X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Fcharconv.tcl;h=d57f2262d7bdab6764c8a7d96ce90194d0979fc4;hb=8b7d8579cda1772ab488813e3e639c6709948db1;hp=7876f9309434ed2180e276d8586a82423fb8a48b;hpb=60fa7941321b3d313bfa0d94e188835a35e93320;p=yaz-moved-to-github.git
diff --git a/src/charconv.tcl b/src/charconv.tcl
old mode 100755
new mode 100644
index 7876f93..d57f226
--- a/src/charconv.tcl
+++ b/src/charconv.tcl
@@ -1,8 +1,4 @@
-#!/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 1.18 2006-12-17 15:34:11 adam Exp $
+#!/usr/bin/tclsh
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@ -39,21 +35,24 @@ proc preamble_trie {ofilehandle ifiles ofile} {
"
puts $f {
static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp,
- size_t inbytesleft, size_t *no_read, int *combining)
+ size_t inbytesleft, size_t *no_read, int *combining, unsigned mask, int boffset)
{
- struct yaz_iconv_trie *t = (ptr > 0) ? ptrs[ptr-1] : 0;
- if (!t || inbytesleft < 1)
+ struct yaz_iconv_trie *t = ptrs[ptr-1];
+ if (inbytesleft < 1)
return 0;
if (t->dir)
{
- size_t ch = inp[0] & 0xff;
- unsigned long code =
- lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining);
- if (code)
+ size_t ch = (inp[0] & mask) + boffset;
+ unsigned long code;
+ if (t->dir[ch].ptr)
{
- (*no_read)++;
- return code;
- }
+ code = lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining, mask, boffset);
+ if (code)
+ {
+ (*no_read)++;
+ return code;
+ }
+ }
if (t->dir[ch].to)
{
code = t->dir[ch].to;
@@ -70,7 +69,13 @@ proc preamble_trie {ofilehandle ifiles ofile} {
size_t len = strlen(flat->from);
if (len <= inbytesleft)
{
- if (memcmp(flat->from, inp, len) == 0)
+ size_t i;
+ for (i = 0; i < len; i++)
+ {
+ if (((unsigned char *) flat->from)[i] != (inp[i] & mask) + boffset)
+ break;
+ }
+ if (i == len)
{
*no_read = len;
*combining = flat->combining;
@@ -147,7 +152,18 @@ proc ins_trie_r {from to combining codename this} {
set trie($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)} {
@@ -248,11 +264,11 @@ proc dump_trie {ofilehandle} {
puts $f ""
puts $f "unsigned long yaz_$trie(prefix)_conv
- (unsigned char *inp, size_t inbytesleft, size_t *no_read, int *combining)
+ (unsigned char *inp, size_t inbytesleft, size_t *no_read, int *combining, unsigned mask, int boffset)
{
unsigned long code;
- code = lookup($trie(prefix)ptrs, 1, inp, inbytesleft, no_read, combining);
+ code = lookup($trie(prefix)ptrs, 1, inp, inbytesleft, no_read, combining, mask, boffset);
if (!code)
{
*no_read = 1;
@@ -282,22 +298,12 @@ proc readfile {fname ofilehandle prefix omits reverse} {
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} {
@@ -411,7 +417,7 @@ if {![info exists ifiles]} {
usage
}
-set ofilehandle [open $ofile w]
+set ofilehandle [open ${ofile}.tmp w]
preamble_trie $ofilehandle $ifiles $ofile
foreach ifile $ifiles {
@@ -419,4 +425,6 @@ foreach ifile $ifiles {
}
close $ofilehandle
+file rename -force ${ofile}.tmp ${ofile}
+