Commit | Line | Data |
a0ed51b3 |
1 | package utf8; |
2 | |
3 | my $DEBUG = 0; |
4 | my $seq = "AAA0000"; |
5 | |
6 | sub DESTROY {} |
7 | |
8 | sub croak { require Carp; Carp::croak(@_) } |
9 | |
10 | sub SWASHNEW { |
11 | my ($class, $type, $list, $minbits, $none) = @_; |
12 | local $^D = 0 if $^D; |
13 | print STDERR "SWASHNEW @_\n" if $DEBUG; |
14 | my $extras; |
15 | my $bits; |
16 | |
17 | if ($type and ref ${"${class}::{$type}"} eq $class) { |
18 | warn qq/Found \${"${class}::{$type}"}\n/ if $DEBUG; |
19 | return ${"${class}::{$type}"}; # Already there... |
20 | } |
21 | |
22 | $type ||= $seq++; |
23 | |
24 | my $caller; |
25 | my $i = 0; |
26 | while (($caller = caller($i)) eq __PACKAGE__) { $i++ } |
55d7b906 |
27 | my $encoding = $enc{$caller} || "unicore"; |
a0ed51b3 |
28 | (my $file = $type) =~ s!::!/!g; |
f3a406cb |
29 | if ($file =~ /^(?:In|in|IN|iN)[- _]?(.+?)\s*$/) { # /i would cause recursion. |
a72deede |
30 | my $In = $1; |
9fdf68be |
31 | defined %utf8::In || do "$encoding/In.pl"; |
d9efae67 |
32 | my $prefix = substr(lc($In), 0, 3); |
33 | if (exists $utf8::InPat{$prefix}) { |
34 | for my $k (keys %{$utf8::InPat{$prefix}}) { |
35 | if ($In =~ /^$k$/i) { |
36 | $In = $utf8::InPat{$prefix}->{$k}; |
37 | if (exists $utf8::In{$In}) { |
38 | $file = "$encoding/In/$utf8::In{$In}"; |
39 | last; |
40 | } |
41 | } |
42 | } |
9fdf68be |
43 | } |
44 | } else { |
45 | $file =~ s#^(Is|To)([A-Z].*)#$1/$2#; |
46 | } |
886ba366 |
47 | |
48 | { |
0064a8a9 |
49 | $list ||= |
50 | ( exists &{"${caller}::${type}"} && |
51 | eval { $caller->$type() } ) |
886ba366 |
52 | || do "$file.pl" |
53 | || do "$encoding/$file.pl" |
54 | || do "$encoding/Is/${type}.pl" |
d9efae67 |
55 | || croak("Can't find Unicode character property \"$type\""); |
886ba366 |
56 | } |
a0ed51b3 |
57 | |
58 | $| = 1; |
59 | |
60 | if ($list) { |
61 | my @tmp = split(/^/m, $list); |
62 | my %seen; |
db376a24 |
63 | no warnings; |
a0ed51b3 |
64 | $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; |
65 | $list = join '', |
66 | sort { hex $a <=> hex $b } |
67 | grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right |
68 | } |
69 | |
70 | if ($none) { |
71 | my $hextra = sprintf "%04x", $none + 1; |
72 | $list =~ s/\tXXXX$/\t$hextra/mg; |
73 | } |
74 | |
75 | if ($minbits < 32) { |
76 | my $top = 0; |
77 | while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) { |
78 | my $min = hex $1; |
79 | my $max = hex(defined $2 ? $2 : $1); |
80 | my $val = hex(defined $3 ? $3 : ""); |
81 | $val += $max - $min if defined $3; |
82 | $top = $val if $val > $top; |
83 | } |
84 | $bits = |
85 | $top > 0xffff ? 32 : |
86 | $top > 0xff ? 16 : |
87 | $top > 1 ? 8 : 1 |
88 | } |
89 | $bits = $minbits if $bits < $minbits; |
90 | |
91 | my @extras; |
92 | for my $x ($extras) { |
93 | pos $x = 0; |
1fef36c7 |
94 | while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) { |
a0ed51b3 |
95 | my $char = $1; |
96 | my $name = $2; |
97 | # print STDERR "$1 => $2\n" if $DEBUG; |
98 | if ($char =~ /[-+!]/) { |
99 | my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really |
100 | my $subobj = $c->SWASHNEW($t, "", 0, 0, 0); |
101 | push @extras, $name => $subobj; |
102 | $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; |
103 | } |
104 | } |
105 | } |
106 | |
107 | print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if $DEBUG; |
108 | |
109 | ${"${class}::{$type}"} = bless { |
110 | TYPE => $type, |
111 | BITS => $bits, |
112 | EXTRAS => $extras, |
113 | LIST => $list, |
114 | NONE => $none, |
115 | @extras, |
116 | } => $class; |
117 | } |
118 | |
119 | # NOTE: utf8.c:swash_init() assumes entries are never modified once generated. |
120 | |
121 | sub SWASHGET { |
