Commit | Line | Data |
a0ed51b3 |
1 | package utf8; |
cf25bb62 |
2 | use strict; |
3 | use warnings; |
a0ed51b3 |
4 | |
15732964 |
5 | sub DEBUG () { 0 } |
a0ed51b3 |
6 | |
7 | sub DESTROY {} |
8 | |
5beb625e |
9 | my %Cache; |
10 | |
12ac2576 |
11 | our (%PropertyAlias, %PA_reverse, %PropValueAlias, %PVA_reverse, %PVA_abbr_map); |
12 | |
86916d66 |
13 | sub croak { require Carp; Carp::croak(@_) } |
14 | |
5beb625e |
15 | ## |
3a2263fe |
16 | ## "SWASH" == "SWATCH HASH". A "swatch" is a swatch of the Unicode landscape. |
17 | ## It's a data structure that encodes a set of Unicode characters. |
5beb625e |
18 | ## |
19 | |
a0ed51b3 |
20 | sub SWASHNEW { |
21 | my ($class, $type, $list, $minbits, $none) = @_; |
22 | local $^D = 0 if $^D; |
15732964 |
23 | |
24 | print STDERR "SWASHNEW @_\n" if DEBUG; |
25 | |
cf25bb62 |
26 | ## |
27 | ## Get the list of codepoints for the type. |
711a919c |
28 | ## Called from swash_init (see utf8.c) or SWASHNEW itself. |
29 | ## |
30 | ## Callers of swash_init: |
31 | ## op.c:pmtrans -- for tr/// and y/// |
32 | ## regexec.c:regclass_swash -- for /[]/, \p, and \P |
33 | ## utf8.c:is_utf8_common -- for common Unicode properties |
34 | ## utf8.c:to_utf8_case -- for lc, uc, ucfirst, etc. and //i |
cf25bb62 |
35 | ## |
36 | ## Given a $type, our goal is to fill $list with the set of codepoint |
711a919c |
37 | ## ranges. If $type is false, $list passed is used. |
38 | ## |
39 | ## $minbits: |
40 | ## For binary properties, $minbits must be 1. |
41 | ## For character mappings (case and transliteration), $minbits must |
42 | ## be a number except 1. |
43 | ## |
44 | ## $list (or that filled according to $type): |
45 | ## Refer to perlunicode.pod, "User-Defined Character Properties." |
46 | ## |
47 | ## For binary properties, only characters with the property value |
48 | ## of True should be listed. The 3rd column, if any, will be ignored. |
cf25bb62 |
49 | ## |
50 | ## To make the parsing of $type clear, this code takes the a rather |
2d92f8a0 |
51 | ## unorthodox approach of last'ing out of the block once we have the |
cf25bb62 |
52 | ## info we need. Were this to be a subroutine, the 'last' would just |
53 | ## be a 'return'. |
54 | ## |
5beb625e |
55 | my $file; ## file to load data from, and also part of the %Cache key. |
56 | my $ListSorted = 0; |
57 | |
cf25bb62 |
58 | if ($type) |
59 | { |
60 | $type =~ s/^\s+//; |
61 | $type =~ s/\s+$//; |
62 | |
711a919c |
63 | print STDERR "type = $type\n" if DEBUG; |
cf25bb62 |
64 | |
cf25bb62 |
65 | GETFILE: |
66 | { |
12ac2576 |
67 | ## |
68 | ## It could be a user-defined property. |
69 | ## |
70 | |
71 | my $caller1 = $type =~ s/(.+)::// ? $1 : caller(1); |
72 | |
73 | if (defined $caller1 && $type =~ /^(?:\w+)$/) { |
74 | my $prop = "${caller1}::$type"; |
75 | if (exists &{$prop}) { |
76 | no strict 'refs'; |
77 | |
78 | $list = &{$prop}; |
79 | last GETFILE; |
80 | } |
81 | } |
82 | |
2d92f8a0 |
83 | my $wasIs; |
84 | |
85 | ($wasIs = $type =~ s/^Is(?:\s+|[-_])?//i) |
5beb625e |
86 | or |
12ac2576 |
87 | $type =~ s/^(?:(?:General(?:\s+|_)?)?Category|gc)\s*[:=]\s*//i |
5beb625e |
88 | or |
12ac2576 |
89 | $type =~ s/^(?:Script|sc)\s*[:=]\s*//i |
5beb625e |
90 | or |
12ac2576 |
91 | $type =~ s/^Block\s*[:=]\s*/In/i; |
92 | |
93 | |
94 | ## |
95 | ## See if it's in some enumeration. |
96 | ## |
7ebf06b3 |
97 | require "unicore/PVA.pl"; |
12ac2576 |
98 | if ($type =~ /^([\w\s]+)[:=]\s*(.*)/) { |
