Bugs in curses.ascii predicates

[serhiy-storchaka]

BPO	27079
Nosy	@Yhg1s, @4kir4, @serhiy-storchaka
Files	curses_ascii.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/serhiy-storchaka'
closed_at = <Date 2016-06-18.19:14:55.922>
created_at = <Date 2016-05-21.20:09:59.924>
labels = ['type-bug', 'library']
title = 'Bugs in curses.ascii predicates'
updated_at = <Date 2016-07-01.16:59:23.201>
user = 'https://github.com/serhiy-storchaka'

bugs.python.org fields:

activity = <Date 2016-07-01.16:59:23.201>
actor = 'akira'
assignee = 'serhiy.storchaka'
closed = True
closed_date = <Date 2016-06-18.19:14:55.922>
closer = 'serhiy.storchaka'
components = ['Library (Lib)']
creation = <Date 2016-05-21.20:09:59.924>
creator = 'serhiy.storchaka'
dependencies = []
files = ['42931']
hgrepos = []
issue_num = 27079
keywords = ['patch']
message_count = 5.0
messages = ['266020', '268825', '269562', '269573', '269679']
nosy_count = 3.0
nosy_names = ['twouters', 'akira', 'serhiy.storchaka']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue27079'
versions = ['Python 2.7', 'Python 3.5', 'Python 3.6']

[serhiy-storchaka]

Most curses.ascii predicates corresponds C functions declared in <ctype.h>. But there are some differences:

1. isblank() returns True for space and backspace instead of space and tab.

2. ispunct() returns True for non-ASCII and control characters.

3. iscntrl() returns False for '\x7f'. Note that there is different function isctrl().

These differences look as bugs. Proposed patch fixes them and adds tests for all curses.ascii functions.

[serhiy-storchaka]

Committed with wrong commit message (damn two clipboards on X11) in ac8338546ca8, d7b6c07bc713, d03c8d0a6a73.

[4kir4]

There is an overlapping issue from 2010: "curses.ascii.isblank() function is broken. It confuses backspace (BS 0x08) with tab (0x09)" http://bugs.python.org/issue9770

Your patch fixes it too (it should be closed). Note: the patch does not pass tests from Lib/test/test_curses_ascii.py attached to bpo-9770 (even if the code: if char_class_name in ('cntrl', 'punct') test = unittest.expectedFailure(test) is removed) e.g., iscntrl(-1) should be False but it returns True:

  $ ./python
  >>> import curses.ascii
  >>> curses.ascii.iscntrl(-1) #XXX expected False
  True

If we ignore negative ints then isblank, ispunct, iscntrl provided in the curses_ascii.patch are ok.

[serhiy-storchaka]

Since screen.getch() can return -1, it looks reasonable to make curses.ascii predicates to work with negative integers.

Do you want to open a new issue and write a patch Akira?

[4kir4]

I'm not sure anything should be done (e.g., it is "undefined behavior" to pass a negative value such as CHAR_MIN (if *char* type is signed) to a character classification function in C. Though EOF value (-1 traditionally) should be handled).

If you want to explore it further; I've enumerated open questions in 2014 (inconsistent TypeError, ord(c) > 0x100, negative ints handling, etc) http://bugs.python.org/issue9770#msg221008