src/etc/2014-06-rewrite-bytes-macros.py

   1 #!/usr/bin/env python
   2 #
   3 # Copyright 2014 The Rust Project Developers. See the COPYRIGHT
   4 # file at the top-level directory of this distribution and at
   5 # http://rust-lang.org/COPYRIGHT.
   6 #
   7 # Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
   8 # http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
   9 # <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
  10 # option. This file may not be copied, modified, or distributed
  11 # except according to those terms.
  12
  13 import sys
  14 import subprocess
  15 import re
  16
  17
  18 def main():
  19     if len(sys.argv) <= 1:
  20         print('Usage: %s [ --apply ] filename1.rs filename2.rs ...'
  21               % sys.argv[0])
  22     elif sys.argv[1] == '--apply':
  23         for filename in sys.argv[2:]:
  24             patch(filename)
  25     else:
  26         for filename in sys.argv[1:]:
  27             diff(filename)
  28
  29
  30 def patch(filename):
  31     source = read(filename)
  32     rewritten = rewrite_bytes_macros(source)
  33     if rewritten is not None and rewritten != source:
  34         write(filename, rewritten)
  35
  36
  37 def diff(filename):
  38     rewritten = rewrite_bytes_macros(read(filename))
  39     if rewritten is not None:
  40         p = subprocess.Popen(['diff', '-u', filename, '-'],
  41                              stdin=subprocess.PIPE)
  42         p.stdin.write(rewritten)
  43         p.stdin.close()
  44         p.wait()
  45
  46
  47 def read(filename):
  48     with open(filename, 'rb') as f:
  49         return f.read()
  50
  51
  52 def write(filename, content):
  53     with open(filename, 'wb') as f:
  54         f.write(content)
  55
  56
  57 def rewrite_bytes_macros(source):
  58     rewritten, num_occurrences = BYTES_MACRO_RE.subn(rewrite_one_macro, source)
  59     if num_occurrences > 0:
  60         return rewritten
  61
  62
  63 BYTES_MACRO_RE = re.compile(br'bytes!\(  (?P<args>  [^)]*  )  \)', re.VERBOSE)
  64
  65
  66 def rewrite_one_macro(match):
  67     try:
  68         bytes = parse_bytes(split_args(match.group('args')))
  69         return b'b"' + b''.join(map(escape, bytes)) + b'"'
  70     except SkipThisRewrite:
  71         print('Skipped: %s' % match.group(0).decode('utf8', 'replace'))
  72         return match.group(0)
  73
  74
  75 class SkipThisRewrite(Exception):
  76     pass
  77
  78
  79 def split_args(args):
  80     previous = b''
  81     for arg in args.split(b','):
  82         if previous:
  83             arg = previous + b',' + arg
  84         if arg.count(b'"') % 2 == 0:
  85             yield arg
  86             previous = b''
  87         else:
  88             previous = arg
  89     if previous:
  90         yield previous
  91
  92
  93 def parse_bytes(args):
  94     for arg in args:
  95         arg = arg.strip()
  96         if (arg.startswith(b'"') and arg.endswith(b'"')) or (
  97                 arg.startswith(b"'") and arg.endswith(b"'")):
  98             # Escaped newline means something different in Rust and Python.
  99             if b'\\\n' in arg:
 100                 raise SkipThisRewrite
 101             for byte in eval(b'u' + arg).encode('utf8'):
 102                 yield ord(byte)
 103         else:
 104             if arg.endswith(b'u8'):
 105                 arg = arg[:-2]
 106             # Assume that all Rust integer literals
 107             # are valid Python integer literals
 108             value = int(eval(arg))
 109             assert value <= 0xFF
 110             yield value
 111
 112
 113 def escape(byte):
 114     c = chr(byte)
 115     escaped = {
 116         b'\0': br'\0',
 117         b'\t': br'\t',
 118         b'\n': br'\n',
 119         b'\r': br'\r',
 120         b'\'': b'\\\'',
 121         b'\\': br'\\',
 122     }.get(c)
 123     if escaped is not None:
 124         return escaped
 125     elif b' ' <= c <= b'~':
 126         return chr(byte)
 127     else:
 128         return ('\\x%02X' % byte).encode('ascii')
 129
 130
 131 if str is not bytes:
 132     # Python 3.x
 133     ord = lambda x: x
 134     chr = lambda x: bytes([x])
 135
 136
 137 if __name__ == '__main__':
 138     main()