From: Adam Dickmeiss <adam@indexdata.dk>
Date: Sat, 22 Sep 2007 18:55:02 +0000 (+0000)
Subject: Remove duplicates (already existing) in flat entries for MARC-8
X-Git-Tag: YAZ.3.0.15.snap.5.oct~8
X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=commitdiff_plain;h=e77f334b9b8b6d0ba01b3ede0f58b99ac561dd2d

Remove duplicates (already existing) in flat entries for MARC-8
tries. Previously only duplicates in index tries were treated.
This ensures that only first sequence (of multiple) is used for
conversion and takes precedence. For example utf E38080 is
converted then converted to marc8 212321 (not marc8 212320).
---

diff --git a/src/charconv.tcl b/src/charconv.tcl
index dc494b1..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 1.19 2007-09-17 19:18:27 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)} {