12ac2576 |
99 | my ($enum, $val) = (lc $1, lc $2); |
100 | $enum =~ tr/ _-//d; |
101 | $val =~ tr/ _-//d; |
102 | |
103 | my $pa = $PropertyAlias{$enum} ? $enum : $PA_reverse{$enum}; |
104 | my $f = $PropValueAlias{$pa}{$val} ? $val : $PVA_reverse{$pa}{lc $val}; |
105 | |
106 | if ($pa and $f) { |
107 | $pa = "gc_sc" if $pa eq "gc" or $pa eq "sc"; |
108 | $file = "unicore/lib/$pa/$PVA_abbr_map{$pa}{lc $f}.pl"; |
109 | last GETFILE; |
110 | } |
111 | } |
112 | else { |
113 | my $t = lc $type; |
114 | $t =~ tr/ _-//d; |
115 | |
116 | if ($PropValueAlias{gc}{$t} or $PropValueAlias{sc}{$t}) { |
117 | $file = "unicore/lib/gc_sc/$PVA_abbr_map{gc_sc}{$t}.pl"; |
118 | last GETFILE; |
119 | } |
120 | } |
cf25bb62 |
121 | |
5beb625e |
122 | ## |
123 | ## See if it's in the direct mapping table. |
124 | ## |
125 | require "unicore/Exact.pl"; |
126 | if (my $base = $utf8::Exact{$type}) { |
12ac2576 |
127 | $file = "unicore/lib/gc_sc/$base.pl"; |
5beb625e |
128 | last GETFILE; |
cf25bb62 |
129 | } |
130 | |
131 | ## |
5beb625e |
132 | ## If not there exactly, try the canonical form. The canonical |
133 | ## form is lowercased, with any separators (\s+|[-_]) removed. |
cf25bb62 |
134 | ## |
5beb625e |
135 | my $canonical = lc $type; |
136 | $canonical =~ s/(?<=[a-z\d])(?:\s+|[-_])(?=[a-z\d])//g; |
711a919c |
137 | print STDERR "canonical = $canonical\n" if DEBUG; |
cf25bb62 |
138 | |
5beb625e |
139 | require "unicore/Canonical.pl"; |
0a0ffbce |
140 | { no warnings "uninitialized"; |
12ac2576 |
141 | if (my $base = ($utf8::Canonical{$canonical} || $utf8::Canonical{ lc $utf8::PropertyAlias{$canonical} })) { |
142 | $file = "unicore/lib/gc_sc/$base.pl"; |
5beb625e |
143 | last GETFILE; |
cf25bb62 |
144 | } |
0a0ffbce |
145 | } |
cf25bb62 |
146 | |
491fd90a |
147 | ## |
3a2263fe |
148 | ## See if it's a user-level "To". |
149 | ## |
150 | |
151 | my $caller0 = caller(0); |
152 | |
153 | if (defined $caller0 && $type =~ /^To(?:\w+)$/) { |
154 | my $map = $caller0 . "::" . $type; |
09e0265a |
155 | |
3a2263fe |
156 | if (exists &{$map}) { |
157 | no strict 'refs'; |
158 | |
159 | $list = &{$map}; |
160 | last GETFILE; |
161 | } |
162 | } |
163 | |
cf25bb62 |
164 | ## |
3a2263fe |
165 | ## Last attempt -- see if it's a standard "To" name |
166 | ## (e.g. "ToLower") ToTitle is used by ucfirst(). |
167 | ## The user-level way to access ToDigit() and ToFold() |
168 | ## is to use Unicode::UCD. |
cf25bb62 |
169 | ## |
711a919c |
170 | if ($type =~ /^To(Digit|Fold|Lower|Title|Upper)$/) { |
cf25bb62 |
171 | $file = "unicore/To/$1.pl"; |
172 | ## would like to test to see if $file actually exists.... |
173 | last GETFILE; |
174 | } |
175 | |
176 | ## |
177 | ## If we reach this line, it's because we couldn't figure |
178 | ## out what to do with $type. Ouch. |
179 | ## |
bc45ce41 |
180 | |
181 | return $type; |
cf25bb62 |
182 | } |
183 | |
491fd90a |
184 | if (defined $file) { |
711a919c |
185 | print STDERR "found it (file='$file')\n" if DEBUG; |
491fd90a |
186 | |
187 | ## |
188 | ## If we reach here, it was due to a 'last GETFILE' above |
3a2263fe |
189 | ## (exception: user-defined properties and mappings), so we |
491fd90a |
190 | ## have a filename, so now we load it if we haven't already. |
191 | ## If we have, return the cached results. The cache key is the |
c49d5ed7 |
192 | ## class and file to load. |
491fd90a |
193 | ## |
c49d5ed7 |
