projects
/
yaz-moved-to-github.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Another fix in extraRecordData handling
[yaz-moved-to-github.git]
/
src
/
charconv.tcl
diff --git
a/src/charconv.tcl
b/src/charconv.tcl
index
8ce0768
..
95a18a3
100755
(executable)
--- a/
src/charconv.tcl
+++ b/
src/charconv.tcl
@@
-2,7
+2,7
@@
# the next line restats using tclsh \
exec tclsh "$0" "$@"
#
# the next line restats using tclsh \
exec tclsh "$0" "$@"
#
-# $Id: charconv.tcl,v 1.3 2004-03-16 13:12:42 adam Exp $
+# $Id: charconv.tcl,v 1.10 2005-11-09 17:48:11 adam Exp $
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
proc usage {} {
puts {charconv.tcl: [-p prefix] [-s split] [-o ofile] file ... }
@@
-12,17
+12,19
@@
proc usage {} {
proc preamble_trie {ofilehandle} {
set f $ofilehandle
proc preamble_trie {ofilehandle} {
set f $ofilehandle
- set totype {unsigned short}
+ set totype {unsigned }
puts $f "\#include <string.h>"
puts $f "
struct yaz_iconv_trie_flat {
puts $f "\#include <string.h>"
puts $f "
struct yaz_iconv_trie_flat {
- char *from;
- $totype to;
+ char from\[6\];
+ unsigned combining : 1;
+ $totype to : 24;
};
struct yaz_iconv_trie_dir {
};
struct yaz_iconv_trie_dir {
- struct yaz_iconv_trie *ptr;
- $totype to;
+ int ptr : 15;
+ unsigned combining : 1;
+ $totype to : 24;
};
struct yaz_iconv_trie {
};
struct yaz_iconv_trie {
@@
-31,16
+33,17
@@
proc preamble_trie {ofilehandle} {
};
"
puts $f {
};
"
puts $f {
- static unsigned long lookup(struct yaz_iconv_trie *t, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
+ 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;
if (!t || inbytesleft < 1)
if (!t || inbytesleft < 1)
- return 0;
+ return 0;
if (t->dir)
{
size_t ch = inp[0] & 0xff;
unsigned long code =
if (t->dir)
{
size_t ch = inp[0] & 0xff;
unsigned long code =
- lookup(t->dir[ch].ptr, inp+1, inbytesleft-1, no_read);
+ lookup(ptrs, t->dir[ch].ptr, inp+1, inbytesleft-1, no_read, combining);
if (code)
{
(*no_read)++;
if (code)
{
(*no_read)++;
@@
-49,6
+52,7
@@
proc preamble_trie {ofilehandle} {
if (t->dir[ch].to)
{
code = t->dir[ch].to;
if (t->dir[ch].to)
{
code = t->dir[ch].to;
+ *combining = t->dir[ch].combining;
*no_read = 1;
return code;
}
*no_read = 1;
return code;
}
@@
-56,7
+60,7
@@
proc preamble_trie {ofilehandle} {
else
{
struct yaz_iconv_trie_flat *flat = t->flat;
else
{
struct yaz_iconv_trie_flat *flat = t->flat;
- while (flat->from)
+ while (flat->to)
{
size_t len = strlen(flat->from);
if (len <= inbytesleft)
{
size_t len = strlen(flat->from);
if (len <= inbytesleft)
@@
-64,6
+68,7
@@
proc preamble_trie {ofilehandle} {
if (memcmp(flat->from, inp, len) == 0)
{
*no_read = len;
if (memcmp(flat->from, inp, len) == 0)
{
*no_read = len;
+ *combining = flat->combining;
return flat->to;
}
}
return flat->to;
}
}
@@
-72,7
+77,6
@@
proc preamble_trie {ofilehandle} {
}
return 0;
}
}
return 0;
}
-
}
}
}
}
@@
-86,11
+90,11
@@
proc reset_trie {} {
set trie(no) 1
set trie(size) 0
set trie(max) 0
set trie(no) 1
set trie(size) 0
set trie(max) 0
- set trie(split) 40
+ set trie(split) 50
set trie(prefix) {}
}
set trie(prefix) {}
}
-proc ins_trie {from to} {
+proc ins_trie {from to combining codename} {
global trie
if {![info exists trie(no)]} {
set trie(no) 1
global trie
if {![info exists trie(no)]} {
set trie(no) 1
@@
-101,7
+105,7
@@
proc ins_trie {from to} {
set trie(max) $to
}
incr trie(size)
set trie(max) $to
