group_mv.py

import re
import os
from typing import Dict, Tuple, List
import sys
import shutil
from argparse import ArgumentParser


def group_file_names(d: str) -> Dict[str, List[Tuple[str, str]]]:
    """
    Groups file in a given directorie by their group names. It is assumed, that the files match the pattern 
    '{group-name}_{5-digit-sequence-nr}.{ext}'. Files not matching the pattern will be ignored

    :param d: path to directory
    :return: dict containing a key for each distinct group name. Each value will be a list with
             (5-digit-sequence-nr, ext)-pairs.
    """
    seqs = {}
    for file in os.listdir(d):
        match = re.match(r"(.*)_(\d{5})\.(.*)", file)

        if not match:
            print(f"WARNING: ignoring file {file} since it does not comply to the pattern", file=sys.stderr)
            continue

        seq_name = match.group(1)
        seq_nr = match.group(2)
        ext = match.group(3)

        if seq_name not in seqs:
            seqs[seq_name] = []

        seqs[seq_name].append((seq_nr, ext))

    # sort each group by the sequence number
    seqs = {k: sorted(v, key=lambda x: int(x[0])) for k, v in seqs.items()}
    return seqs


def group_mv(src: str, target: str, cut_seq_name=True) -> None:
    """
    Groups files in `src` folder that comply to the pattern '{group-name}_{5-digit-sequence-nr}.{ext}' by their group
    name. Then creates a sub folder in `target` folder for each distinct group name and moves the files from the
    `src` folder to its respective group folder.

    :param src: src folder
    :param target: target folder
    :param cut_seq_name: if set, the group name will be cut from the filenames in the target folders
    """
    if not os.path.exists(src) or not os.path.isdir(src):
        print(f"{src} not a directory", file=sys.stderr)
        sys.exit(1)

    seqs = group_file_names(src)

    if not os.path.exists(target):
        os.mkdir(target)

    if not os.path.isdir(target):
        print(f"{target} is not a directory", file=sys.stderr)
        sys.exit(1)

    for seq_name, files in seqs.items():
        seq_folder = os.path.join(target, seq_name)
        os.mkdir(seq_folder)

        for seq_nr, ext in files:
            source_file_name = f"{seq_name}_{seq_nr}.{ext}"
            if cut_seq_name:
                target_file_name = f"{seq_nr}.{ext}"
            else:
                target_file_name = f"{seq_name}_{seq_nr}.{ext}"

            shutil.move(os.path.join(src, source_file_name),
                        os.path.join(seq_folder, target_file_name))


def main():
    parser = ArgumentParser(description="groups files in the `src` folder by their group name and moves all files of "
                                        "one group to a new sub folder in `target`")
    parser.add_argument("src", help="source folder containing files of pattern "
                                    "'{group-name}_{5-digit-sequence-nr}.{ext}'")
    parser.add_argument("target", help="target folder path")
    parser.add_argument("-c", "--cut-seq-name", action="store_true",
                        help="only keep the sequence number instead of the full image name")

    args = parser.parse_args()
    group_mv(args.src, args.target, args.cut_seq_name)


if __name__ == "__main__":
    main()