X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcharconv.tcl;h=5500f595f8ff5faef6f4c4c2bdb0866ce7129cc4;hp=dc494b1229bb32091a1f9a0578035b620205f4e5;hb=a5795a2df2095d5a44b012a59426e7b906db3ec9;hpb=afae1117609712743c1109597d080a53a8dac5d4 diff --git a/src/charconv.tcl b/src/charconv.tcl old mode 100755 new mode 100644 index dc494b1..5500f59 --- 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 1.19 2007-09-17 19:18:27 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 $f { static unsigned long lookup(struct yaz_iconv_trie **ptrs, int ptr, unsigned char *inp, - size_t inbytesleft, size_t *no_read, int *combining) + size_t inbytesleft, size_t *no_read, int *combining, unsigned mask, unsigned int boffset) { - struct yaz_iconv_trie *t = (ptr > 0) ? ptrs[ptr-1] : 0; - if (!t || inbytesleft < 1) + struct yaz_iconv_trie *t = ptrs[ptr-1]; + if (inbytesleft < 1) return 0; if (t->dir) { - size_t ch = inp[0] & 0xff; - unsigned long code = - lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining); - if (code) + size_t ch = (inp[0] & mask) + boffset; + unsigned long code; + if (t->dir[ch].ptr) { - (*no_read)++; - return code; - } + code = lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining, mask, boffset); + if (code) + { + (*no_read)++; + return code; + } + } if (t->dir[ch].to) { code = t->dir[ch].to; @@ -70,7 +69,13 @@ proc preamble_trie {ofilehandle ifiles ofile} { size_t len = strlen(flat->from); if (len <= inbytesleft) { - if (memcmp(flat->from, inp, len) == 0) + size_t i; + for (i = 0; i < len; i++) + { + if (((unsigned char *) flat->from)[i] != (inp[i] & mask) + boffset) + break; + } + if (i == len) { *no_read = len; *combining = flat->combining; @@ -147,7 +152,18 @@ proc ins_trie_r {from to combining codename this} { set trie($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)} { @@ -195,7 +211,7 @@ proc dump_trie {ofilehandle} { set v [lindex $m 3] puts $f "\}, /* $v */" } - puts $f " \{\"\", 0\}" + puts $f " \{\"\", 0, 0\}" puts $f "\};" puts $f "struct yaz_iconv_trie $trie(prefix)page${this} = \{" puts $f " $trie(prefix)page${this}_flat, 0" @@ -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, unsigned 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; @@ -401,7 +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 { @@ -409,4 +425,6 @@ foreach ifile $ifiles { } close $ofilehandle +file rename -force ${ofile}.tmp ${ofile} +