红宝石2 - 540个 447 420字符
运行方式“ruby2.0 jumper.rb'指示'初始化数据'”。1.x Ruby无法正常工作（没有String.bytes方法）。
添加了多行命令和注释，并改善了我的推杆。
i=$*[0].gsub(/\([^)]*\)/m,' ').scan(/(\??)\s*([#=:><+-])\s*(\d*)/m).map{|a|[a[0]!='?',a[1],a[2]==''?/[#=:]/=~a[1]?0:1:a[2].to_i]}
N=i.size
d=$*[1].bytes
r=p=0
while p<N
u,o,x=i[p]
p+=1
d[r]=0 if d[r].nil?
case o
when'#';r=x
when'>';r+=x
when'<';r-=x
when/[=+-]/;eval "d[r]#{o.tr'=',''}=x";d[r]%=256
when':';p=x;abort'Error'if p>=N
end if u||d[r]>0
abort'Error'if r<0
end
printf"%s\n",d.take_while{|v|v&&v!=0}.pack('C*')
这是一个包含一些散布测试的测试套件。使用它的最简单方法是将代码填充到t / jumper.t中并运行“ perl t / jumper.t”。
#/usr/bin/perl
use strict;
use warnings;
# timestamp: 2014 August 3, 19:00
#
# - Assume program takes machine code and initialization string as command
# line options.
# - Assume all required errors reported as "Error\n".
# - Go with the flow and suffix output with \n. Merged terminal newlines are
# unacceptable [I'm talkin' to YOU Ruby puts()!].
# - As per OP - jumping to > end-of-program must be an error.
use Test::More qw(no_plan);
# use Test::More tests => 4;
my $jumper = "jumper.rb";
#
# "happy" path
#
# starter tests provided by OP
is( `$jumper '=72>=101>=108>=108>=111>=32>=119>=111>=114>=108>=100>=33>=' '' 2>&1`, "Hello world!\n", "hello world (from user2992539)");
is( `$jumper '?:2 :4 >1 :0 =33 >1 =0' 'a' 2>&1`, "a!\n", 'append !, #1 (from user2992539)');
# simple variations
is( `$jumper '?:2 :4 >1 :0 =33 >1 =0' '' 2>&1`, "!\n", 'append !, #2');
is( `$jumper '?:2 :4 >1 :0 =33' '' 2>&1`, "!\n", 'append !, #3, no NUL');
# comment delimiters don't nest
is( `$jumper "(()=" 'oops' 2>&1`, "\n", "() don't nest");
# comments and termination
is( `$jumper '(start with a comment)?(comment w/ trailing sp) # (comment w/ surrounding sp) 1 =98' 'a' 2>&1`, "ab\n", 'walk to exit');
is( `$jumper '(start with a comment)? (comment w/ leading sp)= (comment w/ surrounding sp) 97()' '' 2>&1`, "\n", 'skip to exit');
is( `$jumper '#1=0 (actually two instructions, but it scans well) :5 #=(truncate further if not jumped over)' 'a b' 2>&1`, "Error\n", 'truncate & jump to exit');
# is RAM pointer initialized to 0?
is( `$jumper '-103(g-g) ?:1025(exit) =103 #4=10' 'good' 2>&1`, "good\n\n", 'intial string in right place?');
