Perl-compatible regular expression optimizer

Optimizes perl-compatible regular expressions.

The general syntax for running the program is: regex-opt <regexp>

Example:
regex-opt 'xaz|xbz|xcz'
x[a-c]z

Try running regex-opt on Abigail's 7 kilobyte URL regexp. The result should be about 5 kilobytes long.

• * (repeat 0-inf)
• + (repeat 1-inf)
• ? (repeat 0-1)
• {n} (repeat n)
• {n,} (repeat n-inf)
• {n,m} (repeat n-m)
• . (accept any char except \n)
• [a-z] (character sets)
• [^a-z] (inverse character sets)
• [[:alpha:]] (character classes)
• \s (and other character classes and escapes)
• x|y (alternatives)
• (?:x|y) (non-capturing grouping)
• *? (non-greedy repeat)

• ^ (match string-begin)
• $ (match string-end)
• () (capturing is converted to noncapturing)
• Any (? -command that is not mentioned in supported syntax
• Unicode-specific markup

• Character set optimization: [A-Zabcdefgh-yz0-9%] becomes [[:alnum:]%]
• Alternate characters: y|[yp]|[zx] becomes [px-z]
• Counting: aaa* and aa+ become a{2,}
• Combining: abcde|xycde becomes (?:ab|xy)cde
• Parenthesis reduction: ((abc)) becomes abc, (xx|yy)|zz becomes xx|yy|zz
• Compression: xyzyzxyzyz becomes (?:x(?:yz){2}){2}
  • This might not be always a good thing.
• Choice counting: a+|aa+ becomes a+, (b|) becomes b?, dxxxxb|dxxxb|dxxb|dxb becomes dx{1,4}b

• Combining counts:
  • a?|b? should become (?:a|b)?, now becomes a?|b?
• Redundancy removal (removal of alternatives that are subsets of other alternatives):
  • xfooy|x[a-q]+y should become x[a-q]+y, now becomes x(?:foo|[a-q]+)y

Help in solving these shortcomings would be welcome.

regex-opt has been written by Joel Yliluoma, a.k.a. Bisqwit,
and is distributed under the terms of the General Public License (GPL).

For compiling you need the following GNU tools: g++, make.

The official home page of regex-opt is at http://iki.fi/bisqwit/source/regex-opt.html.
Check there for new versions.