Parsing attributes with regex in Perl

Question

Here's a problem I ran into recently. I have attributes strings of the form

"x=1 and y=abc and z=c4g and ..."

Some attributes have numeric values, some have alpha values, some have mixed, some have dates, etc.

Every string is supposed to have "x=someval and y=anotherval" at the beginning, but some don't. I have three things I need to do.

1. Validate the strings to be certain that they have x and y.
2. Actually parse the values for x and y.
3. Get the rest of the string.

Given the example at the top, this would result in the following variables:

$x = 1;
$y = "abc";
$remainder = "z=c4g and ..."

My question is: Is there a (reasonably) simple way to parse these and validate with a single regular expression? i.e.:

if ($str =~ /someexpression/)
{
    $x = $1;
    $y = $2;
    $remainder = $3;
}

Note that the string may consist of only x and y attributes. This is a valid string.

I'll post my solution as an answer, but it doesn't meet my single-regex preference.

Answer 1

Here's basically what I did to solve this:

($x_str, $y_str, $remainder) = split(/ and /, $str, 3);

if ($x_str !~ /x=(.*)/)
{
    # error
}

$x = $1;

if ($y_str !~ /y=(.*)/)
{
    # error
}

$y = $1;

I've omitted some additional validation and error handling. This technique works, but it's not as concise or pretty as I would have liked. I'm hoping someone will have a better suggestion for me.

Answer 2

I'm not the best at regular expressions, but this seems pretty close to what you're looking for:

/x=(.+) and y=([^ ]+)( and (.*))?/

Except you use $1, $2, and $4. In use:

my @strs = ("x=1 and y=abc and z=c4g and w=v4l",
            "x=yes and y=no",
            "z=nox and w=noy");

foreach (@strs) {
    if ($_ =~ /x=(.+) and y=([^ ]+)( and (.*))?/) {
        $x = $1;
        $y = $2;
        $remainder = $4;
        print "x: $x; y: $y; remainder: $remainder\n";
    } else {
        print "Failed.\n";
    }
}

Output:

x: 1; y: abc; remainder: z=c4g and w=v4l
x: yes; y: no; remainder: 
Failed.

This of course leaves out plenty of error checking, and I don't know everything about your inputs, but this seems to work.

Answer 3

@Rudd, nice! That's definitely more concise than what I had.

Answer 4

As a fairly simple modification to Rudd's version,

/^x=(.+) and y=([^ ]+)(?: and (.*))?/

will allow you to use $1, $2 and $3 (the ?: makes it a noncapturing group), and will ensure that the string starts with "x=" rather than allowing a "not_x=" to match

If you have better knowledge of what the x and y values will be, this should be used to tighten the regex further:

my @strs = ("x=1 and y=abc and z=c4g and w=v4l",
        "x=yes and y=no",
        "z=nox and w=noy",
        "not-x=nox and y=present",
        "x=yes and w='there is no and y=something arg here'");

foreach (@strs) {
    if ($_ =~ /^x=(.+) and y=([^ ]+)(?: and (.*))?/) {
        $x = $1;
        $y = $2;
        $remainder = $3;
        print "x: {$x}; y: {$y}; remainder: {$remainder}\n";
    } else {
        print "$_ Failed.\n";
    }
}

Output:

x: {1}; y: {abc}; remainder: {z=c4g and w=v4l}
x: {yes}; y: {no}; remainder: {}
z=nox and w=noy Failed.
not-x=nox and y=present Failed.
x: {yes and w='there is no}; y: {something}; remainder: {}

Note that the missing part of the last test is due to the current version of the y test requiring no spaces, if the x test had the same restriction that string would have failed.

Answer 5

Assuming you also want to do something with the other name=value pairs this is how I would do it ( using Perl version 5.10 ):

use 5.10.0;
use strict;
use warnings;

my %hash;
while(
    $string =~ m{
       (?: ^ | \G )    # start of string or previous match
       \s*

       (?<key>   \w+ ) # word characters
       =
       (?<value> \S+ ) # non spaces

       \s*             # get to the start of the next match
       (?: and )?
    }xgi
){
    $hash{$+{key}} = $+{value};
}

# to make sure that x & y exist
die unless exists $hash{x} and exists $hash{y};

On older Perls ( at least Perl 5.6 );

use strict;
use warnings;

my %hash;
while(
    $string =~ m{
       (?: ^ | \G )   # start of string or previous match
       \s*

       ( \w+ ) = ( \S+ )

       \s*            # get to the start of the next match
       (?: and )?
    }xgi
){
    $hash{$1} = $2;
}

# to make sure that x & y exist
die unless exists $hash{x} and exists $hash{y};

These have the added benefit of continuing to work if you need to work with more data.

Answer 6

Rudd and Cebjyre have gotten you most of the way there but they both have certain problems:

Rudd suggested:

/x=(.+) and y=([^ ]+)( and (.*))?/

Cebjyre modified it to:

/^x=(.+) and y=([^ ]+)(?: and (.*))?/

The second version is better because it will not confuse "not_x=foo" with "x=foo" but will accept things such as "x=foo z=bar y=baz" and set $1 = "foo z=bar" which is undesirable.

This is probably what you are looking for:

/^x=(\w+) and y=(\w+)(?: and (.*))?/

This disallows anything between the x= and y= options, places and allows and optional " and..." which will be in $3