Trouble using 'while' loop to evaluate multiple lines, Perl

Question

Thank you in advance for indulging an amateur Perl question. I'm extracting some data from a large, unformatted text file, and am having trouble combining the use of a 'while' loop and regular expression matching over multiple lines.

First, a sample of the data:

 01-034575 18/12/2007  258,750.00 11,559.00  36       -2     0     6    -3     2    -2     0     2     1    -1     3     0     5    15
                                                      -13   -44   -74  -104  -134  -165  -196  -226  -257  -287  -318  -349  -377  -408  -438
                                                      -469  -510  -541  -572  -602  -633  -663
      Atraso Promedio --->        0.94

The first sequence, XX-XXXXXX is a loan ID number. The date and the following two numbers aren't important. '36' is the number of payments. The following sequence of positive and negative numbers represent how late/early this client was for this loan at each of the 36 payment periods. The '0.94' following 'Atraso Promedio' is the bank's calculation for average delay. The problem is it's wrong, since they substitute all negative (i.e. early) payments in the series with zeros, effectively over-stating how risky a client is. I need to write a program that extracts ID and number of payments, and then dynamically calculates a multi-line average delay.

Here's what I have so far:

#Create an output file
open(OUT, ">out.csv");
print OUT "Loan_ID,Atraso_promedio,Atraso_alt,N_payments,\n";

open(MYINPUTFILE, "<DATA.txt");
while(<MYINPUTFILE>){

    chomp($_);

    if($ID_select != 1 && m/(\d{2}\-\d{6})/){$Loan_ID = $1, $ID_select = 1} 

    if($ID_select == 1 && m/\d{1,2},\d{1,3}\.00\s+\d{1,2},\d{1,3}\.00\s+(\d{1,2})/)  {$N_payments = $1, $Payment_find = 1};

    if($Payment_find == 1 && $ID_select == 1){

            while(m/\s{2,}(\-?\d{1,3})/g){
                $N++; 
                $SUM = $SUM + $1;
                print OUT "$Loan_ID,$1\n"; #THIS SHOWS ME WHAT NUMBERS THE CODE IS GRABBING. ACTUAL OUTPUT WILL BE WRITTEN BELOW
                print $Loan_ID,"\n";
            }


        if(m/---> *(\d*.\d*)/){$Atraso = $1, $Atraso_select = 1}
        if($ID_select == 1 && $Payment_find == 1 && $Atraso_select == 1){
                ...

There's more, but the while loop is where the program is breaking down. The problem is with the pattern modifier, 'g,' which performs a global search of the string. This makes the program grab numbers that I don't want, such as the '1' in loan ID and the '36' for the number of payments. I need the while loop to start from wherever the previous line in the code left off, which should be right after it has identified the number of loans. I've tried every pattern modifier that I've been able to look up, and only 'g' keeps me out of an infinite loop. I need the while loop to go to the end of the line, then start on the next one without combing over the parts of the string already fed through the program.

Thoughts? Does this make sense? Would be immensely grateful for any help you can offer. This work is pro-bono, unpaid: just trying to help out some friends in a micro-lending institution conduct a risk analysis.

Cheers,
Aaron

heijp06 · Accepted Answer · 2011-02-04 09:02:48Z

2

The problem is probably easier using split, for instance something like this:

use strict;
use warnings;

open DATA, "<DATA.txt" or die "$!";

my @payments;
my $numberOfPayments;
my $loanNumber;

while(<DATA>)
{
    if(/\b\d{2}-\d{6}\b/)
    {
        ($loanNumber, undef, undef, undef, $numberOfPayments, @payments) = split;
    }
    elsif(/Atraso Promedio/)
    {
        my (undef, undef, undef, $atrasoPromedio) = split;

        # Calculate average of payments and print results

    }
    else
    {
        push(@payments, split);
    }
}

answered Feb 4, 2011 at 9:02

heijp06

11.8k1 gold badge43 silver badges63 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Bob Lied · Accepted Answer · 2011-02-04 09:03:46Z

0

If the data's clean enough, I might approach it by using split instead of regular expressions. The first line is identifiable if field[0] matches the form of a loan number and field[1] matches the format of a date; then the payment dates are an array slice of field[5..-1]. Similarly testing the first field of each line tells you where you are in the data.

answered Feb 4, 2011 at 9:03

Bob Lied

6414 silver badges11 bronze badges

Comments

typo.pl · Accepted Answer · 2011-02-04 09:17:56Z

0

Peter van her Heijden's answer is a nice simplification for a solution.

To answer the OP's question about getting the regexp to continue where it left off, see Perl operators - regexp-quote-like operators, specifically the section "Matching in list context" and the "\G assertion" section just after that.

Essentially, you can use m//gc along with the \G assertion to use regexps match where previous matches left off.

The example in the "\G assertion" section about lex-like scanners would seem to apply to this question.

answered Feb 4, 2011 at 9:17

typo.pl

8,9302 gold badges31 silver badges29 bronze badges

Collectives™ on Stack Overflow

Trouble using 'while' loop to evaluate multiple lines, Perl

3 Answers 3

Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related