"""Sub-module providing 'keyboard awareness'."""
# std imports
import platform
import time
import re
# 3rd party
import six
try:
from collections import OrderedDict
except ImportError:
# python 2.6 requires 3rd party library (backport)
#
# pylint: disable=import-error
# Unable to import 'ordereddict'
from ordereddict import OrderedDict
# curses
if platform.system() == 'Windows':
# pylint: disable=import-error
import jinxed as curses
from jinxed.has_key import _capability_names as capability_names
else:
import curses
from curses.has_key import _capability_names as capability_names
[docs]class Keystroke(six.text_type):
"""
A unicode-derived class for describing a single keystroke.
A class instance describes a single keystroke received on input,
which may contain multiple characters as a multibyte sequence,
which is indicated by properties :attr:`is_sequence` returning
``True``.
When the string is a known sequence, :attr:`code` matches terminal
class attributes for comparison, such as ``term.KEY_LEFT``.
The string-name of the sequence, such as ``u'KEY_LEFT'`` is accessed
by property :attr:`name`, and is used by the :meth:`__repr__` method
to display a human-readable form of the Keystroke this class
instance represents. It may otherwise by joined, split, or evaluated
just as as any other unicode string.
"""
[docs] def __new__(cls, ucs='', code=None, name=None):
"""Class constructor."""
new = six.text_type.__new__(cls, ucs)
new._name = name
new._code = code
return new
@property
def is_sequence(self):
"""Whether the value represents a multibyte sequence (bool)."""
return self._code is not None
def __repr__(self):
"""Docstring overwritten."""
return (six.text_type.__repr__(self) if self._name is None else
self._name)
__repr__.__doc__ = six.text_type.__doc__
@property
def name(self):
"""String-name of key sequence, such as ``u'KEY_LEFT'`` (str)."""
return self._name
@property
def code(self):
"""Integer keycode value of multibyte sequence (int)."""
return self._code
def get_curses_keycodes():
"""
Return mapping of curses key-names paired by their keycode integer value.
:rtype: dict
Returns dictionary of (name, code) pairs for curses keyboard constant
values and their mnemonic name. Such as code ``260``, with the value of
its key-name identity, ``u'KEY_LEFT'``.
"""
_keynames = [attr for attr in dir(curses)
if attr.startswith('KEY_')]
return dict((keyname, getattr(curses, keyname))
for keyname in _keynames)
[docs]def get_keyboard_codes():
"""
Return mapping of keycode integer values paired by their curses key-name.
:rtype: dict
Returns dictionary of (code, name) pairs for curses keyboard constant
values and their mnemonic name. Such as key ``260``, with the value of
its identity, ``u'KEY_LEFT'``. These are derived from the attributes by
the same of the curses module, with the following exceptions:
* ``KEY_DELETE`` in place of ``KEY_DC``
* ``KEY_INSERT`` in place of ``KEY_IC``
* ``KEY_PGUP`` in place of ``KEY_PPAGE``
* ``KEY_PGDOWN`` in place of ``KEY_NPAGE``
* ``KEY_ESCAPE`` in place of ``KEY_EXIT``
* ``KEY_SUP`` in place of ``KEY_SR``
* ``KEY_SDOWN`` in place of ``KEY_SF``
This function is the inverse of :func:`get_curses_keycodes`. With the
given override "mixins" listed above, the keycode for the delete key will
map to our imaginary ``KEY_DELETE`` mnemonic, effectively erasing the
phrase ``KEY_DC`` from our code vocabulary for anyone that wishes to use
the return value to determine the key-name by keycode.
"""
keycodes = OrderedDict(get_curses_keycodes())
keycodes.update(CURSES_KEYCODE_OVERRIDE_MIXIN)
# invert dictionary (key, values) => (values, key), preferring the
# last-most inserted value ('KEY_DELETE' over 'KEY_DC').
return dict(zip(keycodes.values(), keycodes.keys()))
[docs]def _alternative_left_right(term):
r"""
Determine and return mapping of left and right arrow keys sequences.
:arg blessed.Terminal term: :class:`~.Terminal` instance.
:rtype: dict
This function supports :func:`get_terminal_sequences` to discover
the preferred input sequence for the left and right application keys.
Return dict of sequences ``term._cuf1``, and ``term._cub1``,
valued as ``KEY_RIGHT``, ``KEY_LEFT`` (when appropriate). It is
necessary to check the value of these sequences to ensure we do not
use ``u' '`` and ``u'\b'`` for ``KEY_RIGHT`` and ``KEY_LEFT``,
preferring their true application key sequence, instead.
