use URI::Escape;
use strict;
use warnings;

# Copyright Michael Anthony Puls II, http://shadow2531.com/
# Distributed under the Boost Software License, Version 1.0.
# http://boost.org/LICENSE_1_0.txt .

# Rules to follow http://shadow2531.com/opera/testcases/mailto/rfc2368-3.html
# Example c++ versions of these functions are in  
# http://shadow2531.com/cpp/mailto_funcs1.zip
# http://shadow2531.com/cpp/mailto_funcs2.zip
# http://shadow2531.com/cpp/mailto_funcs3.zip
# http://shadow2531.com/cpp/mailto_funcs4.zip
# http://shadow2531.com/cpp/mailto_funcs5.zip
# http://shadow2531.com/cpp/mailto_funcs6.zip
# http://shadow2531.com/cpp/mailto_funcs7.zip
# http://shadow2531.com/cpp/mailto_funcs8.zip
# http://shadow2531.com/cpp/mailto_funcs9.zip
# http://shadow2531.com/cpp/mailto_funcs10.zip
# http://shadow2531.com/cpp/mailto_funcs11.zip
# Example Javascript versions of these functions are in http://shadow2531.com/opera/testcases/mailto/mailto_funcs.js
# Example Java versions of these functions are in http://shadow2531.com/java/MailtoFuncs.java
# Example Ruby versions of these functions are in http://shadow2531.com/ruby/mailto_funcs.rb
# Example Python versions of these functions are at http://shadow2531.com/py/mailto_funcs.py
# Example TCL versions of these functions are at http://shadow2531.com/tcl/mailto_funcs.tcl
# Newest versions:  http://shadow2531.com/opera/testcases/mailto/MailtoURIParserPack.zip

# Create a modified value that only contains a &-separated list of hname=hvalue pairs (so it can be tokenized)
sub prepareURIForParsing {
    my $uri = $_[0];
    my $check = lc($uri);
    if (index($check, "mailto:?") != -1) {
        return substr($uri, 8);
    } elsif (index($check, "mailto:") != -1) {
        my $prep = $uri;
        $prep =~ s/\?/&/;
        return "to=" . substr($prep, 7);
    }
    return "";
}

# skip_method
# 0 = concat (with delim) all hvalues for the hname (even if they're empty)
# 1 = concat (with delim) all non-empty hvalues for the hname (to, cc and bcc rule)
# 2 = concat (with delim) first non-empty hvalue for the hname and all (even if empty) hvalues after that. (body rule)
# 3 = only use the last hvalue for the hname (even if the hvalue is empty) (subject rule)

sub searchMailtoURI {
    my $uri = $_[0];
    my $hname = $_[1];
    my $delim = $_[2];
    my $skip_method = $_[3];
    my $puri = prepareURIForParsing($uri);
    my $lkey = lc($hname);
    my @hlist = split(/&/, $puri);
    my $zskip = 1; # with skip_method 0, don't want to add a delim to an empty value during the first occurance of the hname
    my $value = "";
    foreach (@hlist) {
        my $eq = index($_, "=");
        if ($eq == -1 or $_ eq "" or lc(substr($_, 0, $eq)) ne  $lkey) {
            next;
        }
        if ($skip_method != 3) {
            if ($value ne "") {
                if ($skip_method eq 1 and substr($_, $eq + 1) eq "") {
                    next;
                }
                $value .= $delim;
            } elsif ($skip_method == 0) {
                if ($zskip == 0) {
                    $value .= $delim;
                }
                $zskip = 0;
            }
            $value .= substr($_, $eq + 1);
        } else {
            $value = substr($_, $eq + 1);
        }
    }
    return $value;
}

sub getEncodedTOValue {
    return searchMailtoURI($_[0], "to", "%2C%20", 1);
}
sub getDecodedTOValue {
    return uri_unescape(getEncodedTOValue($_[0]));
}

