Update headers and omit CVS Ids.

[yaz-moved-to-github.git] / src / charconv.tcl

diff --git a/src/charconv.tcl b/src/charconv.tcl
old mode 100755 (executable)
new mode 100644 (file)
index b509a8d..4412c2a
--- 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 1.17 2006-08-30 20:40:18 adam Exp $
+#!/usr/bin/tclsh
 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@ -39,16 +35,16 @@ proc preamble_trie {ofilehandle ifiles ofile} {
     "
     puts $f {
         static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp,
-                                    size_t inbytesleft, size_t *no_read, int *combining)
+                                    size_t inbytesleft, size_t *no_read, int *combining, unsigned mask, int boffset)
         {
             struct yaz_iconv_trie *t = (ptr > 0) ? ptrs[ptr-1] : 0;
             if (!t || inbytesleft < 1)
                 return 0;
             if (t->dir)
             {
-                size_t ch = inp[0] & 0xff;
+                size_t ch = (inp[0] & mask) + boffset;
                 unsigned long code =
-                lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining);
+                lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining, mask, boffset);
                 if (code)
                 {
                     (*no_read)++;
@@ -70,7 +66,13 @@ proc preamble_trie {ofilehandle ifiles ofile} {
                     size_t len = strlen(flat->from);
                     if (len <= inbytesleft)
                     {
-                        if (memcmp(flat->from, inp, len) == 0)
+                        size_t i;
+                       for (i = 0; i < len; i++)
+                       {
+                           if (((unsigned char *) flat->from)[i] != (inp[i] & mask) + boffset)
+                               break;
+                       }
+                        if (i == len)
                         {
                             *no_read = len;
                            *combining = flat->combining;
@@ -147,7 +149,18 @@ proc ins_trie_r {from to combining codename this} {
         set trie($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 +261,11 @@ 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;
@@ -268,34 +281,26 @@ 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} {
@@ -305,6 +310,7 @@ 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 $i+1]]
@@ -314,10 +320,20 @@ proc readfile {fname ofilehandle prefix omits reverse} {
                    unset hex
                }
            }
+           if {$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]} {
@@ -338,6 +354,8 @@ 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