80}")
print(f"MISSING SEQUENCES ({len(missing)}):")
print(f"{'=' * 80}")
for seq in sorted(missing):
png_count = len(input_sequences[seq])
print(f" - {seq} ({png_count} PNG files)")
# Check if folder exists in output
output_seq_path = output_dir / seq
if output_seq_path.exists():
files_in_output = list(output_seq_path.iterdir())
print(f" Output folder exists but has {len(files_in_output)} files")
if files_in_output:
print(f" Files: {[f.name for f in files_in_output[:5]]}")
else:
print(f" Output folder does not exist")
else:
print("\nAll input sequences have corresponding output sequences.")
# Find sequences with different file counts
print(f"\n{'=' * 80}")
print("SEQUENCE FILE COUNT COMPARISON:")
print(f"{'=' * 80}")
for seq in sorted(input_sequences.keys()):
input_count = len(input_sequences[seq])
output_count = len(output_sequences.get(seq, []))
status = "OK" if input_count == output_count else "DIFF"
print(f"{status:4s} {seq}: {input_count} input -> {output_count} output")
if input_count != output_count and seq in output_sequences:
print(f" Difference: {input_count - output_count} files missing")
# Check for folders without PNG files
print(f"\n{'=' * 80}")
print("ALL FOLDERS IN INPUT DIRECTORY:")
print(f"{'=' * 80}")
input_root = Path('input')
if input_root.exists():
# Get all top-level directories
top_level_dirs = [d for d in input_root.iterdir() if d.is_dir()]
print(f"Total top-level folders: {len(top_level_dirs)}")
print()
folders_with_pngs = set()
for seq_name, png_files in input_sequences.items():
if png_files:
# Get the folder path from the first PNG file
folder_path = png_files[0].parent.relative_to(input_root)
folders_with_pngs.add(str(folder_path))
for folder in sorted(top_level_dirs):
rel_folder = folder.relative_to(input_root)
has_pngs = str(rel_folder) in folders_with_pngs
png_count = len(list(folder.rglob('*.png')))
status = "HAS PNGs" if has_pngs else "NO PNGs"
print(f"{status:10s} {rel_folder} ({png_count} PNG files)")
if __name__ == '__main__':
main()