Skip invalid escape sequence check on 2.6

[sigmavirus24]

Python 2.6 has a bug (https://bugs.python.org/issue15054) that prevents
us from tokenizing strings properly that are b'' strings. This means we
cannot reliably detect future deprecated escape sequences on 2.6

Closes gh-755

[sigmavirus24]

cc @jwilk

[jwilk]

This condition is off-by-one: it's true in Python 2.5 (which doesn't have b'' strings, so not affected by this bug), but false in Python 2.6. How about:

if sys.version_info[:2] == (2, 6):

?

[sigmavirus24]

Thanks, I've updated the range to be correct.

[jones77]

Guessing that a unit test needs updating?

py26 runtests: commands[1] | /home/travis/build/PyCQA/pycodestyle/.tox/py26/bin/python pycodestyle.py --testsuite testsuite
testsuite/W60.py:16:1: error W605 not found
testsuite/W60.py:18:1: error W605 not found
3665 lines tested: 38 files, 405 test cases, 2 failures.
Test failed.
ERROR: InvocationError: '/home/travis/build/PyCQA/pycodestyle/.tox/py26/bin/python pycodestyle.py --testsuite testsuite'
___________________________________ summary ____________________________________
ERROR:   py26: commands failed```

[jones77]

FWIW I can reproduce this when using Python2 and trying to escape square brackets:

WW211     ERROR_LOG_PATTERN = re.compile(
  212         '(?P<date>[^ ]+)'
  213         ' (?P<time>[^ ]+)'
  214         ' \[(?P<error_type>.*)\]'

211 col 7 warning| W605 invalid escape sequence '\[' [pep8]

[sigmavirus24]

FWIW I can reproduce this when using Python2 and trying to escape square brackets:

You should absolutely see this on 2.7. It's not that it's a problem on 2.7 but if your code lives past 2020, you should be well-warned that it will break on the latest versions of 3.7 and there are things you can do today to avoid that.

[jones77]

Thanks, though I still feel like I'm missing something 'cos the docs / internet search seems to suggest escaping square brackets is ok.

https://docs.python.org/3/library/re.html

To match a literal ']' inside a set, precede it with a backslash, or place it at the beginning of the set. For example, both [()[\]{}] and []()[{}] will both match a parenthesis.

Anyhoo, gonna use [[] and []] instead.

[sigmavirus24]

\[ isn't a valid escape sequence. You can also use r'\['.

[jones77]

Thanks again, found the pbk&c, :-S the following are different.

ERROR_LOG_PATTERN_GOOD = re.compile(
    '(?P<date>[^ ]+)'
    ' (?P<time>[^ ]+)'
    r' \[(?P<error_type>.*)\]'
)

ERROR_LOG_PATTERN_BAD = re.compile(
    r'(?P<date>[^ ]+)'
    ' (?P<time>[^ ]+)'
    ' \[(?P<error_type>.*)\]'
)

[sigmavirus24]

@jones77 I'm not certain if you're looking for an explanation or if you're merely commenting on the ability to use r prefixed strings. Either way, I think that discussion is better suited to a different place than this pull request at this point.

[jdufresne]

Given the comments in #720, Python 2.6 support will be dropped. So perhaps this PR is no longer required?