Commit | Line | Data |
ac5ea531 |
1 | #!perl |
2 | # |
3 | # This script generates "unfcan.h", "unfcpt.h", "unfcmb.h", |
4 | # "unfcmp.h", and "unfexc.h" |
5 | # from CombiningClass.pl, Decomposition.pl, CompExcl.txt |
6 | # in lib/unicore or unicode directory |
7 | # for Unicode::Normalize.xs. (cf. Makefile.PL) |
8 | # |
9 | use 5.006; |
10 | use strict; |
11 | use warnings; |
12 | use Carp; |
13 | |
14 | our $PACKAGE = 'Unicode::Normalize, mkheader'; |
15 | |
16 | our $Combin = do "unicore/CombiningClass.pl" |
17 | || do "unicode/CombiningClass.pl" |
18 | || croak "$PACKAGE: CombiningClass.pl not found"; |
19 | |
20 | our $Decomp = do "unicore/Decomposition.pl" |
21 | || do "unicode/Decomposition.pl" |
22 | || croak "$PACKAGE: Decomposition.pl not found"; |
23 | |
24 | our %Combin; # $codepoint => $number : combination class |
25 | our %Canon; # $codepoint => $hexstring : canonical decomp. |
26 | our %Compat; # $codepoint => $hexstring : compat. decomp. |
27 | our %Compos; # $string => $codepoint : composite |
28 | |
29 | our %Exclus; # $codepoint => 1 : composition exclusions |
30 | |
31 | { |
32 | my($f, $fh); |
33 | foreach my $d (@INC) { |
34 | use File::Spec; |
551b6b6f |
35 | $f = File::Spec->catfile($d, "unicore", "CompositionExclusions.txt"); |
ac5ea531 |
36 | last if open($fh, $f); |
37 | $f = File::Spec->catfile($d, "unicode", "CompExcl.txt"); |
38 | last if open($fh, $f); |
39 | $f = undef; |
40 | } |
41 | croak "$PACKAGE: CompExcl.txt not found in @INC" unless defined $f; |
42 | while(<$fh>) { |
43 | next if /^#/ or /^$/; |
44 | s/#.*//; |
45 | $Exclus{ hex($1) } =1 if /([0-9A-Fa-f]+)/; |
46 | } |
47 | close $fh; |
48 | } |
49 | |
50 | while($Combin =~ /(.+)/g) { |
51 | my @tab = split /\t/, $1; |
52 | my $ini = hex $tab[0]; |
53 | if($tab[1] eq '') { |
54 | $Combin{ $ini } = $tab[2]; |
55 | } else { |
56 | $Combin{ $_ } = $tab[2] foreach $ini .. hex($tab[1]); |
57 | } |
58 | } |
59 | |
60 | while($Decomp =~ /(.+)/g) { |
61 | my @tab = split /\t/, $1; |
62 | my $compat = $tab[2] =~ s/<[^>]+>//; |
63 | my $dec = [ _getHexArray($tab[2]) ]; # decomposition |
64 | my $com = pack('U*', @$dec); # composable sequence |
65 | my $ini = hex($tab[0]); |
66 | if($tab[1] eq '') { |
67 | $Compat{ $ini } = $dec; |
68 | if(! $compat) { |
69 | $Canon{ $ini } = $dec; |
70 | $Compos{ $com } = $ini if @$dec > 1; |
71 | } |
72 | } else { |
73 | foreach my $u ($ini .. hex($tab[1])){ |
74 | $Compat{ $u } = $dec; |
75 | if(! $compat){ |
76 | $Canon{ $u } = $dec; |
77 | $Compos{ $com } = $ini if @$dec > 1; |
78 | } |
79 | } |
80 | } |
81 | } |
82 | |
83 | # exhaustive decomposition |
84 | foreach my $key (keys %Canon) { |
85 | $Canon{$key} = [ getCanonList($key) ]; |
86 | } |
87 | |
88 | # exhaustive decomposition |
89 | foreach my $key (keys %Compat) { |
90 | $Compat{$key} = [ getCompatList($key) ]; |
91 | } |
92 | |
93 | sub getCanonList { |
94 | my @src = @_; |
95 | my @dec = map $Canon{$_} ? @{ $Canon{$_} } : $_, @src; |
96 | join(" ",@src) eq join(" ",@dec) ? @dec : getCanonList(@dec); |
97 | # condition @src == @dec is not ok. |
98 | } |
99 | |
100 | sub getCompatList { |
101 | my @src = @_; |
102 | my @dec = map $Compat{$_} ? @{ $Compat{$_} } : $_, @src; |
103 | join(" ",@src) eq join(" ",@dec) ? @dec : getCompatList(@dec); |
104 | # condition @src == @dec is not ok. |
105 | } |
106 | |
107 | sub _getHexArray { |
108 | my $str = shift; |
109 | map hex(), $str =~ /([0-9A-Fa-f]+)/g; |
110 | } |
111 | |
112 | sub _U_stringify { |
113 | sprintf '"%s"', join '', |
d85850a7 |
114 | map sprintf("\\x%02x", $_), unpack 'C*', pack 'U*', @_; |
ac5ea531 |
115 | } |
116 | |
117 | foreach my $hash (\%Canon, \%Compat) { |
118 | foreach my $key (keys %$hash) { |
119 | $hash->{$key} = _U_stringify( @{ $hash->{$key} } ); |
120 | } |
121 | } |
122 | |
ac5ea531 |
123 | my $prefix = "UNF_"; |
124 | |
125 | my $structname = "${prefix}complist"; |
126 | |
127 | our (%Comp1st, %CompList); |
128 | |
129 | foreach(sort keys %Compos) { |
130 | my @a = unpack('U*', $_); |
131 | my $val = $Compos{$_}; |
132 | my $name = sprintf "${structname}_%06x", $a[0]; |
