This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
UCD.t: white-space only
[perl5.git] / lib / Unicode / UCD.t
CommitLineData
25a47338 1#!perl -w
8b731da2 2BEGIN {
a452d459 3 if (ord("A") != 65) {
8b731da2
JH
4 print "1..0 # Skip: EBCDIC\n";
5 exit 0;
6 }
a778afa6
AD
7 chdir 't' if -d 't';
8 @INC = '../lib';
25a47338
NC
9 require Config; import Config;
10 if ($Config{'extensions'} !~ /\bStorable\b/) {
11 print "1..0 # Skip: Storable was not built; Unicode::UCD uses Storable\n";
12 exit 0;
13 }
8b731da2
JH
14}
15
a778afa6
AD
16use strict;
17use Unicode::UCD;
f5c9f3db 18use Test::More;
8b731da2 19
55d7b906 20use Unicode::UCD 'charinfo';
561c79ed 21
ce066323
FC
22$/ = 7;
23
b08cd201
JH
24my $charinfo;
25
9e92970c
KW
26is(charinfo(0x110000), undef, "Verify charinfo() of non-unicode is undef");
27
e10d7780
KW
28$charinfo = charinfo(0); # Null is often problematic, so test it.
29
30is($charinfo->{code}, '0000', '<control>');
31is($charinfo->{name}, '<control>');
32is($charinfo->{category}, 'Cc');
33is($charinfo->{combining}, '0');
34is($charinfo->{bidi}, 'BN');
35is($charinfo->{decomposition}, '');
36is($charinfo->{decimal}, '');
37is($charinfo->{digit}, '');
38is($charinfo->{numeric}, '');
39is($charinfo->{mirrored}, 'N');
40is($charinfo->{unicode10}, 'NULL');
41is($charinfo->{comment}, '');
42is($charinfo->{upper}, '');
43is($charinfo->{lower}, '');
44is($charinfo->{title}, '');
45is($charinfo->{block}, 'Basic Latin');
46is($charinfo->{script}, 'Common');
47
b08cd201
JH
48$charinfo = charinfo(0x41);
49
f5c9f3db
JH
50is($charinfo->{code}, '0041', 'LATIN CAPITAL LETTER A');
51is($charinfo->{name}, 'LATIN CAPITAL LETTER A');
52is($charinfo->{category}, 'Lu');
53is($charinfo->{combining}, '0');
54is($charinfo->{bidi}, 'L');
55is($charinfo->{decomposition}, '');
56is($charinfo->{decimal}, '');
57is($charinfo->{digit}, '');
58is($charinfo->{numeric}, '');
59is($charinfo->{mirrored}, 'N');
60is($charinfo->{unicode10}, '');
61is($charinfo->{comment}, '');
62is($charinfo->{upper}, '');
63is($charinfo->{lower}, '0061');
64is($charinfo->{title}, '');
65is($charinfo->{block}, 'Basic Latin');
66is($charinfo->{script}, 'Latin');
b08cd201
JH
67
68$charinfo = charinfo(0x100);
69
f5c9f3db
JH
70is($charinfo->{code}, '0100', 'LATIN CAPITAL LETTER A WITH MACRON');
71is($charinfo->{name}, 'LATIN CAPITAL LETTER A WITH MACRON');
72is($charinfo->{category}, 'Lu');
73is($charinfo->{combining}, '0');
74is($charinfo->{bidi}, 'L');
75is($charinfo->{decomposition}, '0041 0304');
76is($charinfo->{decimal}, '');
77is($charinfo->{digit}, '');
78is($charinfo->{numeric}, '');
79is($charinfo->{mirrored}, 'N');
80is($charinfo->{unicode10}, 'LATIN CAPITAL LETTER A MACRON');
81is($charinfo->{comment}, '');
82is($charinfo->{upper}, '');
83is($charinfo->{lower}, '0101');
84is($charinfo->{title}, '');
85is($charinfo->{block}, 'Latin Extended-A');
86is($charinfo->{script}, 'Latin');
a196fbfd
JH
87
88# 0x0590 is in the Hebrew block but unused.
561c79ed 89
b08cd201
JH
90$charinfo = charinfo(0x590);
91
f5c9f3db
JH
92is($charinfo->{code}, undef, '0x0590 - unused Hebrew');
93is($charinfo->{name}, undef);
94is($charinfo->{category}, undef);
95is($charinfo->{combining}, undef);
96is($charinfo->{bidi}, undef);
97is($charinfo->{decomposition}, undef);
98is($charinfo->{decimal}, undef);
99is($charinfo->{digit}, undef);
100is($charinfo->{numeric}, undef);
101is($charinfo->{mirrored}, undef);
102is($charinfo->{unicode10}, undef);
103is($charinfo->{comment}, undef);
104is($charinfo->{upper}, undef);
105is($charinfo->{lower}, undef);
106is($charinfo->{title}, undef);
107is($charinfo->{block}, undef);
108is($charinfo->{script}, undef);
a196fbfd
JH
109
110# 0x05d0 is in the Hebrew block and used.
