Perl Guru Forums: Perl Programming Help: Beginner: regex help

CGI/Perl Guide | Learning Center | Forums | Advertise | Login

Site Search: in


	MAIN INDEX	SEARCH POSTS	WHO'S ONLINE	LOG IN

Home: Perl Programming Help: Beginner:
regex help

robertico
Novice

Mar 22, 2013, 12:41 AM

Post #1 of 15 (1126 views)

regex help	Can't Post

I use this regex to extract all lines starting with ' { ' and ending with ' } '

Code

push (@matches,$&) while($line =~ /{(.*?)}/g );

Now I'd like to exclude lines that contains a ' * '
I've already tried several options (also used the Perl Regular Expression Quick Reference Card)
I assume I need to use ' ^ but can't get it done.
Please help

(This post was edited by robertico on Mar 22, 2013, 12:41 AM)

rovf
Veteran

Mar 22, 2013, 2:16 AM

Post #2 of 15 (1122 views)

Re: [robertico] regex help [In reply to]	Can't Post

First, curly braces have special meaning in a Perl regexp; you need to escape them with \

Second, you didn't anchor your regexp, so the pattern is searched anywhere in the line. With ^ and $ you can denote the start and end of the line.

robertico
Novice

Mar 22, 2013, 2:46 AM

Post #3 of 15 (1121 views)

Re: [rovf] regex help [In reply to]	Can't Post

Quote

First, curly braces have special meaning in a Perl regexp; you need to escape them with \

Ok thank you. It works as expected so I didn't noticed.

Quote

Second, you didn't anchor your regexp, so the pattern is searched anywhere in the line. With ^ and $ you can denote the start and end of the line.

That's exactly what I want.

In one line I need to extract multiple sub strings between ' { ' and ' } '.
But some of them contain the character ' * ' somewehere between the curly braces. I need to exclude these sub strings.

rovf
Veteran

Mar 22, 2013, 4:20 AM

Post #4 of 15 (1119 views)

Re: [robertico] regex help [In reply to]	Can't Post

Sorry, but I don't understand your description. It's confusing to me. Which part of each line exactly do you want to extract?

robertico
Novice

Mar 22, 2013, 5:36 AM

Post #5 of 15 (1118 views)

Re: [rovf] regex help [In reply to]	Can't Post

I'll give you an example;

Code

some useless text {"to_id": 0, "message": "This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 1, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more 
some useless text {"to_id": 3, "message": "*This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 4, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more

For line one I need both parts between the curly braces;
Expected result;

Code

{"to_id": 0, "message": "This is a sample", "message_id": 1000, "from_id": 999}  
{"to_id": 1, "message": "This is a sample", "message_id": 1000, "from_id": 999}

For line two I only need one part (Not the part with the ' * ')

Expected result;

Code

{"to_id": 4, "message": "This is a sample", "message_id": 1000, "from_id": 999}

(This post was edited by robertico on Mar 22, 2013, 5:50 AM)

rovf
Veteran

Mar 22, 2013, 5:55 AM

Post #6 of 15 (1112 views)

Re: [robertico] regex help [In reply to]	Can't Post

Well, this makes it clearer. From your examples, I see that you do NOT need to extract lines starting and ending with {} (since none of the example lines have a curly brace in the beginning).

I would do the following approach:

First, collect all the {...} substrings, without worrying that they might contain a '*'. Then drop those which have an asterisk.

IMO this is easiest if you also drop the idea to use a statement modifier. Use a normal while loop, and then a if inside to choose the desired lines.

robertico
Novice

Mar 22, 2013, 6:08 AM

Post #7 of 15 (1111 views)

Re: [rovf] regex help [In reply to]	Can't Post

I'm lost now.

I'm reading a text file line by line.
If a line matches a pattern, I use the regex to extract the sub string between the curly brace (sometimes more than one per line)
But if there's an asterisk between the curly brace, I need to ignore that one.
When it matches the regex I need to do some further processing (JSON and write results to another file)

So I can't simply ignore the one with an asterisk.

Do I need to use a second regex before further processing or is it possible to create a suitable regex to do both at the same time ?

Sorry, it seems a little bit confusing.

BillKSmith
Veteran

Mar 22, 2013, 6:33 AM

Post #8 of 15 (1103 views)

Re: [robertico] regex help [In reply to]	Can't Post

Here is my version of the rovf solution. Use a module to match parens and grep to filter unwanted results.