"""
keymap = dict()
if term._cuf1 and term._cuf1 != u' ':
keymap[term._cuf1] = curses.KEY_RIGHT
if term._cub1 and term._cub1 != u'\b':
keymap[term._cub1] = curses.KEY_LEFT
return keymap
[docs]def get_keyboard_sequences(term):
r"""
Return mapping of keyboard sequences paired by keycodes.
:arg blessed.Terminal term: :class:`~.Terminal` instance.
:returns: mapping of keyboard unicode sequences paired by keycodes
as integer. This is used as the argument ``mapper`` to
the supporting function :func:`resolve_sequence`.
:rtype: OrderedDict
Initialize and return a keyboard map and sequence lookup table,
(sequence, keycode) from :class:`~.Terminal` instance ``term``,
where ``sequence`` is a multibyte input sequence of unicode
characters, such as ``u'\x1b[D'``, and ``keycode`` is an integer
value, matching curses constant such as term.KEY_LEFT.
The return value is an OrderedDict instance, with their keys
sorted longest-first.
"""
# A small gem from curses.has_key that makes this all possible,
# _capability_names: a lookup table of terminal capability names for
# keyboard sequences (fe. kcub1, key_left), keyed by the values of
# constants found beginning with KEY_ in the main curses module
# (such as KEY_LEFT).
#
# latin1 encoding is used so that bytes in 8-bit range of 127-255
# have equivalent chr() and unichr() values, so that the sequence
# of a kermit or avatar terminal, for example, remains unchanged
# in its byte sequence values even when represented by unicode.
#
sequence_map = dict((
(seq.decode('latin1'), val)
for (seq, val) in (
(curses.tigetstr(cap), val)
for (val, cap) in capability_names.items()
) if seq
) if term.does_styling else ())
sequence_map.update(_alternative_left_right(term))
sequence_map.update(DEFAULT_SEQUENCE_MIXIN)
# This is for fast lookup matching of sequences, preferring
# full-length sequence such as ('\x1b[D', KEY_LEFT)
# over simple sequences such as ('\x1b', KEY_EXIT).
return OrderedDict((
(seq, sequence_map[seq]) for seq in sorted(
sequence_map.keys(), key=len, reverse=True)))
def get_leading_prefixes(sequences):
"""
Return a set of proper prefixes for given sequence of strings.
:arg iterable sequences
:rtype: set
Given an iterable of strings, all textparts leading up to the final
string is returned as a unique set. This function supports the
:meth:`~.Terminal.inkey` method by determining whether the given
input is a sequence that **may** lead to a final matching pattern.
>>> prefixes(['abc', 'abdf', 'e', 'jkl'])
set([u'a', u'ab', u'abd', u'j', u'jk'])
"""
return set(seq[:i] for seq in sequences for i in range(1, len(seq)))
def resolve_sequence(text, mapper, codes):
r"""
Return :class:`Keystroke` instance for given sequence ``text``.
The given ``text`` may extend beyond a matching sequence, such as
``u\x1b[Dxxx`` returns a :class:`Keystroke` instance of attribute
:attr:`Keystroke.sequence` valued only ``u\x1b[D``. It is up to
determine that ``xxx`` remains unresolved.
:arg text: string of characters received from terminal input stream.
:arg OrderedDict mapper: unicode multibyte sequences, such as ``u'\x1b[D'``
paired by their integer value (260)
:arg dict codes: a :type:`dict` of integer values (such as 260) paired
by their mnemonic name, such as ``'KEY_LEFT'``.
:rtype: Keystroke
"""
for sequence, code in mapper.items():
if text.startswith(sequence):
return Keystroke(ucs=sequence, code=code, name=codes[code])
return Keystroke(ucs=text and text[0] or u'')
def _time_left(stime, timeout):
"""
Return time remaining since ``stime`` before given ``timeout``.
This function assists determining the value of ``timeout`` for
class method :meth:`~.Terminal.kbhit` and similar functions.
:arg float stime: starting time for measurement
:arg float timeout: timeout period, may be set to None to
indicate no timeout (where None is always returned).
:rtype: float or int
:returns: time remaining as float. If no time is remaining,
then the integer ``0`` is returned.
"""
return max(0, timeout - (time.time() - stime)) if timeout else timeout
def _read_until(term, pattern, timeout):
"""
Convenience read-until-pattern function, supporting :meth:`~.get_location`.
