in reply to Re^3: Using Regexp::Common
in thread Using Regexp::Common

Hi Bill

in my requirement the example: 

10,101,110.11010110

[download]
the numbers can be separated by , or space or tab or ; .

. is a decimal and that particular number should be considered a float. I do not have any requirement to extract out the exponent and the mantissa part.

Replies are listed 'Best First'.
Re^5: Using Regexp::Common
by AnomalousMonk (Archbishop) on Sep 20, 2015 at 16:52 UTC

    The problem I and apparently others had with your example was that if  , (comma) is used as both a number separator and as a three-digit, whole-number group separator, the string  '10,101,110.11010110' is ambiguous. Is it a single binary real 10,101,110.11010110, the whole-number portion of which has three groups? Is it a binary integer 10, followed by a binary real 101,110.11010110? Or is it a binary integer 10, followed by a binary integer 101, followed by a binary real 110.11010110? Or how about a grouped binary integer 10,101 followed by a binary real 110.11010110?

    Trying to use a comma as both a number separator and a group separator seems like a really bad idea. Trying to parse data in which a comma is used in this way seems like a nightmare.

    Give a man a fish:  <%-{-{-{-<

[reply]
[d/l]
[select]
Re^5: Using Regexp::Common
by BillKSmith (Monsignor) on Sep 20, 2015 at 20:00 UTC
    The following program will match every base-10 number in your string. The number may be in any format known to perl. 
    use strict;
use warnings;
use Regexp::Common;
$_ =  '10,101,110.11010110 ';


local $, = "\n";
print /($RE{num}{real})/g;

    [download]
    OUTPUT: 
    10
101
110.11010110

    [download]
    Note that the group and sep options are not specified. They only apply to separations within a single number. The absence of a keep option tells the module not to capture anything. We use our own parenthesis to capture exactly what we want. We use the /g match option to find (and capture) all the matches.
    Bill
[reply]
[d/l]
[select]
      We use our own parenthesis ...

      With use of the  /g regex modifier, capturing parentheses are not needed (although they do no harm): in list context, all matched sub-strings are returned. 

      c:\@Work\Perl>perl -wMstrict -MRegexp::Common -le
"$_ =  '10,101,110.11010110';
 ;;
 print qq{'$_'} for /$RE{num}{real}/g;
"
'10'
'101'
'110.11010110'

      [download]

      The absence of a keep option tells the module not to capture anything.

      But I thought that the idea behind the use of  -keep was that justrajdeep wanted to "... divide [the extracted numbers] into floats/integer etc.", i.e., classify them, and for this the capture of number components (sign, whole number, fractional part, etc.) by  -keep could be made to work nicely.

      ... group and sep options are not specified. They only apply to separations within a single number.

      justrajdeep seems to want to handle comma-separated, grouped whole numbers (and binary ones at that), but I agree that his or her requirements are a bit confusing, at least to me.

      Give a man a fish:  <%-{-{-{-<

[reply]
[d/l]
[select]

        My intention was to help justrajdeep clarify his requirements by providing a base line solution to critique.

        Your use of 'for' neatly solves the problem I had with keep.

        I was not aware that my parenthesis were not needed. I probably still would have specified them.

        Bill
[reply]

In Section Seekers of Perl Wisdom