194 | my $found = $Cache{$class, $file}; |
195 | if ($found and ref($found) eq $class) { |
711a919c |
196 | print STDERR "Returning cached '$file' for \\p{$type}\n" if DEBUG; |
c49d5ed7 |
197 | return $found; |
491fd90a |
198 | } |
199 | |
e81465be |
200 | $list = do $file; die $@ if $@; |
491fd90a |
201 | } |
5beb625e |
202 | |
5beb625e |
203 | $ListSorted = 1; ## we know that these lists are sorted |
886ba366 |
204 | } |
a0ed51b3 |
205 | |
15732964 |
206 | my $extras; |
711a919c |
207 | my $bits = $minbits; |
cf25bb62 |
208 | |
5beb625e |
209 | my $ORIG = $list; |
a0ed51b3 |
210 | if ($list) { |
211 | my @tmp = split(/^/m, $list); |
212 | my %seen; |
db376a24 |
213 | no warnings; |
a0ed51b3 |
214 | $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; |
215 | $list = join '', |
df68b396 |
216 | map { $_->[1] } |
217 | sort { $a->[0] <=> $b->[0] } |
075d4edd |
218 | map { /^([0-9a-fA-F]+)/; [ CORE::hex($1), $_ ] } |
df68b396 |
219 | grep { /^([0-9a-fA-F]+)/ and not $seen{$1}++ } @tmp; # XXX doesn't do ranges right |
a0ed51b3 |
220 | } |
221 | |
222 | if ($none) { |
223 | my $hextra = sprintf "%04x", $none + 1; |
224 | $list =~ s/\tXXXX$/\t$hextra/mg; |
225 | } |
226 | |
711a919c |
227 | if ($minbits != 1 && $minbits < 32) { # not binary property |
a0ed51b3 |
228 | my $top = 0; |
3a2263fe |
229 | while ($list =~ /^([0-9a-fA-F]+)(?:[\t]([0-9a-fA-F]+)?)(?:[ \t]([0-9a-fA-F]+))?/mg) { |
075d4edd |
230 | my $min = CORE::hex $1; |
231 | my $max = defined $2 ? CORE::hex $2 : $min; |
232 | my $val = defined $3 ? CORE::hex $3 : 0; |
a0ed51b3 |
233 | $val += $max - $min if defined $3; |
234 | $top = $val if $val > $top; |
235 | } |
711a919c |
236 | my $topbits = |
a0ed51b3 |
237 | $top > 0xffff ? 32 : |
711a919c |
238 | $top > 0xff ? 16 : 8; |
239 | $bits = $topbits if $bits < $topbits; |
a0ed51b3 |
240 | } |
a0ed51b3 |
241 | |
242 | my @extras; |
243 | for my $x ($extras) { |
244 | pos $x = 0; |
1fef36c7 |
245 | while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) { |
a0ed51b3 |
246 | my $char = $1; |
247 | my $name = $2; |
6d6ae962 |
248 | print STDERR "$1 => $2\n" if DEBUG; |
09e0265a |
249 | if ($char =~ /[-+!&]/) { |
a0ed51b3 |
250 | my ($c,$t) = split(/::/, $name, 2); # bogus use of ::, really |
11ef8fdd |
251 | my $subobj; |
252 | if ($c eq 'utf8') { |
711a919c |
253 | $subobj = utf8->SWASHNEW($t, "", $minbits, 0); |
09e0265a |
254 | } |
255 | elsif (exists &$name) { |
711a919c |
256 | $subobj = utf8->SWASHNEW($name, "", $minbits, 0); |
11ef8fdd |
257 | } |
258 | elsif ($c =~ /^([0-9a-fA-F]+)/) { |
711a919c |
259 | $subobj = utf8->SWASHNEW("", $c, $minbits, 0); |
11ef8fdd |
260 | } |
bc45ce41 |
261 | return $subobj unless ref $subobj; |
a0ed51b3 |
262 | push @extras, $name => $subobj; |
263 | $bits = $subobj->{BITS} if $bits < $subobj->{BITS}; |
264 | } |
265 | } |
266 | } |
267 | |
15732964 |
268 | print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG; |
a0ed51b3 |
269 | |
5beb625e |
270 | my $SWASH = bless { |
a0ed51b3 |
271 | TYPE => $type, |
272 | BITS => $bits, |
273 | EXTRAS => $extras, |
274 | LIST => $list, |
275 | NONE => $none, |
276 | @extras, |
277 | } => $class; |
5beb625e |
278 | |
279 | if ($file) { |
280 | $Cache{$class, $file} = $SWASH; |
281 | } |
282 | |
283 | return $SWASH; |
a0ed51b3 |
284 | } |
285 | |
979f2922 |
286 | # Now SWASHGET is recasted into a C function S_swash_get (see utf8.c). |
a0ed51b3 |
287 | |
288 | 1; |