RegEx Doubt

[mecrazycoder]

Hi Monks, I am having a line like

<xyz>a</xyz><xyz>a3</xyz><xyz>a2</xyz><xyz>a1</xyz>

. Now i want to write a regex to fetch a,a3,a2,a1 alone.How can i do that one. Regex i wrote was something like

$text=`/<xyz>(.*)<\/xyz>/

This doesn't seems to work.Please guide me

[moritz]

Use an XML or HTML parser (like XML::Twig or HTML::TreeBuilder) - parsing HTML with regexes can be very painful and time consuming.

[ccn]

my @a = $text =~ m{<xyz>([^<]*)</xyz>}g;

see more:
• perldoc perlrequick
• perldoc perlretut
• perldoc perlop

[mecrazycoder]

Tanx man

[Bloodnok]

TIMTOWTDI - why not use split ...

$ perl -e '@a = split /<\/?xyz>/, q(<xyz>a</xyz><xyz>a3</xyz><xyz>a2</
+xyz><xyz>a1</xyz>); print qq/@a\n/'
 a  a3  a2  a1

Update:

Ahhh, maybe I see one reason, using Data::Dumper to print the output gives:

$ perl -MData::Dumper -e '@a = split /\<\/?xyz\>/, q(<xyz>a</xyz><xyz>
+a3</xyz><xyz>a2</xyz><xyz>a1</xyz>); print Dumper \@a'
$VAR1 = [
          '',
          'a',
          '',
          'a3',
          '',
          'a2',
          '',
          'a1'
        ];

Question, for me at least is: why doesn't split swallow the sub-strings on which the string is split ? I'm obviously missing something, but can't see it - any enlightenment appreciated.

TIA

[jakobi]

Hm, its non-zero-width, so it's still nice and easy: You've multiple 'split-points' in sequence in the source. Try either grep /./ on the split results or use split /(?:...)+/ to 'combine' them into one 'split-point'.

Aren't they cute, those little regexes? Remembering apocalypse5 fondly :).

[Bloodnok]

...use split /(?:...)+/ to 'combine' them into one 'split-point'. and then grep for empty lines i.e. grep /./, ..., since the first element is still empty, so might as well use grep /./, ... on the lot to start with.

I tried the zero capture approach, but a) transposed the '?' and the ';' and b) didn't use '+' ... doh !!!

Update:

To reduce any confusion, the transposition to which I referred in the above was entirely down to the paucity of my typing i.e. I typed ':?' instead of '?:' and didn't notice .oO(Maybe I ought to use a larger font...) ;-)

[grizzley]

It puzzled me for a minute, but I found explanation. You have two sub-strings every time. Separated by nothing. </xyz>_nothing_<xyz>. And this nothing is what you find in your output.

[dsheroh]

It does swallow the substrings on which the string is split. I don't see any <\/?xyz>s in the Dumper output.

The blank entries being returned are the zero-length substrings in the middle of </xyz><xyz> - that combination is two matches of the split pattern with nothing in between.