561c79ed 111
b08cd201
JH
112$charinfo = charinfo(0x5d0);
113
f5c9f3db
JH
114is($charinfo->{code}, '05D0', '05D0 - used Hebrew');
115is($charinfo->{name}, 'HEBREW LETTER ALEF');
116is($charinfo->{category}, 'Lo');
117is($charinfo->{combining}, '0');
118is($charinfo->{bidi}, 'R');
119is($charinfo->{decomposition}, '');
120is($charinfo->{decimal}, '');
121is($charinfo->{digit}, '');
122is($charinfo->{numeric}, '');
123is($charinfo->{mirrored}, 'N');
124is($charinfo->{unicode10}, '');
125is($charinfo->{comment}, '');
126is($charinfo->{upper}, '');
127is($charinfo->{lower}, '');
128is($charinfo->{title}, '');
129is($charinfo->{block}, 'Hebrew');
130is($charinfo->{script}, 'Hebrew');
561c79ed 131
74f8133e 132# An open syllable in Hangul.
a6fa416b
TS
133
134$charinfo = charinfo(0xAC00);
135
05dbc6f8
KW
136is($charinfo->{code}, 'AC00', 'HANGUL SYLLABLE U+AC00');
137is($charinfo->{name}, 'HANGUL SYLLABLE GA');
f5c9f3db
JH
138is($charinfo->{category}, 'Lo');
139is($charinfo->{combining}, '0');
140is($charinfo->{bidi}, 'L');
05dbc6f8 141is($charinfo->{decomposition}, '1100 1161');
f5c9f3db
JH
142is($charinfo->{decimal}, '');
143is($charinfo->{digit}, '');
144is($charinfo->{numeric}, '');
145is($charinfo->{mirrored}, 'N');
146is($charinfo->{unicode10}, '');
147is($charinfo->{comment}, '');
148is($charinfo->{upper}, '');
149is($charinfo->{lower}, '');
150is($charinfo->{title}, '');
151is($charinfo->{block}, 'Hangul Syllables');
152is($charinfo->{script}, 'Hangul');
a6fa416b 153
74f8133e 154# A closed syllable in Hangul.
a6fa416b
TS
155
156$charinfo = charinfo(0xAE00);
157
05dbc6f8
KW
158is($charinfo->{code}, 'AE00', 'HANGUL SYLLABLE U+AE00');
159is($charinfo->{name}, 'HANGUL SYLLABLE GEUL');
f5c9f3db
JH
160is($charinfo->{category}, 'Lo');
161is($charinfo->{combining}, '0');
162is($charinfo->{bidi}, 'L');
05dbc6f8 163is($charinfo->{decomposition}, "1100 1173 11AF");
f5c9f3db
JH
164is($charinfo->{decimal}, '');
165is($charinfo->{digit}, '');
166is($charinfo->{numeric}, '');
167is($charinfo->{mirrored}, 'N');
168is($charinfo->{unicode10}, '');
169is($charinfo->{comment}, '');
170is($charinfo->{upper}, '');
171is($charinfo->{lower}, '');
172is($charinfo->{title}, '');
173is($charinfo->{block}, 'Hangul Syllables');
174is($charinfo->{script}, 'Hangul');
a6fa416b
TS
175
176$charinfo = charinfo(0x1D400);
177
f5c9f3db
JH
178is($charinfo->{code}, '1D400', 'MATHEMATICAL BOLD CAPITAL A');
179is($charinfo->{name}, 'MATHEMATICAL BOLD CAPITAL A');
180is($charinfo->{category}, 'Lu');
181is($charinfo->{combining}, '0');
182is($charinfo->{bidi}, 'L');
183is($charinfo->{decomposition}, '<font> 0041');
184is($charinfo->{decimal}, '');
185is($charinfo->{digit}, '');
186is($charinfo->{numeric}, '');
187is($charinfo->{mirrored}, 'N');
188is($charinfo->{unicode10}, '');
189is($charinfo->{comment}, '');
190is($charinfo->{upper}, '');
191is($charinfo->{lower}, '');
192is($charinfo->{title}, '');
193is($charinfo->{block}, 'Mathematical Alphanumeric Symbols');
7be0dac3 194is($charinfo->{script}, 'Common');
a6fa416b 195
a452d459
KW
196$charinfo = charinfo(0x9FBA); #Bug 58428
197
198is($charinfo->{code}, '9FBA', 'U+9FBA');
199is($charinfo->{name}, 'CJK UNIFIED IDEOGRAPH-9FBA');
200is($charinfo->{category}, 'Lo');
201is($charinfo->{combining}, '0');
202is($charinfo->{bidi}, 'L');
203is($charinfo->{decomposition}, '');
204is($charinfo->{decimal}, '');
205is($charinfo->{digit}, '');
206is($charinfo->{numeric}, '');
207is($charinfo->{mirrored}, 'N');
208is($charinfo->{unicode10}, '');
209is($charinfo->{comment}, '');
210is($charinfo->{upper}, '');
211is($charinfo->{lower}, '');
212is($charinfo->{title}, '');
213is($charinfo->{block}, 'CJK Unified Ideographs');
214is($charinfo->{script}, 'Han');
215
55d7b906 216use Unicode::UCD qw(charblock charscript);
a196fbfd
JH
217
218# 0x0590 is in the Hebrew block but unused.
