extract_tar_replace_symlinks.py

gistfile1.txt

import tarfile

import os
import shutil

def is_symlink(path):
    try:
        os.readlink(path)
        return True
    except:
        return False


def get_real_path(base, linkpath):
    pre_final_path = ''
    if linkpath.startswith('/'):
        linkpath = linkpath[1:]
    
        pre_final_path = base + os.sep + linkpath
    else:
        pre_final_path = linkpath

    return pre_final_path


def real_file_lookup(path):
    print(f'Handle file: {path}')
    while is_symlink(path):
        link_target = os.readlink(path)

        if link_target.startswith('/'):
            link_target = 'rootfs' + link_target

        folder, file_name = os.path.split(link_target)

        if folder == '':
            folder, _ = os.path.split(path)
        
        link_target = folder + '/' + file_name

        print(f'    Redirect to: {link_target}')

        if link_target.startswith('..'):
            link_target = folder + '/' + link_target

        path = link_target

    return path






def extract_tar_replace_symlinks(tar_path, extract_path):
    with tarfile.open(tar_path, 'r') as tar:

        tar.extractall(path=extract_path)
        
        for r, d, f in os.walk(extract_path):
            for file in f:
                path = os.path.join(r, file)

                if is_symlink(path):
                    real_path = real_file_lookup(path)

                    os.remove(path)

                    try:
                        shutil.copy(real_path, path)
                    except:
                        with open(path, 'w') as file_handle:
                            file_handle.write('->')
                            file_handle.write(real_path)




extract_tar_replace_symlinks('tgpn.tar', 'rootfs')