sub getEncodedSubjectValue {
    return searchMailtoURI($_[0], "subject", "", 3);
}
sub getDecodedSubjectValue {
    return uri_unescape(getEncodedSubjectValue($_[0]));
}

sub getEncodedBodyValue {
    return searchMailtoURI($_[0], "body", "%0D%0A", 2);
}
sub getDecodedBodyValue {
    return uri_unescape(getEncodedBodyValue($_[0]));
}

sub getEncodedCCValue {
    return searchMailtoURI($_[0], "cc", "%2C%20", 1);
}
sub getDecodedCCValue {
    return uri_unescape(getEncodedCCValue($_[0]));
}

sub getEncodedBCCValue {
    return searchMailtoURI($_[0], "bcc", "%2C%20", 1);
}
sub getDecodedBCCValue {
    return uri_unescape(getEncodedBCCValue($_[0]));
}

# Generates a basic, compatible mailto URI with no duplicate hnames
#(for passing to clients with cheap mailto URI parsing)

sub getNormalizedMailtoURI {
    return "mailto:" . getEncodedTOValue($_[0]) . "?subject=" . getEncodedSubjectValue($_[0]) . "&body=" . getEncodedBodyValue($_[0]) . "&cc=" . getEncodedCCValue($_[0]) . "&bcc=" . getEncodedBCCValue($_[0]);
}

# Example of resolving replacement keys in a "open with other application" mailto command field
#  %1$2s = original mailto URI
# %2$2s = Normalized mailto URI
# %3$2s = decoded to value
# %4$2s = encoded to value
# %5$2s = decoded subject value
# %6$2s = encoded subject value
# %7$2s = decoded body value
# %8$2s = encoded body value
# %9$2s = decoded cc value
# %10$2s = encoded cc value
# %11$2s = decoded bcc value
# %12$2s = encoded bcc value
# %% = %

sub openWithOtherApplicationCommandAddInReplacements {
    my $uri = $_[0];
    my $command = $_[1];
    return sprintf($command, $uri, getNormalizedMailtoURI($uri), getDecodedTOValue($uri), getEncodedTOValue($uri), getDecodedSubjectValue($uri), getEncodedSubjectValue($uri), getDecodedBodyValue($uri), getEncodedBodyValue($uri), getDecodedCCValue($uri), getEncodedCCValue($uri), getDecodedBCCValue($uri), getEncodedBCCValue($uri));
}

sub test {
    my $t = $_[0];
    my $in = $_[1];
    my $out = $_[2];
    if ($in eq $out) {
        print "test " . $t . " = pass";
    } else {
        print "test" . $t . "= fail" . "\n";
        print "before_" . $in . "_after";
    }
    print "\n";
}

# tests are the same as the java versions only some of the 2nd arguments to openWithOtherApplicationCommandAddInReplacements are encased with in 
# single quotes with the slashes remove from in front of the double quotes. (so, the sprintf stuff can work)