561c79ed 219
f5c9f3db 220is(charblock(0x590), 'Hebrew', '0x0590 - Hebrew unused charblock');
8079ad82 221is(charscript(0x590), 'Unknown', '0x0590 - Hebrew unused charscript');
c707cf8e 222is(charblock(0x1FFFF), 'No_Block', '0x1FFFF - unused charblock');
561c79ed 223
b08cd201
JH
224$charinfo = charinfo(0xbe);
225
f5c9f3db
JH
226is($charinfo->{code}, '00BE', 'VULGAR FRACTION THREE QUARTERS');
227is($charinfo->{name}, 'VULGAR FRACTION THREE QUARTERS');
228is($charinfo->{category}, 'No');
229is($charinfo->{combining}, '0');
230is($charinfo->{bidi}, 'ON');
231is($charinfo->{decomposition}, '<fraction> 0033 2044 0034');
232is($charinfo->{decimal}, '');
233is($charinfo->{digit}, '');
234is($charinfo->{numeric}, '3/4');
235is($charinfo->{mirrored}, 'N');
236is($charinfo->{unicode10}, 'FRACTION THREE QUARTERS');
237is($charinfo->{comment}, '');
238is($charinfo->{upper}, '');
239is($charinfo->{lower}, '');
240is($charinfo->{title}, '');
241is($charinfo->{block}, 'Latin-1 Supplement');
7be0dac3 242is($charinfo->{script}, 'Common');
10a6ecd2 243
7fcb6f68
KW
244# This is to test a case where both simple and full lowercases exist and
245# differ
246$charinfo = charinfo(0x130);
247
248is($charinfo->{code}, '0130', 'LATIN CAPITAL LETTER I WITH DOT ABOVE');
249is($charinfo->{name}, 'LATIN CAPITAL LETTER I WITH DOT ABOVE');
250is($charinfo->{category}, 'Lu');
251is($charinfo->{combining}, '0');
252is($charinfo->{bidi}, 'L');
253is($charinfo->{decomposition}, '0049 0307');
254is($charinfo->{decimal}, '');
255is($charinfo->{digit}, '');
256is($charinfo->{numeric}, '');
257is($charinfo->{mirrored}, 'N');
258is($charinfo->{unicode10}, 'LATIN CAPITAL LETTER I DOT');
259is($charinfo->{comment}, '');
260is($charinfo->{upper}, '');
261is($charinfo->{lower}, '0069');
262is($charinfo->{title}, '');
263is($charinfo->{block}, 'Latin Extended-A');
264is($charinfo->{script}, 'Latin');
265
266# This is to test a case where both simple and full uppercases exist and
267# differ
268$charinfo = charinfo(0x1F80);
269
270is($charinfo->{code}, '1F80', 'GREEK SMALL LETTER ALPHA WITH PSILI AND YPOGEGRAMMENI');
271is($charinfo->{name}, 'GREEK SMALL LETTER ALPHA WITH PSILI AND YPOGEGRAMMENI');
272is($charinfo->{category}, 'Ll');
273is($charinfo->{combining}, '0');
274is($charinfo->{bidi}, 'L');
275is($charinfo->{decomposition}, '1F00 0345');
276is($charinfo->{decimal}, '');
277is($charinfo->{digit}, '');
278is($charinfo->{numeric}, '');
279is($charinfo->{mirrored}, 'N');
280is($charinfo->{unicode10}, '');
281is($charinfo->{comment}, '');
282is($charinfo->{upper}, '1F88');
283is($charinfo->{lower}, '');
284is($charinfo->{title}, '1F88');
285is($charinfo->{block}, 'Greek Extended');
286is($charinfo->{script}, 'Greek');
287
55d7b906 288use Unicode::UCD qw(charblocks charscripts);
10a6ecd2 289
b08cd201 290my $charblocks = charblocks();
10a6ecd2 291
f5c9f3db
JH
292ok(exists $charblocks->{Thai}, 'Thai charblock exists');
293is($charblocks->{Thai}->[0]->[0], hex('0e00'));
294ok(!exists $charblocks->{PigLatin}, 'PigLatin charblock does not exist');