:arg blessed.Terminal term: :class:`~.Terminal` instance.
:arg float timeout: timeout period, may be set to None to indicate no
timeout (where 0 is always returned).
:arg str pattern: target regular expression pattern to seek.
:rtype: tuple
:returns: tuple in form of ``(match, str)``, *match*
may be :class:`re.MatchObject` if pattern is discovered
in input stream before timeout has elapsed, otherwise
None. ``str`` is any remaining text received exclusive
of the matching pattern).
The reason a tuple containing non-matching data is returned, is that the
consumer should push such data back into the input buffer by
:meth:`~.Terminal.ungetch` if any was received.
For example, when a user is performing rapid input keystrokes while its
terminal emulator surreptitiously responds to this in-band sequence, we
must ensure any such keyboard data is well-received by the next call to
term.inkey() without delay.
"""
stime = time.time()
match, buf = None, u''
# first, buffer all pending data. pexpect library provides a
# 'searchwindowsize' attribute that limits this memory region. We're not
# concerned about OOM conditions: only (human) keyboard input and terminal
# response sequences are expected.
while True:
# block as long as necessary to ensure at least one character is
# received on input or remaining timeout has elapsed.
ucs = term.inkey(timeout=_time_left(stime, timeout))
if ucs:
buf += ucs
# while the keyboard buffer is "hot" (has input), we continue to
# aggregate all awaiting data. We do this to ensure slow I/O
# calls do not unnecessarily give up within the first 'while' loop
# for short timeout periods.
while True:
ucs = term.inkey(timeout=0)
if not ucs:
break
buf += ucs
match = re.search(pattern=pattern, string=buf)
if match is not None:
# match
break
if timeout is not None:
if not _time_left(stime, timeout):
# timeout
break
return match, buf
[docs]def _inject_curses_keynames():
r"""
Inject KEY_NAMES that we think would be useful into the curses module.
This function compliments the global constant
:obj:`DEFAULT_SEQUENCE_MIXIN`. It is important to note that this
function has the side-effect of **injecting** new attributes to the
curses module, and is called from the global namespace at time of
import.
Though we may determine *keynames* and codes for keyboard input that
generate multibyte sequences, it is also especially useful to aliases
a few basic ASCII characters such as ``KEY_TAB`` instead of ``u'\t'`` for
uniformity.
Furthermore, many key-names for application keys enabled only by context
manager :meth:`~.Terminal.keypad` are surprisingly absent. We inject them
here directly into the curses module.
It is not necessary to directly "monkeypatch" the curses module to
contain these constants, as they will also be accessible as attributes
of the Terminal class instance, they are provided only for convenience
when mixed in with other curses code.
"""
_lastval = max(get_curses_keycodes().values())
for key in ('TAB', 'KP_MULTIPLY', 'KP_ADD', 'KP_SEPARATOR', 'KP_SUBTRACT',
'KP_DECIMAL', 'KP_DIVIDE', 'KP_EQUAL', 'KP_0', 'KP_1', 'KP_2',
'KP_3', 'KP_4', 'KP_5', 'KP_6', 'KP_7', 'KP_8', 'KP_9'):
_lastval += 1
setattr(curses, 'KEY_{0}'.format(key), _lastval)
_inject_curses_keynames()
#: In a perfect world, terminal emulators would always send exactly what
#: the terminfo(5) capability database plans for them, accordingly by the
#: value of the ``TERM`` name they declare.
#:
#: But this isn't a perfect world. Many vt220-derived terminals, such as
#: those declaring 'xterm', will continue to send vt220 codes instead of
#: their native-declared codes, for backwards-compatibility.
#:
#: This goes for many: rxvt, putty, iTerm.
#:
#: These "mixins" are used for *all* terminals, regardless of their type.
#:
#: Furthermore, curses does not provide sequences sent by the keypad,
#: at least, it does not provide a way to distinguish between keypad 0
#: and numeric 0.
