canonical_path

canonical_path.py

import sys

MAX_FILENAME_LENGTH = 255
MAX_FILEPATH_LENGTH = 4096
SEP = '/'

"""
Дана строка, которая является абсолютным путем к файлу или директории в системе Unix.
Нужно упростить эту строку до каноничного пути и написать автотесты для своей реализации. Использовать Python + pytest
"""


def canonical_path(path: str) -> str:
    if sys.platform != "linux":
        raise OSError('Works only on linux os')

    canonical = ['']
    if not path.strip():
        raise ValueError('Empty path value!')

    if not path.startswith(SEP):
        raise ValueError(f'Not an absolute path! Path should start with {SEP}')
    
    if path.strip() == SEP:
        return SEP

    tokens = path.split(SEP)

    for token in tokens:
        if token and token != '.':
            if token != '..':
                if len(token) > MAX_FILENAME_LENGTH:
                    raise ValueError('File or directory name too long!')
                canonical.append(f'{token}')
            else:
                del canonical[-1]
                # there are few cases when linux realpath differs from python os.path.realpath behavior
                # as I compare results with python implementation, I took Python approach
                if not canonical:
                    canonical.append('')

    if len(canonical) == 1:
        return f'/{canonical[0]}'
    else:
        res = '/'.join(canonical)
        if len(res) > MAX_FILEPATH_LENGTH:
            raise ValueError('Resulted file path too long!')
        else:
            return res

lib.py

from math import ceil

from canonical_path import MAX_FILEPATH_LENGTH


def generate_long_path(length: int = MAX_FILEPATH_LENGTH) -> str:
    prefix = '/home/user'
    fn = '/{0}'.format('AbCdE' * 51)

    mul = ceil((MAX_FILEPATH_LENGTH - len(prefix)) / len(fn))
    final_path = '{p}{f}'.format(p=prefix, f=fn * mul)

    return final_path[:length]

test_negative.py

import sys

import pytest

from canonical_path import canonical_path, MAX_FILENAME_LENGTH, MAX_FILEPATH_LENGTH
from lib import generate_long_path

MAX_FILE_NAME_PLUS = 'a' * (MAX_FILENAME_LENGTH + 1)
MAX_PATH_PLUS = generate_long_path(MAX_FILEPATH_LENGTH + 1)


@pytest.mark.parametrize(
    'initial_path, msg', [
        pytest.param('user/data', 'Not an absolute path! Path should start with /', id='Valid path'),
        pytest.param(' ', 'Empty path value!', id='Empty path'),
        pytest.param('/home/data/{0}'.format(MAX_FILE_NAME_PLUS), 'File or directory name too long!',
                     id='Exceed linux file length'),
        pytest.param(MAX_PATH_PLUS, 'Resulted file path too long!', id='Exceed linux path length')
    ]
)
def test_negative(initial_path, msg):
    with pytest.raises(ValueError) as excinfo:
        canonical_path(initial_path)

    assert str(excinfo.value) == msg


def test_invalid_platform(monkeypatch):
    monkeypatch.setattr(sys, 'platform', 'win32')

    with pytest.raises(OSError) as excinfo:
        canonical_path('c:/dir')

    assert str(excinfo.value) == 'Works only on linux os'

test_positive.py

from os import path

import pytest

from canonical_path import canonical_path, MAX_FILENAME_LENGTH
from lib import generate_long_path

MAX_FILE_NAME = 'a' * MAX_FILENAME_LENGTH
MAX_PATH = generate_long_path()


@pytest.mark.parametrize(
    'initial_path', [
        pytest.param('/', id='Root path'),
        pytest.param('//', id='Root path with double /'),
        pytest.param('/../data', id='Root dir'),
        pytest.param('/home/data/file.txt', id='Valid path'),
        pytest.param('/home/data///dir/', id='Path ends with /'),
        pytest.param('/home/data/../data.txt', id='Path contains multiple dots'),
        pytest.param('/home/data/{0}'.format(MAX_FILE_NAME), id='Max unix filename length'),
        pytest.param(MAX_PATH, id='Max unix filepath length'),
        pytest.param(f'///{MAX_PATH}', id='Resulted filepath less or equal to limit'),
        pytest.param('/home/user/../test/../../../../data/text.txt', id='Multiple dots in path'),
        pytest.param('/home/user/././text.file', id='file with dot dot')
    ]
)
def test_positive(initial_path):
    assert canonical_path(initial_path) == path.realpath(initial_path)