10a6ecd2 295
b08cd201 296my $charscripts = charscripts();
10a6ecd2 297
f5c9f3db
JH
298ok(exists $charscripts->{Armenian}, 'Armenian charscript exists');
299is($charscripts->{Armenian}->[0]->[0], hex('0531'));
300ok(!exists $charscripts->{PigLatin}, 'PigLatin charscript does not exist');
10a6ecd2
JH
301
302my $charscript;
303
304$charscript = charscript("12ab");
f5c9f3db 305is($charscript, 'Ethiopic', 'Ethiopic charscript');
10a6ecd2
JH
306
307$charscript = charscript("0x12ab");
f5c9f3db 308is($charscript, 'Ethiopic');
10a6ecd2
JH
309
310$charscript = charscript("U+12ab");
f5c9f3db 311is($charscript, 'Ethiopic');
10a6ecd2
JH
312
313my $ranges;
314
315$ranges = charscript('Ogham');
7bccef0b
KW
316is($ranges->[0]->[0], hex('1680'), 'Ogham charscript');
317is($ranges->[0]->[1], hex('169C'));
10a6ecd2 318
55d7b906 319use Unicode::UCD qw(charinrange);
10a6ecd2
JH
320
321$ranges = charscript('Cherokee');
f5c9f3db 322ok(!charinrange($ranges, "139f"), 'Cherokee charscript');
10a6ecd2
JH
323ok( charinrange($ranges, "13a0"));
324ok( charinrange($ranges, "13f4"));
325ok(!charinrange($ranges, "13f5"));
326
ea508aee
JH
327use Unicode::UCD qw(general_categories);
328
329my $gc = general_categories();
330
331ok(exists $gc->{L}, 'has L');
332is($gc->{L}, 'Letter', 'L is Letter');
333is($gc->{Lu}, 'UppercaseLetter', 'Lu is UppercaseLetter');
334
335use Unicode::UCD qw(bidi_types);
336
337my $bt = bidi_types();
338
339ok(exists $bt->{L}, 'has L');
340is($bt->{L}, 'Left-to-Right', 'L is Left-to-Right');
341is($bt->{AL}, 'Right-to-Left Arabic', 'AL is Right-to-Left Arabic');
342
a452d459
KW
343# If this fails, then maybe one should look at the Unicode changes to see
344# what else might need to be updated.
7620cb10 345is(Unicode::UCD::UnicodeVersion, '6.1.0', 'UnicodeVersion');
b08cd201 346
55d7b906 347use Unicode::UCD qw(compexcl);
b08cd201 348
f5c9f3db 349ok(!compexcl(0x0100), 'compexcl');
71a442a8
KW
350ok(!compexcl(0xD801), 'compexcl of surrogate');
351ok(!compexcl(0x110000), 'compexcl of non-Unicode code point');
b08cd201
JH
352ok( compexcl(0x0958));
353
55d7b906 354use Unicode::UCD qw(casefold);
b08cd201
JH
355
356my $casefold;
357
358$casefold = casefold(0x41);
359
a452d459
KW
360is($casefold->{code}, '0041', 'casefold 0x41 code');
361is($casefold->{status}, 'C', 'casefold 0x41 status');
362is($casefold->{mapping}, '0061', 'casefold 0x41 mapping');
363is($casefold->{full}, '0061', 'casefold 0x41 full');
364is($casefold->{simple}, '0061', 'casefold 0x41 simple');
365is($casefold->{turkic}, "", 'casefold 0x41 turkic');
b08cd201
JH
366
367$casefold = casefold(0xdf);
368
a452d459
KW
369is($casefold->{code}, '00DF', 'casefold 0xDF code');
370is($casefold->{status}, 'F', 'casefold 0xDF status');
371is($casefold->{mapping}, '0073 0073', 'casefold 0xDF mapping');
372is($casefold->{full}, '0073 0073', 'casefold 0xDF full');
373is($casefold->{simple}, "", 'casefold 0xDF simple');
374is($casefold->{turkic}, "", 'casefold 0xDF turkic');
375
376# Do different tests depending on if version <= 3.1, or not.
