source: recipecards/trunk/recipe2xml @ 4

#!/usr/bin/perl -w
use strict;

#TODO: look at this existing XML format: http://www.happy-monkey.net/recipebook/doc/author-tutorial.html

$/ = '';

print qq{<recipe xmlns="http://xmlns.grim.ath.cx/recipe">\n};
while (<>) {
    chomp;
    s/^\s+//g;
    s/\s+$//g;
    if (/^##\s+(.*)/) {
        # title
        print "<title>$1</title>";
    } elsif (/^\d/) {
        # ingredient list
        print '<ingredients>';
        foreach (split /\n/, $_) {
            /^\s+/ ? print '<note>' . apply_symbols($_) . '</note>'
                   : print '<ingredient>' . apply_symbols($_) . '</ingredient>';
        }
        print '</ingredients>';
    } else {
        # TODO: catch special body paragraphs, like "Note:", "Serves", "Yields", etc.
        # body
        # fold the lines together
        s/\n/ /;
        print '<para>' . apply_symbols($_) . '</para>';
    }
}
print '</recipe>';


sub apply_symbols {
    my $string = shift;
    for ($string) {
        # numeric dimensions
        s/(\d)x/$1⨯/g;
        # temperature
        s/(\d)\s*deg(\.|rees)(\s*F)?/$1℉/g;
        # fractions
        s{1/3}{⅓}g;
        s{2/3}{⅔}g;
        s{1/4}{¼}g;
        s{1/2}{½}g;
        s{3/4}{¾}g;
        s{1/8}{⅛}g;
        # inches
        s/(\d)[ -]?(in\.|in\b|inch)/$1″/g;
        # numeric range en dashes
        s/(\d)-(\d)/$1–$2/g;
        # remove dash before vulgar fractions
        s/(\d)-([¼½¾⅓⅔])/$1$2/g;
    }
    return $string;
}