#!/bin/sh
# Copyright for the Uniform Repository Service (c) 1995 - 2004,
# by Gerald Banon. All rights reserved.
# Version 2.1
# parseXml.tcl \
exec ../../../../../\
dpi.inpe.br/banon/1997/10.21.17.24\
/doc/tk8.0/unix/wish "$0" ${1+"$@"}

set example {
<?xml version="1.0" encoding="ISO-8859-1" ?>
<t1 a1="1 2">
 <t2 a2="2 4" b2="3" />
 <t2 a2="xx" b2="aa" />
 <t3 a3="3 4" b3="9">
  <t5 a5="vv" />
 </t3>
 <t3 a3="3 6" b3="8">
  <t5 a5="vvcc" />
 </t3>
 <t4>abc</t4>
</t1>
}

source cgi/mirrorfind-.tcl 	;# Load
source utilities1.tcl	;# Store
# Load C:/tmp/curriculo.xml example

# puts $example

if [info exists array] {unset array}

proc ProcessTag {parentName tag} {
	global array
	global tagNumber
	if [regexp {([^ ]*) (.*)} $tag m tagName attributePart] {
		regsub -all {=} $attributePart { } attributeList
		array set attributeArray $attributeList
		foreach attributeName [array names attributeArray] {
			if {[string compare {} $parentName] == 0} {
				set array($tagName|$tagNumber|,$attributeName) $attributeArray($attributeName)
			} else {
				set array($parentName.$tagName|$tagNumber|,$attributeName) $attributeArray($attributeName)
			}
		}
		return $tagName
	} else {
		return $tag
	}
}

proc ParseXml {parentName xml} {
	global array
	global tagNumber
	if [expr [info level] == 1] {set tagNumber 0}
	if [expr [info level] > 99] {return}
# puts [list level: [info level]]
	set xml [string trim $xml "\n 	"]
	regsub {<\?[^?]*\?>} $xml {} rest
	set i 1
	while {$i < 99 && [string compare {} $rest] != 0} {
# puts [list i: $i]
		if [regexp {<([^>]*)>(.*)} $rest m tag rest2] {
			incr tagNumber
			if [regsub {/$} $tag {} tag] {
				ProcessTag $parentName $tag
				set rest $rest2
			} else {
				set tagName [ProcessTag $parentName $tag]
# if [expr [info level] == 2] {return}
#				regexp "(.*)</$tagName>(.*)" $rest2 m tagValue rest	;# time consuming
#				regexp "(.*)</$tagName>" $rest2 m tagValue
# if [expr [info level] == 3 && $i == 2] {puts $rest2; puts $tagName}
#				regexp "</$tagName>(.*)" $rest2 m rest
				regexp -indices "</$tagName>(.*)" $rest2 m rest
#				regsub "$m$" $rest2 {} tagValue	;# doesn't work - tcl aborts
				set tagValue [string range $rest2 0 [expr [lindex $m 0] - 1]]
				set rest [string range $rest2 [lindex $rest 0] [lindex $rest 1]]
				if {[string compare {} $parentName] == 0} {
					ParseXml $tagName|$tagNumber| $tagValue
				} else {
					ParseXml $parentName.$tagName|$tagNumber| $tagValue
				}
			}
			set rest [string trim $rest "\n 	"]
		} else {
			set array($parentName) $rest
			set rest {}
		}
		incr i
	}
}

set errorInfo {}
catch {ParseXml {} $example}
puts $errorInfo
set list [array get array]
# Store list C:/tmp/curriculoList
# puts [array get array CURRICULO-VITAE|1|.PRODUCAO-BIBLIOGRAFICA|68|.LIVROS-E-CAPITULOS|479|.LIVROS-PUBLICADOS-OU-ORGANIZADOS|480|.LIVRO-PUBLICADO-OU-ORGANIZADO|526|*]
 puts [array get array]
puts done