Code

use strict; 
use warnings; 
use Regexp::Common qw /balanced/; 
my $BALANCED = qr/$RE{balanced}{-parens=>'{}'}/; 
while (my $line = <DATA>) { 
    (my @first_pass) = $line =~ /($BALANCED)/g; 
    my @required = grep  {!/[*]/} @first_pass; 
    print join( "\n", @required), "\n\n"; 
} 
__DATA__ 
some useless text {"to_id": 0, "message": "This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 1, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more 
some useless text {"to_id": 3, "message": "*This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 4, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more

Good Luck,
Bill

Kenosis
User

Mar 22, 2013, 6:41 AM

Post #9 of 15 (1101 views)

Re: [robertico] regex help [In reply to]	Can't Post

I think rovf is right on target. Also don't use $& as it has a performance cost within Perl. Use the results of your capture that's contained in $1.

Consider the following:

Code

use warnings; 
use strict; 
use Data::Dumper; 
 
my @matches; 
 
while ( my $line = <DATA> ) { 
	while ( $line =~ /\{(.*?)\}/g ) { 
		push @matches, $1 if $1 and $1 !~ /\*/; 
	} 
} 
print Dumper \@matches; 
 
__DATA__ 
some useless text {"to_id": 0, "message": "This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 1, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more  
some useless text {"to_id": 3, "message": "*This is a sample", "message_id": 1000, "from_id": 999} more useless text {"to_id": 4, "message": "This is a sample", "message_id": 1000, "from_id": 999} and more

Output:

Code

$VAR1 = [ 
          '"to_id": 0, "message": "This is a sample", "message_id": 1000, "from_id": 999', 
          '"to_id": 1, "message": "This is a sample", "message_id": 1000, "from_id": 999', 
          '"to_id": 4, "message": "This is a sample", "message_id": 1000, "from_id": 999' 
        ];

The second regex just checks whether "*" is contained in your capture. The $1 and $1 !~ /\*/ notation just makes sure that you did, in fact, capture something, since you could 'capture' an empty string between "{}" and it would be true that "*" isn't within "".

Hope this helps!

robertico
Novice

Mar 22, 2013, 7:03 AM

Post #10 of 15 (1099 views)

Re: [BillKSmith] regex help [In reply to]	Can't Post

Works almost as expected (unwanted results are filtered out) but the others are printed twice and I've a lot of empty lines.

Already tried to change it this way, but no luck.

Code

		print OUTFILE join( "\n", @required), "\n\n";

Code

foreach (@required) { 
   print OUTFILE "$_"; 
}

robertico
Novice

Mar 22, 2013, 7:11 AM

Post #11 of 15 (1095 views)

Re: [Kenosis] regex help [In reply to]	Can't Post

This one works excellent. Thank you very much !! Laugh

Quote

Also don't use $& as it has a performance cost within Perl. Use the results of your capture that's contained in $1.

When I use $1 the curly brace is ommitted in the result and with $& it's included (as desired)

(This post was edited by robertico on Mar 22, 2013, 7:18 AM)

BillKSmith
Veteran

Mar 22, 2013, 7:33 AM

Post #12 of 15 (1091 views)

Re: [robertico] regex help [In reply to]	Can't Post

My print was designed to duplicate your sample output. It puts a newline after each result and leaves a blank line after the results for each input line. Your first print sends this to OUTFILE.

Your second print sends all the results to a single long line in OUTFILE.

You probably want only the first. (Remove one of the two newlines at the end to eliminate the blank lines.)
Good Luck,
Bill

Kenosis
User

Mar 22, 2013, 7:41 AM

Post #13 of 15 (1090 views)

Re: [robertico] regex help [In reply to]	Can't Post

Ah! Then try the following--and there would be no need to check the returned capture for emptiness, since--at the very least--it would contain "{}":

Code

	while ( $line =~ /(\{.*?\})/g ) { 
		push @matches, $1 if $1 !~ /\*/; 
	}

This just moves the capture to include the braces...

rovf
Veteran

Mar 22, 2013, 8:00 AM

Post #14 of 15 (1085 views)

Re: [BillKSmith] regex help [In reply to]	Can't Post

My I ask why you put parentheses around my @first_pass? It's list context anyway.

BillKSmith
Veteran

Mar 22, 2013, 8:17 AM

Post #15 of 15 (1081 views)

Re: [rovf] regex help [In reply to]	Can't Post

Bad habit! By the time that I noticed it, the OP had already used the code. It seemed to be to late to fix it.
Good Luck,
Bill

Search for (options)

Web Applications & Managed Hosting Powered by Gossamer Threads
Visit our Mailing List Archives