# TBD, do jumps work?
# TBD, do conditional jumps work?
# jump right to a harder case, copy byte 0 to byte 3 and format, e.g. input="Y" output="Y=>Y"
is( `$jumper '#1=61#2=62#4=0#3=#10=#(11:)?:13:20(13:)#3+#10+#0-:11(20:)#10(21:)?:23:28(23:)#0+#10-:21(28:)#' 'Y' 2>&1`, "Y=>Y\n", 'copy a byte');
# test memory allocation by dropping 255s at increasingly large intervals
is( `$jumper '#16=511 #64=511 #256=511 #1024=511 #4096=511 #16384=511 #65536=511 #262144=511 #1048576=511 #65536-255 (20:)?:23(exit) #=' 'wrong' 2>&1`, "\n", 'test alloc()');
# upcase by subtraction
is( `$jumper '-32' 't' 2>&1`, "T\n", 'upcase via subtraction');
# 2 nested loops to upcase a character, like so: #0=2; do { #0--; #1=16; do { #1--; #2--; } while (#1); } while (#0);
is( `$jumper '#=2 (2:)#- #1=16 (6:)#1- #2- #1?:6 #0?:2 #=32 #1=32' ' t' 2>&1`, " T\n", 'upcase via loops');
# downcase by addition
is( `$jumper '+32' 'B' 2>&1`, "b\n", 'downcase via addition');
# same thing with a loop, adjusted to walk the plank instead of jumping off it
is( `$jumper '#1 ?:3 :7 -<+ :0 #' 'B ' 2>&1`, "b\n", 'downcase via adder (from Sieg)');
# base 10 adder with carry
is( `$jumper '#0-48#10=9#11=#5=#0(9:)?:11:22(11:)#10?:14:22(14:)-#11+#5+#0-:9(22:)#0?:110#11(25:)?:27:32(27:)#0+#11-:25(32:)#0+48>-43?:110=43>-48#10=9#11=#2(45:)?:47:58(47:)#10?:50:58(50:)-#11+#5+#2-:45(58:)#2?:110#11(61:)?:63:68(63:)#2+#11-:61(68:)#2+48>-61?:110=61>?:110=32#10=9#11=#5-10(83:)?:85:94(85:)#10?:88:94(88:)-#11+#5-:83(94:)#5?:99#4=49:100(99:)+10(100:)#11(101:)?:103:108(103:)#5+#11-:101(108:)#5+48' '1+1=' 2>&1`, "1+1= 2\n", 'base 10 adder, #1');
is( `$jumper '#0-48#10=9#11=#5=#0(9:)?:11:22(11:)#10?:14:22(14:)-#11+#5+#0-:9(22:)#0?:110#11(25:)?:27:32(27:)#0+#11-:25(32:)#0+48>-43?:110=43>-48#10=9#11=#2(45:)?:47:58(47:)#10?:50:58(50:)-#11+#5+#2-:45(58:)#2?:110#11(61:)?:63:68(63:)#2+#11-:61(68:)#2+48>-61?:110=61>?:110=32#10=9#11=#5-10(83:)?:85:94(85:)#10?:88:94(88:)-#11+#5-:83(94:)#5?:99#4=49:100(99:)+10(100:)#11(101:)?:103:108(103:)#5+#11-:101(108:)#5+48' '9+9=' 2>&1`, "9+9=18\n", 'base 10 adder, #2');
# order of assignment shouldn't affect order of print
is( `$jumper '#1=98 #0=97' '' 2>&1`, "ab\n", 'print order != assignment order');
# are chars modulo 256?
is( `$jumper '#10(#10 defaults to 0) +255+(#10 += 256) ?#(skip if #10==0) =' 'good' 2>&1`, "good\n", 'memory values limited to 0<x<255');
# go for the cycle;
is( `$jumper '(0:)+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ (256:)#4=10' 'BCID' 2>&1`, "ACID\n\n", 'cycle character less 1, PC>255');
# same thing with a loop;
is( `$jumper '#4=255(#4 = 255) (2:)#1+(#1++) #4-(#4--) ?:2(loop 255 times) #4=10(#4 = NL)' 'ADID' 2>&1`, "ACID\n\n", 'cycle character less 1, PC>255');
# Exercise the program counter.
# PC > 255;
is( `$jumper '(0:)= (1:)############################################################################################################################################################################################################################################################### (256:)?:259 (257:)+ (258:):1 (259:)=97#3=10' 'a==' 2>&1`, "a==\n\n", 'program counter range >255');
