#!/usr/bin/perl

################################################

# extract_matches.pl
# Version 1

# Robert D. Cormia
# UCSC Programming for Bioinformatics II
# October 10, 2003

################################################

# Problem 3-1 extract_matches

# Write a Perl program called 'extract_matches' to do the following:
#     a. Read a file called 'strings_data' line by line
#     b. In each line, look for the presence of the pattern in the
#         following string and fetch the offsets where the pattern appears
#         in the string
#     c. Create a hash called 'match_offsets' where the key is colon separated
#         pattern and string and the value is colon separated offsets!
#     d. Print the key-value pairs of this hash into a file called 'matches'
#

use strict;

my $pat;
my $str;
my $pos;
my $line;
my @lines;
my $fname = 'strings_data.txt';
my @file_contents;
my $motifs;
my @seqs;
my @array_of_offsets;
my $dir;
my $date;
my %match_offsets;
my $k;
my $v;

# Pseudocode
# Open file
# Read each line
# Extract from each line the pattern (first part) and string (second part)
# Use the 'matching code' from Motifs and store_matches program
# to extract offsets into a hash
# Print the key-value pairs from hash into file called matches

# Open file strings_data.txt

# open file, if open fails call die
# die will make the program exit after displaying the message
# $! is a special variable that contains the error
# message from the system

open( INPUT, "strings_data.txt" ) or die "Cannot open file: $!";

@lines = <INPUT>;
close INPUT;

# Process lines code block

# read the file one line at a time. <> is called as
# the diamond operator it reads from the file one
# line at a time in a scalar context when the file's
# end is reached, <> returns an EOF which is
# considered as a false value, so the while loop ends

# Block for determining position of offsets

foreach (@lines) {

    # test for possible blank line and skip if true
    chomp $_;
    if ( $_ eq "" ) { next; }

    ( $pat, $str ) = split ( ' ', $_ );

    $pos = 0;
    @array_of_offsets = ();
    while ( $pos != -1 ) {
        $pos = index( $str, $pat, $pos );

        if ( $pos != -1 ) {
            push ( @array_of_offsets, $pos );
	    $match_offsets{"$pat:$str"} = join(':', @array_of_offsets);

            # Start the next index search just past the last found pattern.
            $pos++;
        }
    }

}

    # -----------------------------------------------------------------
    # print it as:
    # $key => $value
    # create an output file.
    # open match.txt
    # -----------------------------------------------------------------

    open( OUTPUT, ">matches.txt" ) or die "Cannot open file: $!";


    while ( ($k, $v) = each %match_offsets ) {
	print "$k => $v\n";
	print OUTPUT "$k => $v";
    }

    print "\n";

    close(OUTPUT);

__END__