133 | $Comp1st{ $a[0] } = $name; |
134 | $CompList{ $name }{ $a[1] } = $val; |
ac5ea531 |
135 | } |
136 | |
137 | my $compinit = |
138 | "typedef struct { UV nextchar; UV composite; } $structname;\n\n"; |
139 | |
140 | foreach my $i (sort keys %CompList) { |
141 | $compinit .= "$structname $i [] = {\n"; |
142 | $compinit .= join ",\n", |
143 | map sprintf("\t{ %d, %d }", $_, $CompList{$i}{$_}), |
144 | sort {$a <=> $b } keys %{ $CompList{$i} }; |
145 | $compinit .= ",\n{0,0}\n};\n\n"; # with sentinel |
146 | } |
147 | |
148 | #################################### |
149 | |
150 | my @Exclus = sort {$a <=> $b} keys %Exclus; |
151 | |
152 | my $file = "unfexc.h"; |
153 | open FH, ">$file" or croak "$PACKAGE: $file can't be made"; |
154 | binmode FH; select FH; |
155 | |
2a204b45 |
156 | print "bool isExclusion (UV uv) \n{\nreturn\n\t"; |
ac5ea531 |
157 | |
158 | while(@Exclus) { |
159 | my $cur = shift @Exclus; |
160 | if(@Exclus && $cur + 1 == $Exclus[0]) { |
20d72259 |
161 | print "($cur <= uv && uv <= "; |
ac5ea531 |
162 | while(@Exclus && $cur + 1 == $Exclus[0]) { |
163 | $cur = shift @Exclus; |
164 | } |
20d72259 |
165 | print "$cur)"; |
ac5ea531 |
166 | print "\n\t|| " if @Exclus; |
167 | } else { |
168 | print "uv == $cur"; |
169 | print "\n\t|| " if @Exclus; |
170 | } |
171 | } |
172 | |
173 | print "\n\t? TRUE : FALSE;\n}\n\n"; |
174 | close FH; |
175 | |
176 | #################################### |
177 | |
178 | my @tripletable = ( |
179 | { |
180 | file => "unfcmb", |
181 | name => "combin", |
3164a1ca |
182 | type => "STDCHAR", |
ac5ea531 |
183 | hash => \%Combin, |
184 | null => 0, |
185 | }, |
186 | { |
187 | file => "unfcan", |
188 | name => "canon", |
189 | type => "char*", |
190 | hash => \%Canon, |
191 | null => "NULL", |
192 | }, |
193 | { |
194 | file => "unfcpt", |
195 | name => "compat", |
196 | type => "char*", |
197 | hash => \%Compat, |
198 | null => "NULL", |
199 | }, |
200 | { |
201 | file => "unfcmp", |
202 | name => "compos", |
203 | type => "$structname *", |
204 | hash => \%Comp1st, |
205 | null => "NULL", |
206 | init => $compinit, |
207 | }, |
208 | ); |
209 | |
210 | foreach my $tbl (@tripletable) { |
211 | my $file = "$tbl->{file}.h"; |
212 | my $head = "${prefix}$tbl->{name}"; |
213 | my $type = $tbl->{type}; |
214 | my $hash = $tbl->{hash}; |
215 | my $null = $tbl->{null}; |
216 | my $init = $tbl->{init}; |
217 | |
218 | open FH, ">$file" or croak "$PACKAGE: $file can't be made"; |
219 | binmode FH; select FH; |
220 | my %val; |
221 | |
222 | print FH << 'EOF'; |
223 | /* |
224 | * This file is auto-generated by mkheader. |
225 | * Any changes here will be lost! |
226 | */ |
227 | EOF |
228 | |
229 | print $init if defined $init; |
230 | |
231 | foreach my $uv (keys %$hash) { |
232 | my @c = unpack 'CCCC', pack 'N', $uv; |
233 | $val{ $c[1] }{ $c[2] }{ $c[3] } = $hash->{$uv}; |
234 | } |
235 | |
236 | foreach my $p (sort { $a <=> $b } keys %val) { |
237 | next if ! $val{ $p }; |
238 | for(my $r = 0; $r < 256; $r++){ |
239 | next if ! $val{ $p }{ $r }; |
240 | printf "$type ${head}_%02x_%02x [256] = {\n", $p, $r; |
241 | for(my $c = 0; $c < 256; $c++){ |
0904bc86 |
242 | print "\t", defined $val{$p}{$r}{$c} ? "($type)".$val{$p}{$r}{$c} : $null; |
ac5ea531 |
243 | print ',' if $c != 255; |
244 | print "\n" if $c % 8 == 7; |
245 | } |
246 | print "};\n\n"; |
247 | } |
248 | } |
249 | foreach my $p (sort { $a <=> $b } keys %val) { |
250 | next if ! $val{ $p }; |
251 | printf "$type* ${head}_%02x [256] = {\n", $p; |
252 | for(my $r = 0; $r < 256; $r++){ |
253 | print $val{ $p }{ $r } ? sprintf("${head}_%02x_%02x", $p, $r) : "NULL"; |
254 | print ',' if $r != 255; |
255 | print "\n" if $val{ $p }{ $r } || ($r+1) % 8 == 0; |
256 | } |
257 | print "};\n\n"; |
258 | } |
259 | print "$type** $head [] = {\n"; |
260 | for(my $p = 0; $p <= 0x10; $p++){ |
261 | print $val{ $p } ? sprintf("${head}_%02x", $p) : "NULL"; |
262 | print ',' if $p != 0x10; |
263 | print "\n"; |
264 | } |
265 | print "};\n\n"; |
266 | close FH; |
267 | } |
268 | |
269 | __END__ |