377(my $version = Unicode::UCD::UnicodeVersion) =~ /^(\d+)\.(\d+)/;
378if (defined $1 && ($1 <= 2 || $1 == 3 && defined $2 && $2 <= 1)) {
379 $casefold = casefold(0x130);
380
381 is($casefold->{code}, '0130', 'casefold 0x130 code');
382 is($casefold->{status}, 'I' , 'casefold 0x130 status');
383 is($casefold->{mapping}, '0069', 'casefold 0x130 mapping');
384 is($casefold->{full}, '0069', 'casefold 0x130 full');
385 is($casefold->{simple}, "0069", 'casefold 0x130 simple');
386 is($casefold->{turkic}, "0069", 'casefold 0x130 turkic');
387
388 $casefold = casefold(0x131);
389
390 is($casefold->{code}, '0131', 'casefold 0x131 code');
391 is($casefold->{status}, 'I' , 'casefold 0x131 status');
392 is($casefold->{mapping}, '0069', 'casefold 0x131 mapping');
393 is($casefold->{full}, '0069', 'casefold 0x131 full');
394 is($casefold->{simple}, "0069", 'casefold 0x131 simple');
395 is($casefold->{turkic}, "0069", 'casefold 0x131 turkic');
396} else {
397 $casefold = casefold(0x49);
398
399 is($casefold->{code}, '0049', 'casefold 0x49 code');
400 is($casefold->{status}, 'C' , 'casefold 0x49 status');
401 is($casefold->{mapping}, '0069', 'casefold 0x49 mapping');
402 is($casefold->{full}, '0069', 'casefold 0x49 full');
403 is($casefold->{simple}, "0069", 'casefold 0x49 simple');
404 is($casefold->{turkic}, "0131", 'casefold 0x49 turkic');
405
406 $casefold = casefold(0x130);
407
408 is($casefold->{code}, '0130', 'casefold 0x130 code');
409 is($casefold->{status}, 'F' , 'casefold 0x130 status');
410 is($casefold->{mapping}, '0069 0307', 'casefold 0x130 mapping');
411 is($casefold->{full}, '0069 0307', 'casefold 0x130 full');
412 is($casefold->{simple}, "", 'casefold 0x130 simple');
413 is($casefold->{turkic}, "0069", 'casefold 0x130 turkic');
414}
415
416$casefold = casefold(0x1F88);
417
418is($casefold->{code}, '1F88', 'casefold 0x1F88 code');
419is($casefold->{status}, 'S' , 'casefold 0x1F88 status');
420is($casefold->{mapping}, '1F80', 'casefold 0x1F88 mapping');
421is($casefold->{full}, '1F00 03B9', 'casefold 0x1F88 full');
422is($casefold->{simple}, '1F80', 'casefold 0x1F88 simple');
423is($casefold->{turkic}, "", 'casefold 0x1F88 turkic');
b08cd201
JH
424
425ok(!casefold(0x20));
426
55d7b906 427use Unicode::UCD qw(casespec);
b08cd201
JH
428
429my $casespec;
430
431ok(!casespec(0x41));
432
433$casespec = casespec(0xdf);
434
435ok($casespec->{code} eq '00DF' &&
436 $casespec->{lower} eq '00DF' &&
437 $casespec->{title} eq '0053 0073' &&
438 $casespec->{upper} eq '0053 0053' &&
2d3cf3ee 439 !defined $casespec->{condition}, 'casespec 0xDF');
b08cd201
JH
440
441$casespec = casespec(0x307);
442
f499c386 443ok($casespec->{az}->{code} eq '0307' &&
2d3cf3ee 444 !defined $casespec->{az}->{lower} &&
f499c386
JH
445 $casespec->{az}->{title} eq '0307' &&
446 $casespec->{az}->{upper} eq '0307' &&
9c3dc587 447 $casespec->{az}->{condition} eq 'az After_I',
f5c9f3db 448 'casespec 0x307');
6c8d78fb
HS
449
450# perl #7305 UnicodeCD::compexcl is weird
451
2d3cf3ee 452for (1) {my $a=compexcl $_}
6c8d78fb 453ok(1, 'compexcl read-only $_: perl #7305');
1f27373c 454map {compexcl $_} %{{1=>2}};
6c8d78fb
HS
455ok(1, 'compexcl read-only hash: perl #7305');
456
d7829152
JH
457is(Unicode::UCD::_getcode('123'), 123, "_getcode(123)");
458is(Unicode::UCD::_getcode('0123'), 0x123, "_getcode(0123)");
459is(Unicode::UCD::_getcode('0x123'), 0x123, "_getcode(0x123)");
460is(Unicode::UCD::_getcode('0X123'), 0x123, "_getcode(0X123)");
461is(Unicode::UCD::_getcode('U+123'), 0x123, "_getcode(U+123)");
462is(Unicode::UCD::_getcode('u+123'), 0x123, "_getcode(u+123)");
463is(Unicode::UCD::_getcode('U+1234'), 0x1234, "_getcode(U+1234)");
464is(Unicode::UCD::_getcode('U+12345'), 0x12345, "_getcode(U+12345)");
