#!/bin/sh
#\
exec tclsh8.0 "$0" "$@"

lappend auto_path [file dirname [info script]]

# Parse XML document and output normalised form

# Process arguments

set filename [lindex $argv 0]
array set options {-parser tclexpat}
catch {array set options [lrange $argv 1 end]}

# Setup procedures to output normalised document

proc elementstart {name attrs} {
    global out
    set normattrs {}
    if {[string length $attrs]} {
	foreach {name value} $attrs {
	    append normattrs \ $name="$value"
	}
    }
    catch {puts -nonewline $out "<$name$normattrs>"}
}

proc elementend {name} {
    global out
    catch {puts -nonewline $out "</$name>"}
}

proc pcdata pcdata {
    global out
    catch {puts -nonewline $out [encodeData $pcdata]}
}

proc encodeData pcdata {
    regsub -all {([^ -~])} \
	$pcdata {[encodeChar {\1}]} pcdata
    return [subst $pcdata]
}

proc encodeChar ch {
    scan $ch %c hex
    return [format {&#%d;} $hex]
}

switch $options(-parser) {
    tclexpat {
        package require expat
        set parser {xmlparser parse}
	set args {}
	if {[info exists options(-o)]} {
	    expat xmlparser -final yes \
                 -elementstartcommand elementstart \
	         -elementendcommand elementend \
		 -characterdatacommand pcdata
	} else {
            expat xmlparser -final yes
        }
    }
    tclxml {
        package require xml
	if {[info exists options(-o)]} {
	    set args {-elementstartcommand elementstart \
		-elementendcommand elementend \
		-characterdatacommand pcdata}
	} else {
	    set args {}
	}
        set parser ::xml::parseEvent
    }
}

# Get input data
set ch [open $filename]
set data [read $ch]
close $ch

# Setup output file
catch {set out [open $options(-o) w]}

if {[catch {eval $parser [list $data] $args} err]} {
    puts stderr $err
}

catch {close $out}