#!/bin/sh
-# the next line restats using tclsh \
-exec tclsh "$0" "$@"
+# 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 1.8 2005-03-16 21:26:37 adam Exp $
+# $Id: charconv.tcl,v 1.17 2006-08-30 20:40:18 adam Exp $
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
exit 1
}
-proc preamble_trie {ofilehandle} {
+proc preamble_trie {ofilehandle ifiles ofile} {
set f $ofilehandle
set totype {unsigned }
+ puts $f "/** \\file $ofile"
+ puts $f " \\brief Character conversion, generated from [lindex $ifiles 0]"
+ puts $f ""
+ puts $f " Generated automatically by charconv.tcl"
+ puts $f "*/"
puts $f "\#include <string.h>"
puts $f "
struct yaz_iconv_trie_flat {
$totype to : 24;
};
struct yaz_iconv_trie_dir {
- short ptr : 15;
+ int ptr : 15;
unsigned combining : 1;
$totype to : 24;
};
static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp,
size_t inbytesleft, size_t *no_read, int *combining)
{
- struct yaz_iconv_trie *t = (ptr >= 0) ? ptrs[ptr] : 0;
+ struct yaz_iconv_trie *t = (ptr > 0) ? ptrs[ptr-1] : 0;
if (!t || inbytesleft < 1)
return 0;
if (t->dir)
}
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
+ }
}
}
}
set ch [format %02X $i]
set null 1
if {[info exist trie($this,ptr,$ch)]} {
- puts -nonewline $f "$trie($this,ptr,$ch), "
+ puts -nonewline $f "[expr $trie($this,ptr,$ch)+1], "
set null 0
} else {
- puts -nonewline $f "-1, "
+ puts -nonewline $f "0, "
}
if {[info exist trie($this,combining,$ch)]} {
puts -nonewline $f "$trie($this,combining,$ch), "
{
unsigned long code;
- code = lookup($trie(prefix)ptrs, 0, inp, inbytesleft, no_read, combining);
+ code = lookup($trie(prefix)ptrs, 1, inp, inbytesleft, no_read, combining);
if (!code)
{
*no_read = 1;
- code = *inp;
}
return code;
}
"
}
-proc readfile {fname ofilehandle prefix omits} {
+proc readfile {fname ofilehandle prefix omits reverse} {
global trie
set marc_lines 0
set ucs_lines 0
+ set utf_lines 0
set codename_lines 0
set lineno 0
set f [open $fname r]
}
} elseif {[regexp {</code>} $line s]} {
if {[string length $ucs]} {
- for {set i 0} {$i < [string length $marc]} {incr i 2} {
- lappend hex [string range $marc $i [expr $i+1]]
+ if {$reverse} {
+ for {set i 0} {$i < [string length $utf]} {incr i 2} {
+ lappend hex [string range $utf $i [expr $i+1]]
+ }
+ # 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]]
+ }
+ # puts "ins_trie $hex $ucs"
+ ins_trie $hex $ucs $combining $codename
+ unset hex
}
- # puts "ins_trie $hex $ucs"
- ins_trie $hex $ucs $combining $codename
- unset hex
}
set marc {}
set uni {}
set combining 1
} elseif {[regexp {<ucs>([0-9A-Fa-f]*)</ucs>} $line s ucs]} {
incr ucs_lines
+ } elseif {[regexp {<utf-8>([0-9A-Fa-f]*)</utf-8>} $line s utf]} {
+ incr utf_lines
}
}
close $f
set ifile {}
set ofile out.c
set prefix {c}
+set reverse_map 0
# Parse command line
set l [llength $argv]
set i 0
}
lappend omits $arg
}
+ -r {
+ set reverse_map 1
+ }
default {
lappend ifiles $arg
}
}
set ofilehandle [open $ofile w]
-preamble_trie $ofilehandle
+preamble_trie $ofilehandle $ifiles $ofile
foreach ifile $ifiles {
- readfile $ifile $ofilehandle $prefix $omits
+ readfile $ifile $ofilehandle $prefix $omits $reverse_map
}
close $ofilehandle