465is(Unicode::UCD::_getcode('123x'), undef, "_getcode(123x)");
466is(Unicode::UCD::_getcode('x123'), undef, "_getcode(x123)");
467is(Unicode::UCD::_getcode('0x123x'), undef, "_getcode(x123)");
468is(Unicode::UCD::_getcode('U+123x'), undef, "_getcode(x123)");
741297c1
JH
469
470{
471 my $r1 = charscript('Latin');
472 my $n1 = @$r1;
7620cb10 473 is($n1, 30, "number of ranges in Latin script (Unicode 6.1.0)");
741297c1
JH
474 shift @$r1 while @$r1;
475 my $r2 = charscript('Latin');
476 is(@$r2, $n1, "modifying results should not mess up internal caches");
477}
478
c5a29f40
LM
479{
480 is(charinfo(0xdeadbeef), undef, "[perl #23273] warnings in Unicode::UCD");
2d3cf3ee 481}
a2bd7410
JH
482
483use Unicode::UCD qw(namedseq);
484
485is(namedseq("KATAKANA LETTER AINU P"), "\x{31F7}\x{309A}", "namedseq");
486is(namedseq("KATAKANA LETTER AINU Q"), undef);
487is(namedseq(), undef);
488is(namedseq(qw(foo bar)), undef);
489my @ns = namedseq("KATAKANA LETTER AINU P");
490is(scalar @ns, 2);
491is($ns[0], 0x31F7);
492is($ns[1], 0x309A);
493my %ns = namedseq();
494is($ns{"KATAKANA LETTER AINU P"}, "\x{31F7}\x{309A}");
495@ns = namedseq(42);
496is(@ns, 0);
497
7319f91d
KW
498use Unicode::UCD qw(num);
499use charnames ":full";
500
501is(num("0"), 0, 'Verify num("0") == 0');
502is(num("98765"), 98765, 'Verify num("98765") == 98765');
503ok(! defined num("98765\N{FULLWIDTH DIGIT FOUR}"), 'Verify num("98765\N{FULLWIDTH DIGIT FOUR}") isnt defined');
c4b3d89d
KW
504is(num("\N{NEW TAI LUE DIGIT TWO}\N{NEW TAI LUE DIGIT ONE}"), 21, 'Verify num("\N{NEW TAI LUE DIGIT TWO}\N{NEW TAI LUE DIGIT ONE}") == 21');
505ok(! defined num("\N{NEW TAI LUE DIGIT TWO}\N{NEW TAI LUE THAM DIGIT ONE}"), 'Verify num("\N{NEW TAI LUE DIGIT TWO}\N{NEW TAI LUE THAM DIGIT ONE}") isnt defined');
7319f91d
KW
506is(num("\N{CHAM DIGIT ZERO}\N{CHAM DIGIT THREE}"), 3, 'Verify num("\N{CHAM DIGIT ZERO}\N{CHAM DIGIT THREE}") == 3');
507ok(! defined num("\N{CHAM DIGIT ZERO}\N{JAVANESE DIGIT NINE}"), 'Verify num("\N{CHAM DIGIT ZERO}\N{JAVANESE DIGIT NINE}") isnt defined');
508is(num("\N{SUPERSCRIPT TWO}"), 2, 'Verify num("\N{SUPERSCRIPT TWO} == 2');
509is(num("\N{ETHIOPIC NUMBER TEN THOUSAND}"), 10000, 'Verify num("\N{ETHIOPIC NUMBER TEN THOUSAND}") == 10000');
510is(num("\N{NORTH INDIC FRACTION ONE HALF}"), .5, 'Verify num("\N{NORTH INDIC FRACTION ONE HALF}") == .5');
98025745 511is(num("\N{U+12448}"), 9, 'Verify num("\N{U+12448}") == 9');
4f143a72 512is(num("\N{U+5146}"), 1000000000000, 'Verify num("\N{U+5146}") == 1000000000000');
eaebe4db 513
7ef25837
KW
514# Create a user-defined property
515sub InKana {<<'END'}
5163040 309F
51730A0 30FF
518END
519
520use Unicode::UCD qw(prop_aliases);
521
522is(prop_aliases(undef), undef, "prop_aliases(undef) returns <undef>");
523is(prop_aliases("unknown property"), undef,
524 "prop_aliases(<unknown property>) returns <undef>");
525is(prop_aliases("InKana"), undef,
526 "prop_aliases(<user-defined property>) returns <undef>");
527is(prop_aliases("Perl_Decomposition_Mapping"), undef, "prop_aliases('Perl_Decomposition_Mapping') returns <undef> since internal-Perl-only");
528is(prop_aliases("Perl_Charnames"), undef,
529 "prop_aliases('Perl_Charnames') returns <undef> since internal-Perl-only");
530is(prop_aliases("isgc"), undef,
531 "prop_aliases('isgc') returns <undef> since is not covered Perl extension");
532is(prop_aliases("Is_Is_Any"), undef,
533 "prop_aliases('Is_Is_Any') returns <undef> since two is's");
534
535require 'utf8_heavy.pl';
536require "unicore/Heavy.pl";
537
538# Keys are lists of properties. Values are defined if have been tested.
539my %props;
540
541# To test for loose matching, add in the characters that are ignored there.