}
incr trie(size)
- ins_trie_r [split $from] $to 0
+ ins_trie_r [split $from] $to $combining $codename 0
}
proc split_trie {this} {
}
proc split_trie {this} {
@@
-110,6
+114,8
@@
proc split_trie {this} {
foreach e $trie($this,content) {
set from [lindex $e 0]
set to [lindex $e 1]
foreach e $trie($this,content) {
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 ch [lindex $from 0]
set rest [lrange $from 1 end]
@@
-119,27
+125,29
@@
proc split_trie {this} {
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
- ins_trie_r $rest $to $trie($this,ptr,$ch)
+ ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
} else {
set trie($this,to,$ch) $to
} else {
set trie($this,to,$ch) $to
+ set trie($this,combining,$ch) $combining
+ set trie($this,codename,$ch) $codename
}
}
set trie($this,content) missing
}
}
}
set trie($this,content) missing
}
-proc ins_trie_r {from to this} {
+proc ins_trie_r {from to combining codename this} {
global trie
if {![info exist trie($this,type)]} {
set trie($this,type) f
}
if {$trie($this,type) == "f"} {
global trie
if {![info exist trie($this,type)]} {
set trie($this,type) f
}
if {$trie($this,type) == "f"} {
- lappend trie($this,content) [list $from $to]
+ lappend trie($this,content) [list $from $to $combining $codename]
# split ?
if {[llength $trie($this,content)] > $trie(split)} {
split_trie $this
# split ?
if {[llength $trie($this,content)] > $trie(split)} {
split_trie $this
- return [ins_trie_r $from $to $this]
+ return [ins_trie_r $from $to $combining $codename $this]
}
} else {
set ch [lindex $from 0]
}
} else {
set ch [lindex $from 0]
@@
-150,9
+158,11
@@
proc ins_trie_r {from to this} {
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
set trie($this,ptr,$ch) $trie(no)
incr trie(no)
}
- ins_trie_r $rest $to $trie($this,ptr,$ch)
+ ins_trie_r $rest $to $combining $codename $trie($this,ptr,$ch)
} else {
set trie($this,to,$ch) $to
} else {
set trie($this,to,$ch) $to
+ set trie($this,combining,$ch) $combining
+ set trie($this,codename,$ch) $codename
}
}
}
}
}
}
@@
-174,10
+184,11
@@
proc dump_trie {ofilehandle} {
foreach d [lindex $m 0] {
puts -nonewline $f "\\x$d"
}
foreach d [lindex $m 0] {
puts -nonewline $f "\\x$d"
}
- puts -nonewline $f "\", 0x[lindex $m 1]"
- puts $f "\},"
+ puts -nonewline $f "\", [lindex $m 2], 0x[lindex $m 1]"
+ set v [lindex $m 3]
+ puts $f "\}, /* $v */"
}
}
- puts $f " \{0, 0\}"
+ puts $f " \{\"\", 0\}"
puts $f "\};"
puts $f "struct yaz_iconv_trie $trie(prefix)page${this} = \{"
puts $f " $trie(prefix)page${this}_flat, 0"
puts $f "\};"
puts $f "struct yaz_iconv_trie $trie(prefix)page${this} = \{"
puts $f " $trie(prefix)page${this}_flat, 0"
@@
-189,9
+200,14
@@
proc dump_trie {ofilehandle} {
set ch [format %02X $i]
set null 1
if {[info exist trie($this,ptr,$ch)]} {
set ch [format %02X $i]
set null 1
if {[info exist trie($this,ptr,$ch)]} {
- puts -nonewline $f "&$trie(prefix)page$trie($this,ptr,$ch), "
+ puts -nonewline $f "$trie($this,ptr,$ch), "
set null 0
} else {
set null 0
} else {
+ puts -nonewline $f "-1, "
+ }
+ if {[info exist trie($this,combining,$ch)]} {
+ puts -nonewline $f "$trie($this,combining,$ch), "
+ } else {
puts -nonewline $f "0, "
}
if {[info exist trie($this,to,$ch)]} {
puts -nonewline $f "0, "
}
if {[info exist trie($this,to,$ch)]} {
@@
-200,8
+216,9
@@
proc dump_trie {ofilehandle} {
} else {
puts -nonewline $f "0\}"
}
