blob: 2c77ee91db5cf1addcc095efb3e792bede7a1266 [file] [log] [blame]
Andrew M. Kuchlingd9b38d22004-04-06 19:42:34 +00001#! /usr/bin/env python
Guido van Rossumf81e5b91997-10-23 22:43:50 +00002
Guido van Rossum7f91cf92000-05-30 13:25:35 +00003r"""Convert old ("regex") regular expressions to new syntax ("re").
Guido van Rossumf81e5b91997-10-23 22:43:50 +00004
5When imported as a module, there are two functions, with their own
6strings:
7
8 convert(s, syntax=None) -- convert a regex regular expression to re syntax
9
10 quote(s) -- return a quoted string literal
11
12When used as a script, read a Python string literal (or any other
13expression evaluating to a string) from stdin, and write the
14translated expression to stdout as a string literal. Unless stdout is
15a tty, no trailing \n is written to stdout. This is done so that it
16can be used with Emacs C-U M-| (shell-command-on-region with argument
17which filters the region through the shell command).
18
19No attempt has been made at coding for performance.
20
21Translation table...
22
23 \( ( (unless RE_NO_BK_PARENS set)
24 \) ) (unless RE_NO_BK_PARENS set)
25 \| | (unless RE_NO_BK_VBAR set)
26 \< \b (not quite the same, but alla...)
27 \> \b (not quite the same, but alla...)
28 \` \A
29 \' \Z
30
31Not translated...
32
33 .
34 ^
35 $
36 *
37 + (unless RE_BK_PLUS_QM set, then to \+)
38 ? (unless RE_BK_PLUS_QM set, then to \?)
39 \
40 \b
41 \B
42 \w
43 \W
44 \1 ... \9
45
46Special cases...
47
48 Non-printable characters are always replaced by their 3-digit
49 escape code (except \t, \n, \r, which use mnemonic escapes)
50
51 Newline is turned into | when RE_NEWLINE_OR is set
52
53XXX To be done...
54
55 [...] (different treatment of backslashed items?)
56 [^...] (different treatment of backslashed items?)
57 ^ $ * + ? (in some error contexts these are probably treated differently)
58 \vDD \DD (in the regex docs but only works when RE_ANSI_HEX set)
59
60"""
61
62
Guido van Rossum81fc7782001-09-04 15:18:54 +000063import warnings
64warnings.filterwarnings("ignore", ".* regex .*", DeprecationWarning, __name__,
65 append=1)
66
Guido van Rossumf81e5b91997-10-23 22:43:50 +000067import regex
Tim Peters0c9886d2001-01-15 01:18:21 +000068from regex_syntax import * # RE_*
Guido van Rossumf81e5b91997-10-23 22:43:50 +000069
Skip Montanaro0de65802001-02-15 22:15:14 +000070__all__ = ["convert","quote"]
71
Guido van Rossumf81e5b91997-10-23 22:43:50 +000072# Default translation table
73mastertable = {
74 r'\<': r'\b',
75 r'\>': r'\b',
76 r'\`': r'\A',
77 r'\'': r'\Z',
78 r'\(': '(',
79 r'\)': ')',
80 r'\|': '|',
81 '(': r'\(',
82 ')': r'\)',
83 '|': r'\|',
84 '\t': r'\t',
85 '\n': r'\n',
86 '\r': r'\r',
87}
88
89
90def convert(s, syntax=None):
91 """Convert a regex regular expression to re syntax.
92
93 The first argument is the regular expression, as a string object,
94 just like it would be passed to regex.compile(). (I.e., pass the
95 actual string object -- string quotes must already have been
96 removed and the standard escape processing has already been done,
97 e.g. by eval().)
98
99 The optional second argument is the regex syntax variant to be
100 used. This is an integer mask as passed to regex.set_syntax();
101 the flag bits are defined in regex_syntax. When not specified, or
102 when None is given, the current regex syntax mask (as retrieved by
103 regex.get_syntax()) is used -- which is 0 by default.
104
105 The return value is a regular expression, as a string object that
106 could be passed to re.compile(). (I.e., no string quotes have
107 been added -- use quote() below, or repr().)
108
109 The conversion is not always guaranteed to be correct. More
110 syntactical analysis should be performed to detect borderline
111 cases and decide what to do with them. For example, 'x*?' is not
112 translated correctly.
113
114 """
115 table = mastertable.copy()
116 if syntax is None:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000117 syntax = regex.get_syntax()
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000118 if syntax & RE_NO_BK_PARENS:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000119 del table[r'\('], table[r'\)']
120 del table['('], table[')']
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000121 if syntax & RE_NO_BK_VBAR:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000122 del table[r'\|']
123 del table['|']
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000124 if syntax & RE_BK_PLUS_QM:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000125 table['+'] = r'\+'
126 table['?'] = r'\?'
127 table[r'\+'] = '+'
128 table[r'\?'] = '?'
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000129 if syntax & RE_NEWLINE_OR:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000130 table['\n'] = '|'
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000131 res = ""
132
133 i = 0
134 end = len(s)
135 while i < end:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000136 c = s[i]
137 i = i+1
138 if c == '\\':
139 c = s[i]
140 i = i+1
141 key = '\\' + c
142 key = table.get(key, key)
143 res = res + key
144 else:
145 c = table.get(c, c)
146 res = res + c
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000147 return res
148
149
150def quote(s, quote=None):
151 """Convert a string object to a quoted string literal.
152
153 This is similar to repr() but will return a "raw" string (r'...'
154 or r"...") when the string contains backslashes, instead of
155 doubling all backslashes. The resulting string does *not* always
156 evaluate to the same string as the original; however it will do
157 just the right thing when passed into re.compile().
158
159 The optional second argument forces the string quote; it must be
160 a single character which is a valid Python string quote.
161
162 """
163 if quote is None:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000164 q = "'"
165 altq = "'"
166 if q in s and altq not in s:
167 q = altq
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000168 else:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000169 assert quote in ('"', "'")
170 q = quote
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000171 res = q
172 for c in s:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000173 if c == q: c = '\\' + c
174 elif c < ' ' or c > '~': c = "\\%03o" % ord(c)
175 res = res + c
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000176 res = res + q
177 if '\\' in res:
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000178 res = 'r' + res
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000179 return res
180
181
182def main():
183 """Main program -- called when run as a script."""
184 import sys
185 s = eval(sys.stdin.read())
186 sys.stdout.write(quote(convert(s)))
187 if sys.stdout.isatty():
Guido van Rossum45e2fbc1998-03-26 21:13:24 +0000188 sys.stdout.write("\n")
Guido van Rossumf81e5b91997-10-23 22:43:50 +0000189
190
191if __name__ == '__main__':
192 main()