Help with regular expressions [2]

displace · 09-07-2015, 08:05 AM

Hey, I'm just gonna be brief about it.

Are there any ways of using variables inside a regular expression? For example when matching a particular text I'd like to construct a regular expression in a dynamic way. In particular a part of the regular expression should change, but I do not know in advance what the change will be. All I have is a string variable that should become a part of the regex.

Take it I have a regular expression like "123someREGEX", but I want the word "some" to be variable. So is there a way I can 'inject' a variable inside the regex so that I can instead write the regex as "123\1REGEX". Of course this wouldn't work because "\1" is a backreference to a capturing group that doesn't exist. But perhaps there is a way to programatically create a number of named groups before the match takes place and then use those named groups as variables in the regex?

I deliberately haven't specified a language because I'd like to know, if any languages support this.

HMW · 09-07-2015, 09:53 AM

Yeah, it's possible. Check out this example with sed:

Code:

$ foo="a word"

Code:

$ echo "foobar" | sed "s/foo/this is $foo /"
this is a word bar

Note the double quotation marks in sed, otherwise the use of variables will fail.

Edit:
The same applies for grep:

Code:

$ bar="string"

$ echo "A long string with basically nothing in it" | grep -o "long $bar"
long string

Best regards,
HMW

displace · 09-07-2015, 12:12 PM

It's probably also possible to do with awk scripting language, but what about higher languages like c++ (boost::regex) and python?

norobro · 09-07-2015, 01:27 PM

Here's a simple c++ example:

Code:

#include <iostream>
#include <string>
#include <regex>

int main (int argc, char *argv[]) 
{
    std::string pattern { argv[1] };
    static const std::regex re("(" + pattern + ")") ;
    std::string line  { "Now is the time for all good men to come to the aid of their party" };
    std::smatch match ;
     while(std::regex_search( line, match, re ) ) {
        std::cout << match[1] << std::endl;
        line = match.suffix().str();
    }
}

Code:

$ clang++ -std=c++11 main.cpp && ./a.out the
the
the
the

Code:

$./a.out "\bthe\b"
the
the

Rinndalir · 09-07-2015, 07:18 PM

In python you could do

Code:

import re
for line in text:
    rx = '(%s%s%s)' % ('123',dynamicword,'REGEX')
    dynrx = re.compile(rx,re.I)
    for dynmatch in dynrx.findall(line):
        print dynmatch
        dynamicword = dynmatch

displace · 09-09-2015, 01:28 AM

So basically I have to manually assemble the regex string programatically. There is no way to do it within the regex library itself (i.e. named groups)?

Rinndalir · 09-09-2015, 01:42 AM

Quote:

Originally Posted by displace

So basically I have to manually assemble the regex string programatically. There is no way to do it within the regex library itself (i.e. named groups)?

Manually and programatically are not the same. Write some pseudo-code with pseudo-data and it would be easier to provide an answer.
As far as I know you can do almost anything with regex and some code. Limited only by abilities and creativity.

chrism01 · 09-10-2015, 03:48 AM

You can do that in Perl, but I agree that providing a couple of examples of what you are trying to do would be helpful.
http://perldoc.perl.org/perlretut.html