DEFAULT_SEQUENCE_MIXIN = (
# these common control characters (and 127, ctrl+'?') mapped to
# an application key definition.
(six.unichr(10), curses.KEY_ENTER),
(six.unichr(13), curses.KEY_ENTER),
(six.unichr(8), curses.KEY_BACKSPACE),
(six.unichr(9), curses.KEY_TAB),
(six.unichr(27), curses.KEY_EXIT),
(six.unichr(127), curses.KEY_DC),
(u"\x1b[A", curses.KEY_UP),
(u"\x1b[B", curses.KEY_DOWN),
(u"\x1b[C", curses.KEY_RIGHT),
(u"\x1b[D", curses.KEY_LEFT),
(u"\x1b[F", curses.KEY_END),
(u"\x1b[H", curses.KEY_HOME),
# not sure where these are from .. please report
(u"\x1b[K", curses.KEY_END),
(u"\x1b[U", curses.KEY_NPAGE),
(u"\x1b[V", curses.KEY_PPAGE),
# keys sent after term.smkx (keypad_xmit) is emitted, source:
# http://www.xfree86.org/current/ctlseqs.html#PC-Style%20Function%20Keys
# http://fossies.org/linux/rxvt/doc/rxvtRef.html#KeyCodes
#
# keypad, numlock on
(u"\x1bOM", curses.KEY_ENTER), # return
(u"\x1bOj", curses.KEY_KP_MULTIPLY), # *
(u"\x1bOk", curses.KEY_KP_ADD), # +
(u"\x1bOl", curses.KEY_KP_SEPARATOR), # ,
(u"\x1bOm", curses.KEY_KP_SUBTRACT), # -
(u"\x1bOn", curses.KEY_KP_DECIMAL), # .
(u"\x1bOo", curses.KEY_KP_DIVIDE), # /
(u"\x1bOX", curses.KEY_KP_EQUAL), # =
(u"\x1bOp", curses.KEY_KP_0), # 0
(u"\x1bOq", curses.KEY_KP_1), # 1
(u"\x1bOr", curses.KEY_KP_2), # 2
(u"\x1bOs", curses.KEY_KP_3), # 3
(u"\x1bOt", curses.KEY_KP_4), # 4
(u"\x1bOu", curses.KEY_KP_5), # 5
(u"\x1bOv", curses.KEY_KP_6), # 6
(u"\x1bOw", curses.KEY_KP_7), # 7
(u"\x1bOx", curses.KEY_KP_8), # 8
(u"\x1bOy", curses.KEY_KP_9), # 9
# keypad, numlock off
(u"\x1b[1~", curses.KEY_FIND), # find
(u"\x1b[2~", curses.KEY_IC), # insert (0)
(u"\x1b[3~", curses.KEY_DC), # delete (.), "Execute"
(u"\x1b[4~", curses.KEY_SELECT), # select
(u"\x1b[5~", curses.KEY_PPAGE), # pgup (9)
(u"\x1b[6~", curses.KEY_NPAGE), # pgdown (3)
(u"\x1b[7~", curses.KEY_HOME), # home
(u"\x1b[8~", curses.KEY_END), # end
(u"\x1b[OA", curses.KEY_UP), # up (8)
(u"\x1b[OB", curses.KEY_DOWN), # down (2)
(u"\x1b[OC", curses.KEY_RIGHT), # right (6)
(u"\x1b[OD", curses.KEY_LEFT), # left (4)
(u"\x1b[OF", curses.KEY_END), # end (1)
(u"\x1b[OH", curses.KEY_HOME), # home (7)
# The vt220 placed F1-F4 above the keypad, in place of actual
# F1-F4 were local functions (hold screen, print screen,
# set up, data/talk, break).
(u"\x1bOP", curses.KEY_F1),
(u"\x1bOQ", curses.KEY_F2),
(u"\x1bOR", curses.KEY_F3),
(u"\x1bOS", curses.KEY_F4),
)
#: Override mixins for a few curses constants with easier
#: mnemonics: there may only be a 1:1 mapping when only a
#: keycode (int) is given, where these phrases are preferred.
CURSES_KEYCODE_OVERRIDE_MIXIN = (
('KEY_DELETE', curses.KEY_DC),
('KEY_INSERT', curses.KEY_IC),
('KEY_PGUP', curses.KEY_PPAGE),
('KEY_PGDOWN', curses.KEY_NPAGE),
('KEY_ESCAPE', curses.KEY_EXIT),
('KEY_SUP', curses.KEY_SR),
('KEY_SDOWN', curses.KEY_SF),
('KEY_UP_LEFT', curses.KEY_A1),
('KEY_UP_RIGHT', curses.KEY_A3),
('KEY_CENTER', curses.KEY_B2),
('KEY_BEGIN', curses.KEY_BEG),
)
__all__ = ('Keystroke', 'get_keyboard_codes', 'get_keyboard_sequences',)