122 | my ($self, $start, $len) = @_; |
123 | local $^D = 0 if $^D; |
a0ed51b3 |
124 | my $type = $self->{TYPE}; |
125 | my $bits = $self->{BITS}; |
126 | my $none = $self->{NONE}; |
1fef36c7 |
127 | print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if $DEBUG; |
a0ed51b3 |
128 | my $end = $start + $len; |
129 | my $swatch = ""; |
130 | my $key; |
131 | vec($swatch, $len - 1, $bits) = 0; # Extend to correct length. |
132 | if ($none) { |
133 | for $key (0 .. $len - 1) { vec($swatch, $key, $bits) = $none } |
134 | } |
135 | |
136 | for ($self->{LIST}) { |
137 | pos $_ = 0; |
138 | if ($bits > 1) { |
139 | LINE: |
140 | while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0-9a-fA-F]+))?/mg) { |
141 | my $min = hex $1; |
142 | my $max = (defined $2 ? hex $2 : $min); |
143 | my $val = hex $3; |
144 | next if $max < $start; |
145 | # print "$min $max $val\n"; |
146 | if ($none) { |
147 | if ($min < $start) { |
710250cb |
148 | $val += $start - $min if $val < $none; |
a0ed51b3 |
149 | $min = $start; |
150 | } |
151 | for ($key = $min; $key <= $max; $key++) { |
152 | last LINE if $key >= $end; |
153 | # print STDERR "$key => $val\n" if $DEBUG; |
154 | vec($swatch, $key - $start, $bits) = $val; |
155 | ++$val if $val < $none; |
156 | } |
157 | } |
158 | else { |
159 | if ($min < $start) { |
160 | $val += $start - $min; |
161 | $min = $start; |
162 | } |
163 | for ($key = $min; $key <= $max; $key++, $val++) { |
164 | last LINE if $key >= $end; |
165 | # print STDERR "$key => $val\n" if $DEBUG; |
166 | vec($swatch, $key - $start, $bits) = $val; |
167 | } |
168 | } |
169 | } |
170 | } |
171 | else { |
172 | LINE: |
173 | while (/^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+))?/mg) { |
174 | my $min = hex $1; |
175 | my $max = (defined $2 ? hex $2 : $min); |
176 | next if $max < $start; |
177 | if ($min < $start) { |
178 | $min = $start; |
179 | } |
180 | for ($key = $min; $key <= $max; $key++) { |
181 | last LINE if $key >= $end; |
182 | # print STDERR "$key => 1\n" if $DEBUG; |
183 | vec($swatch, $key - $start, 1) = 1; |
184 | } |
185 | } |
186 | } |
187 | } |
188 | for my $x ($self->{EXTRAS}) { |
189 | pos $x = 0; |
1fef36c7 |
190 | while ($x =~ /^([-+!])(.*)/mg) { |
a0ed51b3 |
191 | my $char = $1; |
192 | my $name = $2; |
193 | print STDERR "INDIRECT $1 $2\n" if $DEBUG; |
1fef36c7 |
194 | my $otherbits = $self->{$name}->{BITS}; |
195 | croak("SWASHGET size mismatch") if $bits < $otherbits; |
196 | my $other = $self->{$name}->SWASHGET($start, $len); |
197 | if ($char eq '+') { |
198 | if ($bits == 1 and $otherbits == 1) { |
199 | $swatch |= $other; |
a0ed51b3 |
200 | } |
1fef36c7 |
201 | else { |
202 | for ($key = 0; $key < $len; $key++) { |
203 | vec($swatch, $key, $bits) = vec($other, $key, $otherbits); |
a0ed51b3 |
204 | } |
1fef36c7 |
205 | } |
206 | } |
207 | elsif ($char eq '!') { |
208 | if ($bits == 1 and $otherbits == 1) { |
209 | $swatch |= ~$other; |
210 | } |
211 | else { |
212 | for ($key = 0; $key < $len; $key++) { |
213 | if (!vec($other, $key, $otherbits)) { |
214 | vec($swatch, $key, $bits) = 1; |
a0ed51b3 |
215 | } |
216 | } |
217 | } |
1fef36c7 |
218 | } |
219 | elsif ($char eq '-') { |
220 | if ($bits == 1 and $otherbits == 1) { |
221 | $swatch &= ~$other; |
222 | } |
223 | else { |
224 | for ($key = 0; $key < $len; $key++) { |
225 | if (vec($other, $key, $otherbits)) { |
226 | vec($swatch, $key, $bits) = 0; |
a0ed51b3 |
227 | } |
228 | } |
229 | } |
230 | } |
231 | } |
232 | } |
233 | if ($DEBUG) { |
234 | print STDERR "CELLS "; |
235 | for ($key = 0; $key < $len; $key++) { |
236 | print STDERR vec($swatch, $key, $bits), " "; |
237 | } |
238 | print STDERR "\n"; |
239 | } |
240 | $swatch; |
241 | } |
242 | |
243 | 1; |