Do you mean something like this?

my $string = '...blah...<file fiop="foo" length="bar"/>baz</file>...bl
+ah...';

my ( $foo, $bar, $baz );
if ( $string =~ s{<file fiop="([^"]+)" length="([^"]+)"/>([^<]+)</file
+>}{} ) {
    ( $foo, $bar, $baz ) = ( $1, $2, $3 );
    print "extracted $foo, $bar, $baz; left $string\n";
}
[download]

Actually you hit it right on the spot; it's home-grown XML from Cupertino...
The baz part is raw binary data, inserted in the XML; that's why I want to extract it before parsing the valid XML.
I hadn't even noticed the close-angle-bracket (thanks - but it's really there).

I've tried your code; but it doesn't seem to get me there.
Any further suggestions ?

When I run my snippet as posted, I get the following output:

extracted foo, bar, baz; left ...blah......blah...
[download]

Do you get something different when you run it? Or do you want something different from that?

When you try to use the "s{...}{}" expression in your own code, is it possible that your "raw binary data" (in "the baz part") might contain a byte value of 0x3C? This would be treated as a "<" character in the regex match, which would cause trouble. Something like this might work better in that case:

s{<file fiop="([^"]+)" length="([^"]+)"/>(.*?)</file>}{}s
[download]

(update: added the "s" modifier at the end, in case the raw binary stuff might contain a line-feed)

Note the question mark after ".*" -- that's the important thing that was missing from your initial attempt: it makes the wildcard match non-greedy (stops matching as soon as possible).

I find it funny that the right solution ("use a parser") is shot down because this isn't exactly XML. Are you sure an HTML parser wouldn't parse it properly? Weird how everyone gets stuck on regex.

duff

Nobody answered after 16 min? Oh, graff did (and was faster than me) ;-)

 ...
 my ($fiop, $length, $data)
    = $string =~ m{<file                 # tag anchor
                   \s+ fiop="([^"]+)"    # (foo)
                   \s+ length="([^"]+)"  # (bar)
                   />                    # end: start file tag
                   \s* (.*?)             # (baz) - note the "nongreedy
+ness" .*?
                   </file>               # end: file tag
                  }x;

 print "$fiop, $length, $data\n";
 ...
[download]

Addendum: forgot the tag-cleaning part:

 
  ...
 (my $notags = $string) =~ s{<file.+?</file>}{};
  print "$notags\n";
  ...
[download]

Your mistake was basically to take the "greedy modifier" (.*), which matches until the end of the string - and backtracks then - and matches from the rear ...

Regards

mwa

I'm not sure if I should rephrase...
The data I get is from a POST request; it's almost XML except for the <file/> part that I need take out, as parser.pm barfs on it..
in "<file fiop="foo" length="bar"/>baz</file>", baz is raw binary;
that's why (I think) the preg matching (m^) doesn't work.
FYI baz looks like: µÜ¡3õ§©AEurope/Amsterdam$...

Ohh, *if* there is some binary within the tag and *if* the "length" field says sth. about its *length* you could easily construct a regex that extracts binary data of that length:

 my $binary = pack 'F*', (3.141592) x 10;  # make binary vector of len
+gth 80 bytes
 my $string = '...blah...<file fiop="foo" length="' . length($binary) 
+.'"/>' . $binary . '</file>...blah...';

 my ($fiop, $length, $data)
    = $string =~ m{<file                   # tag anchor
                   \s+ fiop="([^"]+)"      # (fiop)
                   \s+ length="([^"]+)"    # (length)
                   />                      # end: start file tag
                   ((??{ "\\C{$2}" }))     # self modifying regex for 
+binary stuff
                   </file>                 # end: file tag
                  }sx;

 print "$fiop, $length (data comes below)\n";
 print join ',', unpack("F*", $data);      # extract binary data again
 
 (my $notags = $string) =~ s{<file.+</file>}{};
 print "\n$notags\n";
[download]

In the above I pack a binary sequence of 10 Pi-Numbers (double, 10 x 8 bytes) into the tag, match a binary sequence of its length ($2) and unpack it afterwards.

Regards

mwa

Using the fubarred xml from your question in the CB:

use warnings;
use strict;

my $string = q {...blah...<file fiop="foo" length="304507"/> SyncBookm
+arkRecordType12345·bjbjUU</file>...blah};

my $foo = qr {<file fiop="([a-z]+)" length="(\d+)"/> (.+)(?=</file>)};
 
if ($string =~ m#$foo#) {
   print "\nMatched \"fiop\" label: |$1|\nMatched Length: |$2|\nMatche
+d data: |$3|\n";
} else {
   print "no match \n";
}

print "\nDone \n\n";
[download]

OUTPUT
Matched "fiop" label: |foo|
Matched Length: |304507|
Matched data: |SyncBookmarkRecordType12345?+bjbjU?U?|

Done

Thanks ww for your solution;
as mwah pointed out in Re^5: extracting a substring from a string - multiple variables it'd be safer to read in the binary data with the given length.
How would I do that here ?