542my $extra_chars = "-_ ";
543
544# The one internal property we accept
545$props{'Perl_Decimal_Digit'} = 1;
546my @list = prop_aliases("perldecimaldigit");
547is_deeply(\@list,
548 [ "Perl_Decimal_Digit",
549 "Perl_Decimal_Digit"
550 ], "prop_aliases('perldecimaldigit') returns Perl_Decimal_Digit as both short and full names");
551
552# Get the official Unicode property name synonyms and test them.
553open my $props, "<", "../lib/unicore/PropertyAliases.txt"
554 or die "Can't open Unicode PropertyAliases.txt";
ce066323 555$/ = "\n";
7ef25837
KW
556while (<$props>) {
557 s/\s*#.*//; # Remove comments
558 next if /^\s* $/x; # Ignore empty and comment lines
559
560 chomp;
561 my $count = 0; # 0th field in line is short name; 1th is long name
562 my $short_name;
563 my $full_name;
564 my @names_via_short;
565 foreach my $alias (split /\s*;\s*/) { # Fields are separated by
566 # semi-colons
567 # Add in the characters that are supposed to be ignored, to test loose
568 # matching, which the tested function does on all inputs.
569 my $mod_name = "$extra_chars$alias";
570
e72b6605 571 my $loose = &utf8::_loose_name(lc $alias);
7ef25837
KW
572
573 # Indicate we have tested this.
574 $props{$loose} = 1;
575
576 my @all_names = prop_aliases($mod_name);
577 if (grep { $_ eq $loose } @Unicode::UCD::suppressed_properties) {
578 is(@all_names, 0, "prop_aliases('$mod_name') returns undef since $alias is not installed");
579 next;
580 }
581 elsif (! @all_names) {
582 fail("prop_aliases('$mod_name')");
583 diag("'$alias' is unknown to prop_aliases()");
584 next;
585 }
586
587 if ($count == 0) { # Is short name
588
589 @names_via_short = prop_aliases($mod_name);
590
591 # If the 0th test fails, no sense in continuing with the others
592 last unless is($names_via_short[0], $alias,
593 "prop_aliases: '$alias' is the short name for '$mod_name'");
594 $short_name = $alias;
595 }
596 elsif ($count == 1) { # Is full name
597
598 # Some properties have the same short and full name; no sense
599 # repeating the test if the same.
600 if ($alias ne $short_name) {
601 my @names_via_full = prop_aliases($mod_name);
602 is_deeply(\@names_via_full, \@names_via_short, "prop_aliases() returns the same list for both '$short_name' and '$mod_name'");
603 }
604
605 # Tests scalar context
606 is(prop_aliases($short_name), $alias,
607 "prop_aliases: '$alias' is the long name for '$short_name'");
608 }
609 else { # Is another alias
610 is_deeply(\@all_names, \@names_via_short, "prop_aliases() returns the same list for both '$short_name' and '$mod_name'");
611 ok((grep { $_ =~ /^$alias$/i } @all_names),
612 "prop_aliases: '$alias' is listed as an alias for '$mod_name'");
613 }
614
615 $count++;
616 }
617}
618
619# Now test anything we can find that wasn't covered by the tests of the
620# official properties. We have no way of knowing if mktables omitted a Perl
621# extension or not, but we do the best we can from its generated lists
622
623foreach my $alias (keys %utf8::loose_to_file_of) {
624 next if $alias =~ /=/;
625 my $lc_name = lc $alias;
e72b6605 626 my $loose = &utf8::_loose_name($lc_name);
7ef25837
KW
627 next if exists $props{$loose}; # Skip if already tested
628 $props{$loose} = 1;
629 my $mod_name = "$extra_chars$alias"; # Tests loose matching
630 my @aliases = prop_aliases($mod_name);
e72b6605 631 my $found_it = grep { &utf8::_loose_name(lc $_) eq $lc_name } @aliases;
7ef25837
KW
632 if ($found_it) {
633 pass("prop_aliases: '$lc_name' is listed as an alias for '$mod_name'");
634 }
635 elsif ($lc_name =~ /l[_&]$/) {
636
637 # These two names are special in that they don't appear in the
638 # returned list because they are discouraged from use. Verify
639 # that they return the same list as a non-discouraged version.
640 my @LC = prop_aliases('Is_LC');
641 is_deeply(\@aliases, \@LC, "prop_aliases: '$lc_name' returns the same list as 'Is_LC'");
642 }
643 else {
644 my $stripped = $lc_name =~ s/^is//;
645
646 # Could be that the input includes a prefix 'is', which is rarely
647 # returned as an alias, so having successfully stripped it off above,
648 # try again.
649 if ($stripped) {
e72b6605 650 $found_it = grep { &utf8::_loose_name(lc $_) eq $lc_name } @aliases;
7ef25837
KW
651 }
652
653 # If that didn't work, it could be that it's a block, which is always
654 # returned with a leading 'In_' to avoid ambiguity. Try comparing
655 # with that stripped off.
