0001 #! /usr/bin/env python 0002 0003 r"""Convert old ("regex") regular expressions to new syntax ("re"). 0004 0005 When imported as a module, there are two functions, with their own 0006 strings: 0007 0008 convert(s, syntax=None) -- convert a regex regular expression to re syntax 0009 0010 quote(s) -- return a quoted string literal 0011 0012 When used as a script, read a Python string literal (or any other 0013 expression evaluating to a string) from stdin, and write the 0014 translated expression to stdout as a string literal. Unless stdout is 0015 a tty, no trailing \n is written to stdout. This is done so that it 0016 can be used with Emacs C-U M-| (shell-command-on-region with argument 0017 which filters the region through the shell command). 0018 0019 No attempt has been made at coding for performance. 0020 0021 Translation table... 0022 0023 \( ( (unless RE_NO_BK_PARENS set) 0024 \) ) (unless RE_NO_BK_PARENS set) 0025 \| | (unless RE_NO_BK_VBAR set) 0026 \< \b (not quite the same, but alla...) 0027 \> \b (not quite the same, but alla...) 0028 \` \A 0029 \' \Z 0030 0031 Not translated... 0032 0033 . 0034 ^ 0035 $ 0036 * 0037 + (unless RE_BK_PLUS_QM set, then to \+) 0038 ? (unless RE_BK_PLUS_QM set, then to \?) 0039 \ 0040 \b 0041 \B 0042 \w 0043 \W 0044 \1 ... \9 0045 0046 Special cases... 0047 0048 Non-printable characters are always replaced by their 3-digit 0049 escape code (except \t, \n, \r, which use mnemonic escapes) 0050 0051 Newline is turned into | when RE_NEWLINE_OR is set 0052 0053 XXX To be done... 0054 0055 [...] (different treatment of backslashed items?) 0056 [^...] (different treatment of backslashed items?) 0057 ^ $ * + ? (in some error contexts these are probably treated differently) 0058 \vDD \DD (in the regex docs but only works when RE_ANSI_HEX set) 0059 0060 """ 0061 0062 0063 import warnings 0064 warnings.filterwarnings("ignore", ".* regex .*", DeprecationWarning, __name__, 0065 append=1) 0066 0067 import regex 0068 from regex_syntax import * # RE_* 0069 0070 __all__ = ["convert","quote"] 0071 0072 # Default translation table 0073 mastertable = { 0074 r'\<': r'\b', 0075 r'\>': r'\b', 0076 r'\`': r'\A', 0077 r'\'': r'\Z', 0078 r'\(': '(', 0079 r'\)': ')', 0080 r'\|': '|', 0081 '(': r'\(', 0082 ')': r'\)', 0083 '|': r'\|', 0084 '\t': r'\t', 0085 '\n': r'\n', 0086 '\r': r'\r', 0087 } 0088 0089 0090 def convert(s, syntax=None): 0091 """Convert a regex regular expression to re syntax. 0092 0093 The first argument is the regular expression, as a string object, 0094 just like it would be passed to regex.compile(). (I.e., pass the 0095 actual string object -- string quotes must already have been 0096 removed and the standard escape processing has already been done, 0097 e.g. by eval().) 0098 0099 The optional second argument is the regex syntax variant to be 0100 used. This is an integer mask as passed to regex.set_syntax(); 0101 the flag bits are defined in regex_syntax. When not specified, or 0102 when None is given, the current regex syntax mask (as retrieved by 0103 regex.get_syntax()) is used -- which is 0 by default. 0104 0105 The return value is a regular expression, as a string object that 0106 could be passed to re.compile(). (I.e., no string quotes have 0107 been added -- use quote() below, or repr().) 0108 0109 The conversion is not always guaranteed to be correct. More 0110 syntactical analysis should be performed to detect borderline 0111 cases and decide what to do with them. For example, 'x*?' is not 0112 translated correctly. 0113 0114 """ 0115 table = mastertable.copy() 0116 if syntax is None: 0117 syntax = regex.get_syntax() 0118 if syntax & RE_NO_BK_PARENS: 0119 del table[r'\('], table[r'\)'] 0120 del table['('], table[')'] 0121 if syntax & RE_NO_BK_VBAR: 0122 del table[r'\|'] 0123 del table['|'] 0124 if syntax & RE_BK_PLUS_QM: 0125 table['+'] = r'\+' 0126 table['?'] = r'\?' 0127 table[r'\+'] = '+' 0128 table[r'\?'] = '?' 0129 if syntax & RE_NEWLINE_OR: 0130 table['\n'] = '|' 0131 res = "" 0132 0133 i = 0 0134 end = len(s) 0135 while i < end: 0136 c = s[i] 0137 i = i+1 0138 if c == '\\': 0139 c = s[i] 0140 i = i+1 0141 key = '\\' + c 0142 key = table.get(key, key) 0143 res = res + key 0144 else: 0145 c = table.get(c, c) 0146 res = res + c 0147 return res 0148 0149 0150 def quote(s, quote=None): 0151 """Convert a string object to a quoted string literal. 0152 0153 This is similar to repr() but will return a "raw" string (r'...' 0154 or r"...") when the string contains backslashes, instead of 0155 doubling all backslashes. The resulting string does *not* always 0156 evaluate to the same string as the original; however it will do 0157 just the right thing when passed into re.compile(). 0158 0159 The optional second argument forces the string quote; it must be 0160 a single character which is a valid Python string quote. 0161 0162 """ 0163 if quote is None: 0164 q = "'" 0165 altq = "'" 0166 if q in s and altq not in s: 0167 q = altq 0168 else: 0169 assert quote in ('"', "'") 0170 q = quote 0171 res = q 0172 for c in s: 0173 if c == q: c = '\\' + c 0174 elif c < ' ' or c > '~': c = "\\%03o" % ord(c) 0175 res = res + c 0176 res = res + q 0177 if '\\' in res: 0178 res = 'r' + res 0179 return res 0180 0181 0182 def main(): 0183 """Main program -- called when run as a script.""" 0184 import sys 0185 s = eval(sys.stdin.read()) 0186 sys.stdout.write(quote(convert(s))) 0187 if sys.stdout.isatty(): 0188 sys.stdout.write("\n") 0189 0190 0191 if __name__ == '__main__': 0192 main() 0193
Generated by PyXR 0.9.4