X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcharconv.tcl;h=5732f8ddac6bea05a6f3b90ad79b0d34edebd27d;hb=2dbe1df5facb92bac31b5da0ca2e2e8ee6f9b358;hp=0f1ddc283095e0dc77ccb4ef1f0ec49b568c053a;hpb=00abdd969f52b04d5d53e4645c30596f4040060b;p=yaz-moved-to-github.git

diff --git a/src/charconv.tcl b/src/charconv.tcl
index 0f1ddc2..5732f8d 100755
--- a/src/charconv.tcl
+++ b/src/charconv.tcl
@@ -1,8 +1,5 @@
-#!/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 1.16 2006-05-23 09:43:37 mike 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 ... }
@@ -147,7 +144,18 @@ 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)} {
@@ -165,9 +173,11 @@ proc ins_trie_r {from to combining codename this} {
             }
             ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
         } else {
-            set trie($this,to,$ch) $to
-            set trie($this,combining,$ch) $combining
-            set trie($this,codename,$ch) $codename
+	    if {![info exist trie($this,to,$ch)]} {
+                set trie($this,to,$ch) $to
+                set trie($this,combining,$ch) $combining
+                set trie($this,codename,$ch) $codename
+	    }
         }
     }
 }
@@ -266,34 +276,26 @@ proc readfile {fname ofilehandle prefix omits reverse} {
     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} {
@@ -303,6 +305,7 @@ proc readfile {fname ofilehandle prefix omits 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 $i+1]]
@@ -312,10 +315,20 @@ proc readfile {fname ofilehandle prefix omits reverse} {
 		    unset hex
 		}
 	    }
+	    if {$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]} {
@@ -336,6 +349,8 @@ proc readfile {fname ofilehandle prefix omits reverse} {
 	    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