#
# "sad" path
#
# Error checking required by the specification.
#
# simplest test case of PC going out of bounds
is( `$jumper ':2' '' 2>&1`, "Error\n", 'program counter too big by 1');
is( `$jumper ':1024' '' 2>&1`, "Error\n", 'program counter in space');
is( `$jumper ':1073741824' '' 2>&1`, "Error\n", 'program counter in hyperspace');
# try to drive program counter negative, if 32-bit signed integer
is( `$jumper ':2147483648(exit)' 'ridiculous speed' 2>&1`, "Error\n", 'program counter goes negative?, #1');
# try to drive program counter negative, if 64-bit signed integer
is( `$jumper ':9223372036854775808 (exit)' 'ludicrous speed' 2>&1`, "Error\n", 'program counter goes negative?, #2');
# spaces not allowed in operand; error or silently ignore (my choice)
isnt(`$jumper '#= #= #= #= #= +1 4 ' 'aops' 2>&1`, "oops\n", 'do not accept spaces in operands');
# ditto w/ a comment ; error or silently ignore (my choice)
isnt(`$jumper '#= #= #= #= #= +1(not valid)4 ' 'aops' 2>&1`, "oops\n", 'do not accept spaces in operands');
# RAM pointer error-checking; "Error" or "" are OK
isnt( `$jumper '<>=' 'oops' 2>&1 | grep -v Error`, "oops\n", 'unused negative RAM pointer behavior unspecified');
# RAM pointer negative and use it
is( `$jumper '<=' '' 2>&1`, "Error\n", 'cannot use negative RAM pointer, #1');
# check for RAM pointer wrap-around
is( `$jumper '<=' '0123456789' 2>&1`, "Error\n", 'cannot use negative RAM pointer, #2');
# The way I read this
# "Commands and arguments may be delimited with spaces or new lines but
# not necessary."
# multi-line commands are legit.
is( `$jumper "#4#?\n=" 'oops' 2>&1`, "\n", 'multi-line commands allowed');
# Multi-line comments would be consistent with multi-line commands, but I can't
# find something I can translate into a "must" or "must not" requirement in
# "Program can have comments between (). ... Comments can be placed
# anywhere."
# Until uncertainty resolved, no test case.
#
# "bad" path
#
# These tests violate the assumption that the instruction stream is wellll-farmed.
#
# characters not in the language; error or (my choice) silently skip
isnt(`$jumper 'x =' 'oops' 2>&1`, "oops\n", 'opcode discrimination');
# is ? accepted as an operator (vs operation modifier); error or (my choice) silently skip
is(`$jumper '(bad 0, good 0:)??0 (bad 1, good 0:):3 (bad 2, good 1:)#0' '' 2>&1`, "Error\n", '? not accepted as an opcode');
exit 0;
非高尔夫版本。
#
# Turing Machine Mach 2.0.
# Tape? Tape? We don't need no stinkin' tape! We gots RAM!
#
# dM = data memory
# iM = instruction memory
# pC = program counter
# rP = RAM pointer
# u, o, x = current instruction being executed
#
# N = number of instructions in instruction memory
#
# instruction decoder
iM = $*[0].gsub(/\([^)]*\)/m,' ').scan(/(\??)\s*([#=:><+-])\s*(\d*)/m).map { |a|
[
a[0] != '?',
a[1],
(a[2] == '') ? (/[#=:]/ =~ a[1] ? 0 : 1) : a[2].to_i
]
}
pC = 0
N = iM.size
dM = $*[1].bytes
rP = 0
while pC < N do
# u, unconditional instruction, execute if true || (dM[rP] > 0)
# skip if false && (dM[rP] == 0)
# o, operator
# x, operand
(u, o, x) = iM[pC]
pC += 1
dM[rP] = 0 if dM[rP].nil?