} 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 $f ","
}
if {$i < 255} {
puts $f ","
@@
-215,16
+232,23
@@
proc dump_trie {ofilehandle} {
puts $f "\};"
}
}
puts $f "\};"
}
}
+
+ puts $f "struct yaz_iconv_trie *$trie(prefix)ptrs \[\] = {"
+ for {set this 0} {$this < $trie(no)} {incr this} {
+ puts $f " &$trie(prefix)page$this,"
+ }
+ puts $f "0, };"
+ puts $f ""
+
puts $f "unsigned long yaz_$trie(prefix)_conv
puts $f "unsigned long yaz_$trie(prefix)_conv
- (unsigned char *inp, size_t inbytesleft, size_t *no_read)
+ (unsigned char *inp, size_t inbytesleft, size_t *no_read, int *combining)
{
unsigned long code;
{
unsigned long code;
- code = lookup(&$trie(prefix)page0, inp, inbytesleft, no_read);
+ code = lookup($trie(prefix)ptrs, 0, inp, inbytesleft, no_read, combining);
if (!code)
{
*no_read = 1;
if (!code)
{
*no_read = 1;
- code = *inp;
}
return code;
}
}
return code;
}
@@
-236,9
+260,12
@@
proc readfile {fname ofilehandle prefix omits} {
set marc_lines 0
set ucs_lines 0
set marc_lines 0
set ucs_lines 0
+ set codename_lines 0
set lineno 0
set f [open $fname r]
set tablenumber x
set lineno 0
set f [open $fname r]
set tablenumber x
+ set combining 0
+ set codename {}
while {1} {
incr lineno
set cnt [gets $f line]
while {1} {
incr lineno
set cnt [gets $f line]
@@
-248,16
+275,15
@@
proc readfile {fname ofilehandle prefix omits} {
if {[regexp {<entitymap>} $line s]} {
reset_trie
set trie(prefix) "${prefix}"
if {[regexp {<entitymap>} $line s]} {
reset_trie
set trie(prefix) "${prefix}"
- puts "new table $tablenumber"
} elseif {[regexp {</entitymap>} $line s]} {
dump_trie $ofilehandle
} elseif {[regexp {<character hex="([^\"]*)".*<unientity>([0-9A-Fa-f]*)</unientity>} $line s hex ucs]} {
} elseif {[regexp {</entitymap>} $line s]} {
dump_trie $ofilehandle
} elseif {[regexp {<character hex="([^\"]*)".*<unientity>([0-9A-Fa-f]*)</unientity>} $line s hex ucs]} {
- ins_trie $hex $ucs
+ ins_trie $hex $ucs $combining {}
unset hex
unset hex
- } elseif {[regexp {<codeTable number="([0-9]+)"} $line s tablenumber]} {
+ } elseif {[regexp {<codeTable .*number="([0-9]+)"} $line s tablenumber]} {
reset_trie
set trie(prefix) "${prefix}_$tablenumber"
reset_trie
set trie(prefix) "${prefix}_$tablenumber"
- puts "new table $tablenumber"
+ set combining 0
} elseif {[regexp {</codeTable>} $line s]} {
if {[lsearch $omits $tablenumber] == -1} {
dump_trie $ofilehandle
} elseif {[regexp {</codeTable>} $line s]} {
if {[lsearch $omits $tablenumber] == -1} {
dump_trie $ofilehandle
@@
-268,13
+294,29
@@
proc readfile {fname ofilehandle prefix omits} {
lappend hex [string range $marc $i [expr $i+1]]
}
# puts "ins_trie $hex $ucs"
lappend hex [string range $marc $i [expr $i+1]]
}
# puts "ins_trie $hex $ucs"
- ins_trie $hex $ucs
+ ins_trie $hex $ucs $combining $codename
unset hex
}
set marc {}
set uni {}
unset hex
}
set marc {}
set uni {}
+ set codename {}
+ set combining 0
} elseif {[regexp {<marc>([0-9A-Fa-f]*)</marc>} $line s marc]} {
incr marc_lines
} 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 {<ucs>([0-9A-Fa-f]*)</ucs>} $line s ucs]} {
incr ucs_lines
}
@@
-285,7
+327,6
@@
proc readfile {fname ofilehandle prefix omits} {
set verbose 0
set ifile {}
set ofile out.c
set verbose 0
set ifile {}
set ofile out.c
-set trie(split) 40
set prefix {c}
# Parse command line
set l [llength $argv]
set prefix {c}
# Parse command line
set l [llength $argv]