test(1, getEncodedTOValue("mailto:?to=email1%40site.com&to=email2%40site.com"), "email1%40site.com%2C%20email2%40site.com");
test(2, getEncodedTOValue("mailto:email1%40site.com%2C%20email2%40site.com"), "email1%40site.com%2C%20email2%40site.com");
test(3, getEncodedSubjectValue("mailto:?subject=one"), "one");
test(4, getEncodedSubjectValue("mailto:?subject=one&subject=two"), "two");
test(5, getEncodedSubjectValue("mailto:?subject=one&subject="), "");
test(6, getEncodedBodyValue("mailto:?body=line1%0D%0Aline2"), "line1%0D%0Aline2");
test(7, getEncodedBodyValue("mailto:?body=line1%0D%0Aline2&body=line3%0D%0Aline4"), "line1%0D%0Aline2%0D%0Aline3%0D%0Aline4");
test(8, getEncodedBodyValue("mailto:?body=&body=&body=&body=line1%0D%0Aline2&body=line3%0D%0Aline4&body=&body=&body=line8"), "line1%0D%0Aline2%0D%0Aline3%0D%0Aline4%0D%0A%0D%0A%0D%0Aline8");
test(9, getEncodedBodyValue("mailto:?body=&body="), "");
test(10, getEncodedCCValue("mailto:?cc=email1%40site.com"), "email1%40site.com");
test(11, getEncodedCCValue("mailto:?cc=email1%40site.com%2C%20email2%40site.com&cc=email5%40site.com%2C%20email6%40site.com"), "email1%40site.com%2C%20email2%40site.com%2C%20email5%40site.com%2C%20email6%40site.com");
test(12, getEncodedCCValue("mailto:"), "");
test(13, getEncodedCCValue("mailto:?cc=&cc=&cc=&cc=email1%40site.com"), "email1%40site.com");
test(14, getEncodedBCCValue("mailto:?bcc=email1%40site.com"), "email1%40site.com");
test(15, getEncodedBCCValue("mailto:?bcc=email1%40site.com%2C%20email2%40site.com&bcc=email5%40site.com%2C%20email6%40site.com"), "email1%40site.com%2C%20email2%40site.com%2C%20email5%40site.com%2C%20email6%40site.com");
test(16, getEncodedBCCValue("mailto:"), "");
test(17, getEncodedBCCValue("mailto:?bcc=&bcc=&bcc=&bcc=email1%40site.com"), "email1%40site.com");
test(18, getNormalizedMailtoURI("mailto:"), "mailto:?subject=&body=&cc=&bcc=");
test(19, getNormalizedMailtoURI("mailto:email1%40site.com%2C%20email2%40site.com?subject=1&subject=2&subject=3&body=line1%0D%0Aline2&body=line3%0D%0Aline4&cc=test1%40site.com%2C%20test2%40site.com&cc=test3%40site.com%2C%20test4%40site.com&bcc=zap1%40site.com%2C%20zap2%40site.com&bcc=zap3%40site.com%2C%20zap4%40site.com&to=email3%40site.com&to=email4%40site.com"), "mailto:email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com?subject=3&body=line1%0D%0Aline2%0D%0Aline3%0D%0Aline4&cc=test1%40site.com%2C%20test2%40site.com%2C%20test3%40site.com%2C%20test4%40site.com&bcc=zap1%40site.com%2C%20zap2%40site.com%2C%20zap3%40site.com%2C%20zap4%40site.com");
test(20, getNormalizedMailtoURI("mailto:email%40site.com?subject=1&body=2&cc=test%40site.com&bcc=zap%40site.com"), "mailto:email%40site.com?subject=1&body=2&cc=test%40site.com&bcc=zap%40site.com");
test(21, getEncodedSubjectValue("mailto:?subject=1&subject="), "");
test(22, searchMailtoURI("mailto:?foo=&foo=&foo=", "foo", "^", 0), "^^");
test(23, searchMailtoURI("mailto:?foo=1&foo=&foo=", "foo", "^", 0), "1^^");
my $big = "mailto:%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com?to=&to=email3%40site.com%2C%20email4%40site.com&to=email5%40site.com%2C%20email6%40site.com&to=&to=&subject=You%20should%20not%20see%20me%2E&subject=last%20call%20for%20cats%20%26%20dogs%2E&body=&body=&body=&body=line1%0D%0Aline2&body=&body=&body=&body=line6&body=line7%0D%0A1+2+3+4+5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A&cc=%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com&cc=&cc=someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar%2Ecom%3E&cc=&cc=&bcc=%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com&bcc=&bcc=someoneelse3%40site.com%2C%20someoneelse4%40site.com&bcc=&bcc=";
