X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcharconv.tcl;h=dc494b1229bb32091a1f9a0578035b620205f4e5;hb=afae1117609712743c1109597d080a53a8dac5d4;hp=b509a8de57f7e08691dc22b5e386e7b8b512bece;hpb=6f17341a80d2c4558e4944c0731231994b3da8bc;p=yaz-moved-to-github.git
diff --git a/src/charconv.tcl b/src/charconv.tcl
index b509a8d..dc494b1 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 1.17 2006-08-30 20:40:18 adam Exp $
+# $Id: charconv.tcl,v 1.19 2007-09-17 19:18:27 adam Exp $
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@ -268,34 +268,26 @@ 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 {} $line s]} {
- reset_trie
- set trie(prefix) "${prefix}"
- } elseif {[regexp {} $line s]} {
+ if {[regexp {} $line s]} {
dump_trie $ofilehandle
- } elseif {[regexp {([0-9A-Fa-f]*)} $line s hex ucs]} {
- ins_trie $hex $ucs $combining {}
- unset hex
- } elseif {[regexp {} $line s]} {
- if {[lsearch $omits $tablenumber] == -1} {
- dump_trie $ofilehandle
- }
} elseif {[regexp {} $line s]} {
if {[string length $ucs]} {
if {$reverse} {
@@ -305,6 +297,7 @@ 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 $i+1]]
@@ -314,10 +307,20 @@ proc readfile {fname ofilehandle prefix omits reverse} {
unset hex
}
}
+ if {$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 {([0-9A-Fa-f]*)} $line s marc]} {
incr marc_lines
} elseif {[regexp {(.*)} $line s codename]} {
@@ -338,6 +341,8 @@ proc readfile {fname ofilehandle prefix omits reverse} {
incr ucs_lines
} elseif {[regexp {([0-9A-Fa-f]*)} $line s utf]} {
incr utf_lines
+ } elseif {[regexp {([0-9A-Fa-f]*)} $line s altutf]} {
+ incr altutf_lines
}
}
close $f