# -----------------------------------------------------------------------------------------------
# Author: Dean Stringer
# Description:
#   takes an XML filename as input, parses each line escaping all bracket
#   characters and wrapping tags and attributes in <span> tahs so the XML
#   can be displayed in a webpage
#
#   can output CSS styles to render a page in Mozilla/Firefox colours
#   or the Microsoft/IE engine colours
#
# Notes:
#   (1) this script would be more reliably implemented as an XSLT
#       transform, or at least using an XML parser then DOM methods
#       to step through and output the results, but hey, its simple,
#       has no dependencies, and works in many cases just fine.
#
#   (2) watch file bloat, the output will be considerable bigger than
#       the input XML file
#
#   (3) have tested the output in Mozilla, Firefox, IE6, Netscape4.7
#       Opera7.54 and seemed to render OK
# -----------------------------------------------------------------------------------------------

unless (@ARGV > 0) {
    die "XMLPretty: converts XML to a colorised browser friendly HTML\nSyntax: $0 <fileToPretty.xml>";
}

# set Mozilla-like XML rendering colours/styles.
my $style = 'Mozilla';
my $styleTag = 'color: #990099; font-weight: bold;';
my $styleAttrib = 'color: #000000; font-weight: bold;';
my $styleAttribVal = 'color: #0000ff;';
my $styleBracket = 'color: #0000CC;';
my $styleBody = 'color: #000000; font-size: 0.95em;';
my $styleComment = 'color: #006600; font-style: italic;';
my $stylePI = 'color: #990099; ';

# override if you want the Microsoft/IE XML rendering colours
if ($style ne 'Mozilla') {
    $styleTag = 'color: #660000;';
    $styleAttrib = 'color: #ee0000;';
    $styleBracket = 'color: #0000CC;';
    $styleAttribVal = 'color: #000000;';
    $styleBody = 'color: #000000; font-family: courier, sans-serif; font-size: 0.9em;';
    $styleComment = 'color: #666666; ';
    $stylePI = 'color: #006633; ';
}

my $tabSize = 2;
my $fileName = shift;
open INFILE, "< $fileName";

print "<div style='$styleBody'>";
while (my $line = <INFILE>) {

    # replace left and right hand '>'s
    $line =~ s/>/&gt;/g;
    $line =~ s/</&lt;/g;

    # check if a processing instruction
    if ($line =~ /^&lt;\?xml(.+)\?&gt;$/) {
        print "<br><span style='$stylePI'>" . $line . '</span>';
        next;   # dont bother with the other checks for PI's
    }

    # check if a comment
    if ($line =~ /\s*&lt;!--(.+)--&gt;\s*/) {
        print "<br><span style='$styleComment'>&lt;!-- $1 --&gt;</span>";
        next;   # dont bother with the other checks for comments
    }
    
    # attriubute names,  look for name=" or name=' or name = ' etc
    $line =~ s/([a-z0-9_:\-]+)\s*=\s*('|"){1}/<span style='$styleAttrib'>$1<\/span>=$2/ig;

    # attriubute values
    $line =~ s/='([\w\/\[\]\@="]+)'/<span style='$styleBracket'>='<\/span><span style='$styleAttribVal'>$1<\/span><span style='$styleBracket'>'<\/span>/ig;
    $line =~ s/="([\w\/\[\]\@=']+)"/<span style='$styleBracket'>="<\/span><span style='$styleAttribVal'>$1<\/span><span style='$styleBracket'>"<\/span>/ig;

    # opening tags,   look for '<' followed by a tagname
    $line =~ s/&lt;([a-z0-9_:\-]+)/&lt;<span style='$styleTag'>$1<\/span>/ig;

    # closing tags,    look for '<' followed by a tagname
    $line =~ s/([a-z0-9_:\-]+)&gt;/<span style='$styleTag'>$1<\/span>&gt;/ig;

    $line =~ s/("|')\s+/$1/g;   # remove space after quotes
    $line =~ s/(\s){$tabSize}/&nbsp;&nbsp;/g;   # pad/indent stuff
    $line =~ s/^\s+/&nbsp;/g;   # remove any leading whitespace

    # IE colours brackets, so <span> them as well
    $line =~ s/&lt;/<span style='$styleBracket'>&lt;<\/span>/g;
    $line =~ s/&gt;/<span style='$styleBracket'>&gt;<\/span>/g;

    print "<br>" . $line;
}
print "</div>";
close INFILE