-#!/bin/sh
-# the next line restats using tclsh \
-exec tclsh "$0" "$@"
-#
-# $Id: charconv.tcl,v 1.7 2004-08-07 08:06:57 adam 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 ... }
exit 1
}
-proc preamble_trie {ofilehandle} {
+proc preamble_trie {ofilehandle ifiles ofile} {
set f $ofilehandle
- set totype {unsigned short}
+ 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 {
char from\[6\];
unsigned combining : 1;
- $totype to;
+ $totype to : 24;
};
struct yaz_iconv_trie_dir {
- short ptr : 15;
+ int ptr : 15;
unsigned combining : 1;
- $totype to;
+ $totype to : 24;
};
struct yaz_iconv_trie {
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)
set trie(prefix) {}
}
-proc ins_trie {from to combining} {
+proc ins_trie {from to combining codename} {
global trie
if {![info exists trie(no)]} {
set trie(no) 1
set trie(max) $to
}
incr trie(size)
- ins_trie_r [split $from] $to $combining 0
+ ins_trie_r [split $from] $to $combining $codename 0
}
proc split_trie {this} {
set from [lindex $e 0]
set to [lindex $e 1]
set combining [lindex $e 2]
+ set codename [lindex $e 3]
set ch [lindex $from 0]
set rest [lrange $from 1 end]
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
- ins_trie_r $rest $to $combining $trie($this,ptr,$ch)
+ 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) $codename
}
}
set trie($this,content) missing
}
-proc ins_trie_r {from to combining this} {
+proc ins_trie_r {from to combining codename this} {
global trie
if {![info exist trie($this,type)]} {
set trie($this,type) f
}
if {$trie($this,type) == "f"} {
- lappend trie($this,content) [list $from $to $combining]
+ set dup 0
+ if {[info exists trie($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)} {
split_trie $this
- return [ins_trie_r $from $to $combining $this]
+ return [ins_trie_r $from $to $combining $codename $this]
}
} else {
set ch [lindex $from 0]
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
- ins_trie_r $rest $to $combining $trie($this,ptr,$ch)
+ ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
} else {
- set trie($this,to,$ch) $to
- set trie($this,combining,$ch) $combining
+ if {![info exist trie($this,to,$ch)]} {
+ set trie($this,to,$ch) $to
+ set trie($this,combining,$ch) $combining
+ set trie($this,codename,$ch) $codename
+ }
}
}
}
puts -nonewline $f "\\x$d"
}
puts -nonewline $f "\", [lindex $m 2], 0x[lindex $m 1]"
- puts $f "\},"
+ set v [lindex $m 3]
+ puts $f "\}, /* $v */"
}
puts $f " \{\"\", 0\}"
puts $f "\};"
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), "
} else {
puts -nonewline $f "0\}"
}
- if {!$null} {
- puts -nonewline $f " /* $ch */"
+ if {[info exist trie($this,codename,$ch)]} {
+ set v $trie($this,codename,$ch)
+ puts -nonewline $f " /* $v */"
}
if {$i < 255} {
puts $f ","
{
unsigned long code;
- code = lookup($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 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 $ucs]} {
- for {set i 0} {$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 $codename
+ unset hex
+ }
+ }
+ if {$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
}
- # puts "ins_trie $hex $ucs"
- ins_trie $hex $ucs $combining
- 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]} {
+ incr codename_lines
+ } elseif {[regexp {<name>(.*)} $line s codename]} {
+ incr codename_lines
+ incr lineno
+ set cnt [gets $f line]
+ if {$cnt < 0} {
+ break
+ }
+ if {[regexp {(.*)</name>} $line s codename_ex]} {
+ set codename "${codename} ${codename_ex}"
+ }
} elseif {[regexp {<isCombining>true</isCombining>} $line s]} {
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
+ } elseif {[regexp {<altutf-8>([0-9A-Fa-f]*)</altutf-8>} $line s altutf]} {
+ incr altutf_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