ECMAScript Regex for a multilined string

Question

I am writing the loading procedure for my application and it involves reading data from a file and creating an appropriate object with appropriate properties.

The file consists of sequential entries (separated by a newline) in the following format:

=== OBJECT TYPE ===
<Property 1>: Value1
<Property 2>: Value2
=== END OBJECT TYPE ===

Where the values are often strings which may consist of arbitrary characters, new-lines, etc.

I want to create a std::regex which can match this format and allow me to use std::regex_iterator to read each of the objects into the file in turn.

However, I am having trouble creating a regex which matches this type of format; I have looked at the ECMAScript syntax and create my regex in the following way, but it does not match the string in my test application:

const std::regex regexTest( "=== ([^=]+) ===\\n([.\\n]*)\\n=== END \\1 ===" );

And when using this in the following test application, it fails to match the regex to the string:

int main()
{
    std::string testString = "=== TEST ===\n<Random Example>:This is a =test=\n<Another Example>:Another Test||\n=== END TEST ===";

    std::cout << testString << std::endl;

    const std::regex regexTest( "=== ([^=]+) ===\\n([.\\n]*)\\n=== END \\1 ===" );
    std::smatch regexMatch;

    if( std::regex_match( testString, regexMatch, regexTest ) )
    {
        std::cout << "Prefix: \"" << regexMatch[1] << "\"" << std::endl;
        std::cout << "Main Body: \"" << regexMatch[2] << "\"" << std::endl;
    }

    return 0;
}

What compiler are you using? Are you aware of that `str::regex` is not implemented (fully/at all) in some compilers? `g++` specifically (last I checked). — Qtax, Jun 18 '13 at 20:58
@Qtax I am using Microsoft Visual Studio 2012 and the standard library implementation which ships with that, which as far as I'm aware, provides a full implementation of `std::regex` and associated functions. — Thomas Russell, Jun 18 '13 at 20:59
@Shaktal can you try removing parts of the expression (to get parts of the match) to see which part breaks the pattern? — Martin Ender, Jun 19 '13 at 01:25

score 1 · Accepted Answer · answered Jun 18 '13 at 23:31

Your problem is quite simpler than it looks. This:

const std::regex regexTest( "=== ([^=]+) ===\\n((?:.|\\n)*)\\n=== END \\1 ===" );

worked perfectly on clang++/libc++. It seems that \n does not fit into [] brackets in ECMAscript regexen. Remember to use while regex_search instead of if regex_match if you want to look for more than one instance of the regex inside the string!

Casimir et Hippolyte · Answer 2 · 2013-06-16T14:18:47.033

0

Try to use:

lazy quantifiers:

=== (.+?) ===\\n([\\s\\S]*?)\\n=== END \\1 ===
negative classes and negative lookaheads:

=== ((?:[^ ]+| (?!===))+) ===\\n((?:[^\\n]+|\\n(?!=== END \\1 ===))*)
POSIX:

=== (.+?) ===\n((.|\n)*?)\n=== END [^=]+? ===

edited Jun 16 '13 at 14:18

answered Jun 16 '13 at 12:35

Casimir et Hippolyte

88,009
5
94
125

Neither of these regular expressions matched the string, though both compiled without throwing an exception. Thanks for the help!! – Thomas Russell Jun 16 '13 at 12:49
I have tried with the edit and the first one fails to match, and the second one results in a run-time error being generated. – Thomas Russell Jun 16 '13 at 13:40
I have tried your POSIX implementation and no run time error was produced and the regex compiled successfully, but it didn't match my test string. Thank you for all your help so far! – Thomas Russell Jun 17 '13 at 17:34
@Shaktal: have you made a test with a silly example like `a` to match `a`? – Casimir et Hippolyte Jun 17 '13 at 17:36
Yeah, if I have say the string `"abbc"` and the regex `"a([b]+)c"` it matches perfectly. – Thomas Russell Jun 17 '13 at 17:40

ECMAScript Regex for a multilined string

2 Answers2

Linked