Parsing HTML-attributes like strings - perl

Question

Have strings like HTML attributes
key1="value1 value2" key2="va3" key4
need parse such string to get HoA:
$parsed = {
key1' => [
'value1',
'value2'
],
key2 => [ 'val3' ], #or key2 => 'val3' doesn't matter..
key4 => undef,
};
Creating the function myself, like:
#!/usr/bin/env perl
use 5.014;
use strict;
use warnings;
use Data::Dumper;
while(<DATA>) {
my $parsed;
chomp;
next if m/\A\s*#/;
while( m/(\w+)(\s*=\s*(["'])(.*?)(\3))?/g ) {
my $k = $1;
if( $4 ) {
my #v = split(/\s+/, $4);
$parsed->{$k} = \#v;
}
else {
$parsed->{$k} = undef;
}
}
say Dumper $parsed;
}
__DATA__
key1="value1 value2" key2 key3="val3"
key1='value1 "value2"' key8 key3='val3'
key1='value1 i\'m' key2 key3="val3"
key1='value1 value2' key8 key3=val3
works and prints correct results for the first 2 lines.
$VAR1 = {
'key1' => [
'value1',
'value2'
],
'key3' => [
'val3'
],
'key2' => undef
};
$VAR1 = {
'key1' => [
'value1',
'"value2"'
],
'key3' => [
'val3'
],
'key8' => undef
};
Unfortunately, it fails on 3rd line - don't know how to handle the escaped quotes. (And just figured out than the key=val (without quotes) is valid too))
Additionally, because don't want reinvent the wheel again, probably exists some module on CPAN for this, only haven't any idea what to search. ;(
EDIT
#mpapec suggested a module, what could greatly help for parsing the RHS part of the "assignement". My problem is than the string contains multiple space delimited LHS=RHS, where the RHS could be quoted (in single and double) or not quoted (in the case of one value) and the RHS values (in the quotes) are space delimited too..
key1="value1 value2" key2="va3" key4 key5=val5 key6='val6' key7='val x\'y zzz'
So I don't know how to break the string into multiple LHS=RHS parts, because can't split at space and can't use my regex, because it fails in escaped quotes. (maybe some more complicated regex what handles escapes could work).
Any suggestions, please?

Answer 1

You can use Text::ParseWords as mpapec suggested:
use strict;
use warnings;
use 5.010;
use Data::Dumper;
use Text::ParseWords;
$Data::Dumper::Sortkeys = 1;
my $string = q{key1="value1 value2" key2="va3" key4 key5=val5 key6='val6' key7='val x\'y zzz'};
my #words = shellwords $string;
my %parsed;
foreach my $word (#words) {
my ($key, $values) = split /=/, $word, 2;
$parsed{$key} //= [];
push #{ $parsed{$key} }, $_ for shellwords $values;
}
print Dumper \%parsed;
Output:
$VAR1 = {
'key1' => [
'value1',
'value2'
],
'key2' => [
'va3'
],
'key4' => [],
'key5' => [
'val5'
],
'key6' => [
'val6'
],
'key7' => [
'val',
'x\'y',
'zzz'
]
};
Note that for consistency, I assigned keys without values an empty array instead of undef. I think this will make the data structure easier to use.
Also note that I called shellwords twice. I did this to remove the backslashes from escaped quotes, so that
key7='val x\'y zzz'
gets split into
val x'y zzz
instead of
val x\'y zzz
(The backslash in x\'y in the output above is added by Data::Dumper; there's no backslash in the variable itself.)

Answer 2

To fix your current issue, you can setup an alteration to handle backslashes in a special way.
#!/usr/bin/env perl
use 5.014;
use strict;
use warnings;
use Data::Dump;
my $parsed;
while (<DATA>) {
chomp;
next if m/\A\s*#/;
while (
m{
(\w+)
(?:
\s* = \s*
(["'])
( (?: (?!\2)[^\\] | \\. )* )
\2
)?
}gx
)
{
my $k = $1;
if ($2) {
( my $val = $3 ) =~ s/\\(.)/$1/g; # Unescape backslashes
$parsed->{$k} = [ split /\s+/, $val ]; # Split words
} else {
$parsed->{$k} = undef;
}
}
dd $parsed;
print "\n";
}
__DATA__
key1="value1 value2" key2 key3="val3"
key1='value1 "value2"' key2 key3='val3'
key1='value1 i\'m' key2 key3="val3"
Outputs:
{ key1 => ["value1", "value2"], key2 => undef, key3 => ["val3"] }
{ key1 => ["value1", "\"value2\""], key2 => undef, key3 => ["val3"] }
{ key1 => ["value1", "i'm"], key2 => undef, key3 => ["val3"] }
There are still other issues to take into account, but perhaps this will help you get further along.

Answer 3

You might consider something based on Parser::MGC. Your examples look like a nice simple loop on
my $key = $self->token_ident;
$self->expect( '=' );
my $value = $self->token_string;