-#!/bin/sh
-# the next line restarts using tclsh \
-if [ -f /usr/local/bin/tclsh8.4 ]; then exec tclsh8.4 "$0" "$@"; else exec tclsh "$0" "$@"; fi
-#
-# $Id: charconv.tcl,v 1.17 2006-08-30 20:40:18 adam Exp $
+#!/usr/bin/tclsh
+# $Id: charconv.tcl,v 1.21 2008-01-06 13:02:48 adam Exp $
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
set trie($this,type) f
}
if {$trie($this,type) == "f"} {
- lappend trie($this,content) [list $from $to $combining $codename]
+ set dup 0
+ if {[info exists trie($this,content)]} {
+ foreach e $trie($this,content) {
+ set efrom [lindex $e 0]
+ if { $efrom == $from } {
+ set dup 1
+ }
+ }
+ }
+ if { $dup == 0 } {
+ lappend trie($this,content) [list $from $to $combining $codename]
+ }
# split ?
if {[llength $trie($this,content)] > $trie(split)} {
set marc_lines 0
set ucs_lines 0
set utf_lines 0
+ set altutf_lines 0
set codename_lines 0
set lineno 0
set f [open $fname r]
set tablenumber x
set combining 0
set codename {}
+ set altutf {}
while {1} {
incr lineno
set cnt [gets $f line]
if {$cnt < 0} {
break
}
- if {[regexp {<entitymap>} $line s]} {
- reset_trie
- set trie(prefix) "${prefix}"
- } elseif {[regexp {</entitymap>} $line s]} {
+ if {[regexp {</characterSet>} $line s]} {
dump_trie $ofilehandle
- } elseif {[regexp {<character hex="([^\"]*)".*<unientity>([0-9A-Fa-f]*)</unientity>} $line s hex ucs]} {
- ins_trie $hex $ucs $combining {}
- unset hex
- } elseif {[regexp {<codeTable .*number="([0-9]+)"} $line s tablenumber]} {
+ } elseif {[regexp {<characterSet .*ISOcode="([0-9A-Fa-f]+)"} $line s tablenumber]} {
reset_trie
set trie(prefix) "${prefix}_$tablenumber"
set combining 0
- } elseif {[regexp {</codeTable>} $line s]} {
- if {[lsearch $omits $tablenumber] == -1} {
- dump_trie $ofilehandle
- }
} elseif {[regexp {</code>} $line s]} {
if {[string length $ucs]} {
if {$reverse} {
# puts "ins_trie $hex $marc
ins_trie $hex $marc $combining $codename
unset hex
+
} else {
for {set i 0} {$i < [string length $marc]} {incr i 2} {
lappend hex [string range $marc $i [expr $i+1]]
unset hex
}
}
+ if {$reverse && [string length $marc]} {
+ for {set i 0} {$i < [string length $altutf]} {incr i 2} {
+ lappend hex [string range $altutf $i [expr $i+1]]
+ }
+ if {[info exists hex]} {
+ ins_trie $hex $marc $combining $codename
+ unset hex
+ }
+ }
set marc {}
set uni {}
set codename {}
set combining 0
+ set altutf {}
} elseif {[regexp {<marc>([0-9A-Fa-f]*)</marc>} $line s marc]} {
incr marc_lines
} elseif {[regexp {<name>(.*)</name>} $line s codename]} {
incr ucs_lines
} elseif {[regexp {<utf-8>([0-9A-Fa-f]*)</utf-8>} $line s utf]} {
incr utf_lines
+ } elseif {[regexp {<altutf-8>([0-9A-Fa-f]*)</altutf-8>} $line s altutf]} {
+ incr altutf_lines
}
}
close $f