if u || (dM[rP] > 0)
case o
when '#'
rP = x
when '>'
rP += x
when '<'
rP -= x
when /[=+-]/
eval "dM[rP]#{o.tr'=',''}=x"
dM[rP] %= 256
when ':'
pC = x
abort 'Error' if pC >= N
end
end
abort 'Error' if rP < 0
end
printf "%s\n", dM.take_while{|v|v&&v!=0}.pack('C*')
快速原型装配器。
#
# Jumper "assembler" - symbolic goto labels.
#
# what it does:
# - translates labels/targets into absolute position
# @label ?:good_exit
# ...
# :label
#
# - a label is [a-zA-Z][a-zA-Z0-9_]*
# - a target is @label
# - one special label:
# - "hyperspace" is last instruction index + 1
# - strips out user comments
# - everything from "//" to EOL is stripped
# - jumper comments are stripped
# - adds "label" comments of the form "(ddd:)"
# limitations & bugs:
# - multi-line jumper comments aren't alway handled gracefully
# - a target not followed by an instruction will reference
# the previous instruction. this can only happen
# at the end of the program. recommended idiom to
# avoid this:
# @good_exit #
# what it doesn't do:
# - TBD, simple error checking
# - labels defined and not used
# - TBD, symbolic memory names
#
# Example:
#
# input -
# (
# adder from Sieg
# )
# @loop_head # 1 // while (*(1)) {
# ?:continue
# :good_exit
#
# @continue - // *(1) -= 1;
# <- // *(0) += 1;
# +
# :loop_head // }
# @good_exit #
#
# output -
# (0:) #1 ?:3 :7 (3:) - < + :0 (7:)#
rawSource = ARGF.map do |line|
line.gsub(/\([^)]*\)/, ' ') # eat intra-line jumper comments
.gsub(/\/\/.*/, ' ') # eat C99 comments
.gsub(/^/, "#{$<.filename}@#{$<.file.lineno}\n") # add line ID
end.join
rawSource.gsub! /\([^)]*\)/m, '' # eat multi-line jumper comments
#
# Using example from above
#
# rawSource =
# "sieg.ja@1\n \n" +
# "sieg.ja@4\n@loop_head # 1\n"
# ...
# "sieg.ja@12\n@good_exit # \n"
instructionPattern = %r{
(?<label> [[:alpha:]]\w* ){0}
(?<operator> \??\s*[#=:><+-]) {0}
(?<operand> \d+|[[:alpha:]]\w* ){0}
\G\s*(@\g<label>\s*)?(\g<operator>\s*)?(\g<operand>)?
}x
FAIL = [nil, nil, nil]
instructionOffset = 0
iStream = Array.new
target = Hash.new
targetComment = nil
for a in rawSource.lines.each_slice(2) do
# only parse non-empty lines
if /\S/ =~ a[1]
m = nil
catch( :parseError ) do
chopped = a[1]
while m = instructionPattern.match(chopped)
if m.captures.eql?(FAIL) || (!m[:operator] && m[:operand])
m = nil
throw :parseError
end
if m[:label]
if target.has_key?(m[:label].to_sym)
printf $stderr, a[0].chomp + ": error: label '#{m[:label]}' is already defined"
abort a[1]
end
target[ m[:label].to_sym ] = instructionOffset
targetComment = "(#{instructionOffset}:)"
end
if m[:operator]
iStream[instructionOffset] = [
targetComment,
m[:operator],
/\A[[:alpha:]]/.match(m[:operand]) ? m[:operand].to_sym : m[:operand]
]
targetComment = nil
instructionOffset += 1
end
chopped = m.post_match
if /\A\s*\Z/ =~ chopped
# nothing parseable left
break
end
end
end
if !m
printf $stderr, a[0].chomp + ": error: parse failure"
abort a[1]
end
end
end
# inject hyperspace label
target[:hyperspace] = instructionOffset
# replace operands that are labels
iStream.each do |instruction|
if instruction[2]
if !(/\A\d/ =~ instruction[2]) # its a label
if target.has_key?(instruction[2])
instruction[2] = target[instruction[2]]
else
abort "error: label '@#{instruction[2]}' is used but not defined"
end
end
end
puts instruction.join
end