night/regex/bitwise.sed

# Common bitwise operations using regular expressions
#
# Copyright (C) 2018 Mike Gerwitz
#
#  This program is free software: you can redistribute it and/or modify
#  it under the terms of the GNU General Public License as published by
#  the Free Software Foundation, either version 3 of the License, or
#  (at your option) any later version.
#
#  This program is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#  GNU General Public License for more details.
#
#  You should have received a copy of the GNU General Public License
#  along with this program.  If not, see <http://www.gnu.org/licenses/>.
#
# This script implements the most common unary and binary bitwise operations
# on 8-bit (1-byte) values.  The format of the input is:
#
#   C BYTE[ BYTE]
#
# Where C is one of the commands defined below and, BYTE is an 8-bit value
# represented by 0s and 1s.  The square brackets denote an optional
# value---some operators are unary (require one argument) and others are
# binary (require two arguments).
#
# For example:
#
#   ^ 11100001 11110000
#
# will XOR the two bytes to produce `00010001'.  Whereas:
#
#   < 11111111
#
# will perform a logical left shift to produce `11111110'.
#
# The regexes below all follow common patterns.  To make that pattern clear,
# some regexes may do useless things (e.g. `.\{0\}') so that they are
# well-aligned.
#
# Transformations use `:' and `.' as intermediate values to represent 1 and
# 0 respectively.  This is necessary to ensure that one regex does not
# operate on the replacement of another (for example, NOT replacing 0 with 1
# and then replacing 1 with 0 immediately thereafter; or the dual-use of OR
# with XOR).
#
# Below, A denotes the first byte and B the second.  The term ``set'' refers
# to a bit with a value of 1 and ``clear'' a value of 0.
#
# (This could have been implemented more concisely by branching in a loop,
# but I want to be clear that this is being done with vanilla replacements
# using regexes and that such loops are not needed.)
#
# Note that all regexes operate at a fixed position; this makes them
# suitable as a template for general-purpose use in larger pattern spaces.
##

# Bitwise AND (&).  If the value of the bit in A is already clear, then we
# need not do anything, because the result will always be clear.  Otherwise,
# we need only clear the bit if the respective bit of B is clear.
s/^\(& .\{0\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{1\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{2\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{3\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{4\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{5\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{6\}\)1\(.\{8\}0\)/\1.\2/
s/^\(& .\{7\}\)1\(.\{8\}0\)/\1.\2/

# Bitwise OR (|) or XOR (^).  This logic is shared for both operations (see
# XOR below).  If the bit in A is already set, then we need not do anything,
# because the result will always be set.  Otherwise, we need only set the bit
# if the respective bit in B is set.
s/^\([|^] .\{0\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{1\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{2\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{3\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{4\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{5\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{6\}\)0\(.\{8\}1\)/\1:\2/
s/^\([|^] .\{7\}\)0\(.\{8\}1\)/\1:\2/

# Bitwise XOR (^).  We must perform two steps: first, if a bit in A is clear,
# then it should be set if the respective bit in B is set; this logic
# is handled above in OR.  Otherwise, if A is set, then it should be cleared
# if the respective bit in B is also set.
s/^\(\^ .\{0\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{1\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{2\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{3\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{4\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{5\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{6\}\)1\(.\{8\}1\)/\1.\2/
s/^\(\^ .\{7\}\)1\(.\{8\}1\)/\1.\2/

# Bitwise NOT (~).  This is a unary operation.  A bit in A is set if it is
# clear and vice-versa.
s/^\(~ .\{0\}\)1/\1./;  s/^\(~ .\{0\}\)0/\1:/;
s/^\(~ .\{1\}\)1/\1./;  s/^\(~ .\{1\}\)0/\1:/;
s/^\(~ .\{2\}\)1/\1./;  s/^\(~ .\{2\}\)0/\1:/;
s/^\(~ .\{3\}\)1/\1./;  s/^\(~ .\{3\}\)0/\1:/;
s/^\(~ .\{4\}\)1/\1./;  s/^\(~ .\{4\}\)0/\1:/;
s/^\(~ .\{5\}\)1/\1./;  s/^\(~ .\{5\}\)0/\1:/;
s/^\(~ .\{6\}\)1/\1./;  s/^\(~ .\{6\}\)0/\1:/;
s/^\(~ .\{7\}\)1/\1./;  s/^\(~ .\{7\}\)0/\1:/;

# Logical left shift (<), right shift (>).  For left shifts, the first bit
# in A is discarded and a 0 is added to the end.  For right shifts, the last
# bit of A is discarded and a 0 is added to the beginning.
s/^< .\(.\{7\}\)/< \10/
s/^> \(.\{7\}\)./> 0\1/

# Arithmetic right shift (a).  Similar to a logical right shift, except that
# instead of shifting in a clear bit, the sign is maintained (in a two's
# complement system, the most significant bit is the sign bit).  An
# arithmetic left shit is the same as a logical left shift, so we do not
# provide such an operator.
s/^a \(.\)\(.\{6\}\)/a \1\1\2/

# Circular shift (rot8) left (r), right (R).  Rather than shifting in a
# clear bit, the bit that is shifted off of the end is re-added to the other
# end.  This is also called a rotation.
s/^r \(.\)\(.\{7\}\)/r \2\1/
s/^R \(.\{7\}\)\(.\)/R \2\1/


# Prepare the final output by discarding the command and second byte, and
# then replacing the temporary values `:' and `.' with their respective bits.
s/^. \(.\{8\}\).*/\1/
s/:/1/g;  s/\./0/g