X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=src%2Fcharconv.tcl;h=b37b0d80bd08aaa756bf97ab49db17c01248588f;hb=cf30376bd9e43ad653c6292f275ee4dd10b70a6c;hp=7876f9309434ed2180e276d8586a82423fb8a48b;hpb=60fa7941321b3d313bfa0d94e188835a35e93320;p=yaz-moved-to-github.git

diff --git a/src/charconv.tcl b/src/charconv.tcl
index 7876f93..b37b0d8 100755
--- a/src/charconv.tcl
+++ b/src/charconv.tcl
@@ -2,7 +2,7 @@
 # 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 1.18 2006-12-17 15:34:11 adam Exp $
+# $Id: charconv.tcl,v 1.20 2007-09-22 18:55:02 adam Exp $
 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@ -147,7 +147,18 @@ proc ins_trie_r {from to combining codename this} {
         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)] > $trie(split)} {
@@ -282,22 +293,12 @@ proc readfile {fname ofilehandle prefix omits reverse} {
         if {$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} {