test(24, getNormalizedMailtoURI($big), "mailto:%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com%2C%20email5%40site.com%2C%20email6%40site.com?subject=last%20call%20for%20cats%20%26%20dogs%2E&body=line1%0D%0Aline2%0D%0A%0D%0A%0D%0A%0D%0Aline6%0D%0Aline7%0D%0A1+2+3+4+5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A&cc=%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com%2C%20someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar%2Ecom%3E&bcc=%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com%2C%20someoneelse3%40site.com%2C%20someoneelse4%40site.com");
test(25, openWithOtherApplicationCommandAddInReplacements("mailto:email%40site.com", 'program.exe %4$2s'), "program.exe email%40site.com");
test(26, openWithOtherApplicationCommandAddInReplacements("mailto:email%40site.com", 'program.exe %1$2s'), "program.exe mailto:email%40site.com");
test(27, openWithOtherApplicationCommandAddInReplacements("mailto:?to=email%40site.com", 'program.exe "%2$2s"'), "program.exe \"mailto:email%40site.com?subject=&body=&cc=&bcc=\"");
test(28, openWithOtherApplicationCommandAddInReplacements($big, '"c:\\program files\\program\\program.exe" "%4$2s" "%6$2s" "%8$2s" "%10$2s" "%12$2s"'), "\"c:\\program files\\program\\program.exe\" \"%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com%2C%20email5%40site.com%2C%20email6%40site.com\" \"last%20call%20for%20cats%20%26%20dogs%2E\" \"line1%0D%0Aline2%0D%0A%0D%0A%0D%0A%0D%0Aline6%0D%0Aline7%0D%0A1+2+3+4+5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A\" \"%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com%2C%20someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar%2Ecom%3E\" \"%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com%2C%20someoneelse3%40site.com%2C%20someoneelse4%40site.com\"");
test(29, openWithOtherApplicationCommandAddInReplacements($big, "%%1%%"), "%1%");
test(30, searchMailtoURI("mailto:?bla=test&foo=&foo=&foo=", "foo", "^", 0), "^^");
test(31, searchMailtoURI("mailto:?bla=test&foo=&foo=1&foo=&bam=this", "foo", "^", 0), "^1^");
test(32, searchMailtoURI("mailto:?bla=test&foo=1&foo=1&foo=&bam=this", "foo", "^", 0), "1^1^");
test(33, searchMailtoURI("mailto:?bla=test&foo=1&foo=2&foo=3&bam=this", "foo", "^", 0), "1^2^3");
test(34, getEncodedSubjectValue("mailto:?subject=1+2+3+4+5"), "1+2+3+4+5");
test(35, searchMailtoURI("mailto:?foo=", "foo", "^", 0), "");
test(36, getDecodedSubjectValue("mailto:?subject=%E2%88%9A"), "\xE2\x88\x9A");
test(37, getEncodedBodyValue("mailto:?body=line1&body=line2"), "line1%0D%0Aline2");
test(38, getEncodedTOValue("mailto:email1%40site.com?to=email2%40site.com"), "email1%40site.com%2C%20email2%40site.com");
test(39, getEncodedSubjectValue("mailto:?subject=first&subject=last"), "last");
test(40, getNormalizedMailtoURI("mailto:?to=email%40site.com&subject=first&subject=second&subject=last&body=line1&body=line2"), 'mailto:email%40site.com?subject=last&body=line1%0D%0Aline2&cc=&bcc=');
test(41, openWithOtherApplicationCommandAddInReplacements("mailto:email1%40site.com?to=email2%40site.com", 'program.exe -to "%4$2s"'), "program.exe -to \"email1%40site.com%2C%20email2%40site.com\"");
test(42, openWithOtherApplicationCommandAddInReplacements("mailto:email1%40site.com?to=email2%40site.com", 'program.exe %2$2s'), "program.exe mailto:email1%40site.com%2C%20email2%40site.com?subject=&body=&cc=&bcc=");
test(43, getEncodedBodyValue("mailto:?body=&body=&body=line1&body=&body=&body=line4&body=&body="), "line1%0D%0A%0D%0A%0D%0Aline4%0D%0A%0D%0A");