Mercurial > repo
annotate bin/gs2c.py @ 12256:821155c00e34 draft
<fizzie> ` sed -e \'s|wisdom|bin|\' < ../bin/culprits > ../bin/cblprits; chmod a+x ../bin/cblprits
| author | HackEso <hackeso@esolangs.org> |
|---|---|
| date | Sat, 07 Dec 2019 23:36:22 +0000 |
| parents | c9f5afad5250 |
| children |
| rev | line source |
|---|---|
| 9075 | 1 # gs2 compiler (version 0.2) |
| 2 # (c) nooodl 2014 | |
| 3 | |
| 4 import re | |
| 5 import struct | |
| 6 import sys | |
| 7 | |
| 8 if sys.platform == "win32": | |
| 9 import os, msvcrt | |
| 10 msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY) | |
| 11 | |
| 12 mnemonics = {} | |
|
12097
c9f5afad5250
<fizzie> sled /hackenv/bin/gs2c.py//s|bin/gs2.py|/hackenv/bin/gs2.py|
HackEso <hackeso@esolangs.org>
parents:
9075
diff
changeset
|
13 with open('/hackenv/bin/gs2.py') as f: |
| 9075 | 14 for line in f: |
| 15 if '#=' in line: | |
| 16 a, b = line.split('#=') | |
| 17 a = re.findall(r'\\x(..)', a.strip()) | |
| 18 b = b.strip().split(', ') | |
| 19 assert len(a) == len(b) | |
| 20 for i, j in zip(a, b): | |
| 21 for k in j.split(): | |
| 22 mnemonics[k] = chr(int(i, 16)) | |
| 23 | |
| 24 mnemonics["'"] = '\xe0' # block1 | |
| 25 | |
| 26 for i in xrange(16): | |
| 27 mnemonics['@%d' % i] = chr(0xA0 | i) | |
| 28 mnemonics['junk%d' % i] = chr(0xA0 | i) | |
| 29 for i, c in enumerate('abcd'): | |
| 30 mnemonics['save-%s' % c] = chr(0xC8 | i) | |
| 31 mnemonics['pop-%s' % c] = chr(0xCC | i) | |
| 32 mnemonics['push-%s' % c] = chr(0xD0 | i) | |
| 33 mnemonics['nip-%s' % c] = chr(0xD4 | i) | |
| 34 mnemonics['tuck-%s' % c] = chr(0xD8 | i) | |
| 35 mnemonics['show-%s' % c] = chr(0xDC | i) | |
| 36 for i in xrange(8): | |
| 37 mnemonics['b%d' % (i+1)] = chr(0xE0 | i) | |
| 38 mnemonics['block%d' % (i+1)] = chr(0xE0 | i) | |
| 39 mnemonics['m%d' % (i+1)] = chr(0xE8 | i) | |
| 40 mnemonics['map%d' % (i+1)] = chr(0xE8 | i) | |
| 41 mnemonics['f%d' % (i+1)] = chr(0xF0 | i) | |
| 42 mnemonics['filter%d' % (i+1)] = chr(0xF0 | i) | |
| 43 if 0xF8 | i < 0xFD: | |
| 44 mnemonics['t%d' % (i+1)] = chr(0xF8 | i) | |
| 45 mnemonics['both%d' % (i+1)] = chr(0xF8 | i) | |
| 46 | |
| 47 def compile_num(i): | |
| 48 if 0 <= i <= 10: | |
| 49 return chr(i + 0x10) | |
| 50 elif i == 100: | |
| 51 return '\x1b' | |
| 52 elif i == 1000: | |
| 53 return '\x1c' | |
| 54 elif i == 16: | |
| 55 return '\x1d' | |
| 56 elif i == 64: | |
| 57 return '\x1e' | |
| 58 elif i == 256: | |
| 59 return '\x1f' | |
| 60 | |
| 61 elif 0x00 <= i <= 0xFF: | |
| 62 return '\x01' + struct.pack('B', i) | |
| 63 elif -0x8000 <= i <= 0x7FFF: | |
| 64 return '\x02' + struct.pack('<h', i) | |
| 65 elif -0x80000000 <= i <= 0x7FFFFFFF: | |
| 66 return '\x03' + struct.pack('<l', i) | |
| 67 else: | |
| 68 raise Exception("couldn't compile number: %s" % i) | |
| 69 | |
| 70 def compile_gs2(s): | |
| 71 s = '\n'.join(l for l in s.split('\n') if l and l[0] != '#') | |
| 72 tokens = re.findall(r'"[^"]*"|\S+', s) | |
| 73 string_mode = False | |
| 74 string_array = False | |
| 75 strings = [] | |
| 76 output_code = [] | |
| 77 | |
| 78 for i in tokens: | |
| 79 # sys.stderr.write('[%s]\n' % i) | |
| 80 try: | |
| 81 v = compile_num(int(i)) | |
| 82 output_code.append(v) | |
| 83 continue | |
| 84 except ValueError: pass | |
| 85 | |
| 86 if i[0] == "'" and len(i) > 1: | |
| 87 v = compile_num(ord(i[1])) | |
| 88 output_code.append(v) | |
| 89 elif i == ')': | |
| 90 string_mode = False | |
| 91 s_open = '\x04' | |
| 92 s_close = { | |
| 93 'regular': '\x05', | |
| 94 'array': '\x06', | |
| 95 'printf': '\x9b', | |
| 96 'regex-match': '\x9c', | |
| 97 'regex-sub': '\x9d', | |
| 98 'regex-find': '\x9e', | |
| 99 'regex-split': '\x9f', | |
| 100 }[string_type] | |
| 101 if len(strings) == 1 and len(strings[0]) == 1: | |
| 102 output_code.append('\x07' + strings[0]) | |
| 103 else: | |
| 104 output_code.append(s_open + '\x07'.join(strings) + s_close) | |
| 105 elif string_mode or i[0] == '"': | |
| 106 if i[0] == '"': | |
| 107 if string_mode: | |
| 108 strings.append(eval(i)) | |
| 109 elif len(eval(i)) == 1: | |
| 110 output_code.append('\x07' + eval(i)) | |
| 111 else: | |
| 112 output_code.append('\x04' + eval(i) + '\x05') | |
| 113 else: | |
| 114 strings.append(i) | |
| 115 elif i in ['(', 'w(', 'p(', 'm(', 's(', 'f(', 'v(']: | |
| 116 string_mode = True | |
| 117 string_type = { | |
| 118 '(': 'regular', | |
| 119 'w(': 'array', | |
| 120 'p(': 'printf', | |
| 121 'm(': 'regex-match', | |
| 122 's(': 'regex-sub', | |
| 123 'f(': 'regex-find', | |
| 124 'v(': 'regex-split', | |
| 125 }[i] | |
| 126 strings = [] | |
| 127 elif i.lower() in mnemonics: | |
| 128 output_code.append(mnemonics[i.lower()]) | |
| 129 else: | |
| 130 raise Exception('unknown symbol: ' + i) | |
| 131 # shortcut: strip leading \x04 | |
| 132 return ''.join(output_code).lstrip('\x04') | |
| 133 | |
| 134 if __name__ == '__main__': | |
| 135 sys.stdout.write(compile_gs2(sys.stdin.read())) |