citi_stmt_to_csv-simple

#!/usr/bin/perl

# run on output of `pdftotext -simple Citi_CreditCardStatement.pdf Citi_CreditCardStatement.p2t`
# see %~dp0/CCStmtP2tToCsv.pm for details

use strict;
use warnings;

use FindBin;
use lib $FindBin::Bin;
use CCStmtP2tToCsv;

my $opts = CCStmtP2tToCsv::getopts();

my $rdt = '\d{2}/\d{2}';
my $rdesc = '\w.*\w';
my $rdamtcapt = '([\d,]*\.\d{2})';
my $rtxn = qr"\s+($rdt)\s+($rdesc)\s+(\-?)\$$rdamtcapt\b";
my $retxn = "^(?:$rdt)?$rtxn";

for my $ifnm ( sort @ARGV ) {  # sort so that write_all_csv content will be sorted (assumes stmt filenames are validly date-sortable)
   my $lp_Payments = sub { # my $self = shift; $self->parse_new_txn( $retxn, 'credit' );  no way to cross-check credits in Citi stmts (because Citi stmts provide no total for credits)
      };
   my ($cardholder);  # multiline parse support
   my $lp_CardholderSummary = sub { my $self = shift;
    # print "chs: $_\n" if $opts->{v};
      if( !defined($cardholder) ) {
         if( m"^($rdesc)\s+Card ending in\s+\d{4}" ) {
            $cardholder = $1;
            print "cardholder = $cardholder\n" if $opts->{v};
            }
         }
      elsif( m"^New Charges\s+\$$rdamtcapt" ) {
         my $cardtotal = CCStmtP2tToCsv::tocents( $1 );  # for cross-checking
         print "cardholder = $cardholder; cardtotal = $cardtotal\n" if $opts->{v};
         $self->set_total( $cardholder, $cardtotal );
         my $ch_capt = $cardholder;  # required!
         $self->add_section_hdr( $cardholder, sub { my $self = shift; $self->parse_new_txn( $retxn, 'charge', $ch_capt ); } );
         $cardholder = undef;  # we're done with this $cardholder
         }
      };
   my $lp_ignore = sub {;};
   my $lp_BillPeriod = sub { my $self = shift;
      if( m"^\s*Billing\s+Period:\s+\d{2}/\d{2}/(\d{2})\-(\d{2})/(\d{2})/(\d{2})" ) { print "yrMin/yrMax raw $1/$4\n";
         $self->set_stmtOpenCloseDates();
         }
      };
   my $init_key = '(start_of_file)';
   CCStmtP2tToCsv::process_stmt_p2t( shift,
      {
      $init_key                                 => $lp_BillPeriod,
      'CARDHOLDER\s+SUMMARY'                    => $lp_CardholderSummary,
      'Payments,\s+Credits\s+and\s+Adjustments' => $lp_Payments,
      'Fees\s+Charged'                          => $lp_ignore,
      }, $init_key, [ qw( charge ) ], $opts
      );
   }
CCStmtP2tToCsv::write_all_csv() if @ARGV > 1;