unicode: Store recursive decompositions directly in the tables.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This commit is contained in:
Alexandre Julliard 2020-02-04 10:14:28 +01:00
parent dba4bd756a
commit a646e4e6f2
3 changed files with 3064 additions and 2199 deletions

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff

View File

@ -416,7 +416,7 @@ my @direction_table = ();
my @decomp_table = ();
my @compose_table = ();
my @combining_class_table = ();
my @full_compatmap_table = ();
my @decomp_compat_table = ();
my $default_char;
my $default_wchar;
@ -446,8 +446,40 @@ sub open_data_file($$)
}
################################################################
# read in the defaults file
sub READ_DEFAULTS($)
# recursively get the decomposition for a character
sub get_decomposition($$);
sub get_decomposition($$)
{
my ($char, $table) = @_;
my @ret;
return $char unless defined ${$table}[$char];
foreach my $ch (@{${$table}[$char]})
{
push @ret, get_decomposition( $ch, $table );
}
return @ret;
}
################################################################
# recursively build decompositions
sub build_decompositions(@)
{
my @src = @_;
my @dst;
for (my $i = 0; $i < 65536; $i++)
{
next unless defined $src[$i];
my @decomp = get_decomposition( $i, \@src );
$dst[$i] = \@decomp;
}
return @dst;
}
################################################################
# read in the Unicode database files
sub load_data($)
{
my $filename = shift;
my $start;
@ -535,12 +567,8 @@ sub READ_DEFAULTS($)
if ($decomp =~ /^<([a-zA-Z]+)>\s+([0-9a-fA-F]+)/)
{
my @seq = ();
for my $ch (split /\s+/, (split /\s+/, $decomp, 2)[1])
{
push @seq, (hex $ch);
}
$full_compatmap_table[$src] = \@seq;
my @seq = map { hex $_; } (split /\s+/, (split /\s+/, $decomp, 2)[1]);
$decomp_compat_table[$src] = \@seq;
}
if ($decomp =~ /^<([a-zA-Z]+)>\s+([0-9a-fA-F]+)$/)
@ -580,7 +608,7 @@ sub READ_DEFAULTS($)
# store decomposition if it contains two chars
if ($decomp =~ /^([0-9a-fA-F]+)\s+([0-9a-fA-F]+)$/)
{
$decomp_table[$src] = [ hex $1, hex $2 ];
$decomp_table[$src] = $decomp_compat_table[$src] = [ hex $1, hex $2 ];
push @compose_table, [ hex $1, hex $2, $src ];
}
elsif ($decomp =~ /^([0-9a-fA-F]+)$/)
@ -588,7 +616,7 @@ sub READ_DEFAULTS($)
# Single char decomposition
if (hex $1 < 65536)
{
$decomp_table[$src] = [ hex $1 ];
$decomp_table[$src] = $decomp_compat_table[$src] = [ hex $1 ];
}
if (($src >= 0xf900 && $src < 0xfb00) || ($src >= 0xfe30 && $src < 0xfffd))
@ -2352,6 +2380,8 @@ sub dump_decompositions($@)
{
my ($name, @decomp) = @_;
@decomp = build_decompositions( @decomp );
# first determine all the 16-char subsets that contain something
my @filled = (0) x 4096;
@ -2457,74 +2487,45 @@ sub dump_decompose_table($$)
print OUTPUT "/* DO NOT EDIT!! */\n\n";
print OUTPUT "#include \"windef.h\"\n\n";
dump_decompositions( "table", @decomp_table );
dump_decompositions( $compat ? "decomp_table" : "table", @decomp_table );
if ($compat)
{
dump_decompositions( "compatmap_table", @full_compatmap_table );
dump_decompositions( "compatmap_table", @decomp_compat_table );
print OUTPUT <<"EOF";
#include "wine/unicode.h"
static const WCHAR *get_table_entry( const WCHAR *table, WCHAR ch, unsigned int *len )
{
unsigned short offset_offs = table[table[ch >> 8] + ((ch >> 4) & 0xf)] + (ch & 0xf);
unsigned short start = table[offset_offs];
unsigned short end = table[offset_offs + 1];
*len = end - start;
return table + start;
}
unsigned int DECLSPEC_HIDDEN wine_decompose( int flags, WCHAR ch, WCHAR *dst, unsigned int dstlen )
{
const WCHAR *ptr;
unsigned int res, len = 0, dst_pos = 0;
const WCHAR *table = (flags & WINE_DECOMPOSE_COMPAT) ? compatmap_table : decomp_table;
unsigned short offset = table[table[ch >> 8] + ((ch >> 4) & 0xf)] + (ch & 0xf);
unsigned short start = table[offset];
unsigned short end = table[offset + 1];
unsigned int len = end - start;
*dst = ch;
if (flags & WINE_DECOMPOSE_COMPAT) ptr = get_table_entry( compatmap_table, ch, &len );
if (!len) ptr = get_table_entry( table, ch, &len );
if (!len) return 1;
if (dstlen < len) return 0;
/* apply the decomposition recursively */
while (len--)
{
if (!(res = wine_decompose( flags, *ptr++, dst + dst_pos, dstlen - dst_pos - len ))) return 0;
dst_pos += res;
}
return dst_pos;
memcpy( dst, table + start, len * sizeof(WCHAR) );
return len;
}
EOF
}
else
{
print OUTPUT <<"EOF";
static const WCHAR *get_table_entry( const WCHAR *table, WCHAR ch, unsigned int *len )
{
unsigned short offset_offs = table[table[ch >> 8] + ((ch >> 4) & 0xf)] + (ch & 0xf);
unsigned short start = table[offset_offs];
unsigned short end = table[offset_offs + 1];
*len = end - start;
return table + start;
}
unsigned int DECLSPEC_HIDDEN wine_decompose( int flags, WCHAR ch, WCHAR *dst, unsigned int dstlen )
{
unsigned int res, len = 0, dst_pos = 0;
const WCHAR *ptr = get_table_entry( table, ch, &len );
unsigned short offset = table[table[ch >> 8] + ((ch >> 4) & 0xf)] + (ch & 0xf);
unsigned short start = table[offset];
unsigned short end = table[offset + 1];
unsigned int len = end - start;
*dst = ch;
if (!len) return 1;
if (dstlen < len) return 0;
/* apply the decomposition recursively */
while (len--)
{
if (!(res = wine_decompose( flags, *ptr++, dst + dst_pos, dstlen - dst_pos - len ))) return 0;
dst_pos += res;
}
return dst_pos;
memcpy( dst, table + start, len * sizeof(WCHAR) );
return len;
}
EOF
}
@ -2777,7 +2778,7 @@ sub REPLACE_IN_FILE($@)
# main routine
chdir ".." if -f "./make_unicode";
READ_DEFAULTS( $DEFAULTS );
load_data( $DEFAULTS );
dump_case_mappings( "libs/port/casemap.c" );
dump_sortkeys( "libs/port/collation.c" );
dump_sortkeys( "dlls/kernelbase/collation.c" );