#!/usr/bin/perl

$docxfile = $ARGV[0];
$newline = "\n";

open(DOCUMENT, "unzip -p '$docxfile' word/document.xml |");

while ($line = <DOCUMENT>) {

    # convert paragraphs to newlines
    $line =~ s/<w:p>/\n\n/g;
    $line =~ s/<w:p [^>]*>/\n\n/g;

    # chop some extra newlines out
    $line =~ s/\n\n\n\n/\n\n/g;
    $line =~ s/\n\n\n/\n\n/g;

    # substitute all other OOXML tags to nothing
    $line =~ s/<[^>]+>//g;

    # substitute all non-printing characters to nothing
    $line =~ s/[^[:print::space:]]+//g;

    # print finished product
    print $line;
}

print "\n";

close(DOCUMENT);