656 if (! $found_it) {
e72b6605 657 $found_it = grep { &utf8::_loose_name(s/^In_(.*)/\L$1/r) eq $lc_name }
7ef25837
KW
658 @aliases;
659 # Could check that is a real block, but tests for invmap will
660 # likely pickup any errors, since this will be tested there.
661 $lc_name = "in$lc_name" if $found_it; # Change for message below
662 }
663 my $message = "prop_aliases: '$lc_name' is listed as an alias for '$mod_name'";
664 ($found_it) ? pass($message) : fail($message);
665 }
666}
667
668my $done_equals = 0;
669foreach my $alias (keys %utf8::stricter_to_file_of) {
670 if ($alias =~ /=/) { # Only test one case where there is an equals
671 next if $done_equals;
672 $done_equals = 1;
673 }
674 my $lc_name = lc $alias;
675 my @list = prop_aliases($alias);
676 if ($alias =~ /^_/) {
677 is(@list, 0, "prop_aliases: '$lc_name' returns an empty list since it is internal_only");
678 }
679 elsif ($alias =~ /=/) {
680 is(@list, 0, "prop_aliases: '$lc_name' returns an empty list since is illegal property name");
681 }
682 else {
683 ok((grep { lc $_ eq $lc_name } @list),
684 "prop_aliases: '$lc_name' is listed as an alias for '$alias'");
685 }
686}
687
688use Unicode::UCD qw(prop_value_aliases);
689
690is(prop_value_aliases("unknown property", "unknown value"), undef,
691 "prop_value_aliases(<unknown property>, <unknown value>) returns <undef>");
692is(prop_value_aliases(undef, undef), undef,
693 "prop_value_aliases(undef, undef) returns <undef>");
694is((prop_value_aliases("na", "A")), "A", "test that prop_value_aliases returns its input for properties that don't have synonyms");
695is(prop_value_aliases("isgc", "C"), undef, "prop_value_aliases('isgc', 'C') returns <undef> since is not covered Perl extension");
696is(prop_value_aliases("gc", "isC"), undef, "prop_value_aliases('gc', 'isC') returns <undef> since is not covered Perl extension");
697
698# We have no way of knowing if mktables omitted a Perl extension that it
699# shouldn't have, but we can check if it omitted an official Unicode property
700# name synonym. And for those, we can check if the short and full names are
701# correct.
702
703my %pva_tested; # List of things already tested.
704open my $propvalues, "<", "../lib/unicore/PropValueAliases.txt"
705 or die "Can't open Unicode PropValueAliases.txt";
706while (<$propvalues>) {
707 s/\s*#.*//; # Remove comments
708 next if /^\s* $/x; # Ignore empty and comment lines
709 chomp;
710
711 my @fields = split /\s*;\s*/; # Fields are separated by semi-colons
712 my $prop = shift @fields; # 0th field is the property,
713 my $count = 0; # 0th field in line (after shifting off the property) is
714 # short name; 1th is long name
715 my $short_name;
716 my @names_via_short; # Saves the values between iterations
717
718 # The property on the lhs of the = is always loosely matched. Add in
719 # characters that are ignored under loose matching to test that
720 my $mod_prop = "$extra_chars$prop";
721
722 if ($fields[0] eq 'n/a') { # See comments in input file, essentially
723 # means full name and short name are identical
724 $fields[0] = $fields[1];
725 }
726 elsif ($fields[0] ne $fields[1]
e72b6605
KW
727 && &utf8::_loose_name(lc $fields[0])
728 eq &utf8::_loose_name(lc $fields[1])
7ef25837
KW
729 && $fields[1] !~ /[[:upper:]]/)
730 {
731 # Also, there is a bug in the file in which "n/a" is omitted, and
732 # the two fields are identical except for case, and the full name
733 # is all lower case. Copy the "short" name unto the full one to
734 # give it some upper case.
735
736 $fields[1] = $fields[0];
737 }
738
739 # The ccc property in the file is special; has an extra numeric field
740 # (0th), which should go at the end, since we use the next two fields as
741 # the short and full names, respectively. See comments in input file.
742 splice (@fields, 0, 0, splice(@fields, 1, 2)) if $prop eq 'ccc';
743
e72b6605 744 my $loose_prop = &utf8::_loose_name(lc $prop);
7ef25837
KW
745 my $suppressed = grep { $_ eq $loose_prop }
746 @Unicode::UCD::suppressed_properties;
747 foreach my $value (@fields) {
748 if ($suppressed) {
749 is(prop_value_aliases($prop, $value), undef, "prop_value_aliases('$prop', '$value') returns undef for suppressed property $prop");
750 next;
751 }
e72b6605 752 elsif (grep { $_ eq ("$loose_prop=" . &utf8::_loose_name(lc $value)) } @Unicode::UCD::suppressed_properties) {
7ef25837
KW
753 is(prop_value_aliases($prop, $value), undef, "prop_value_aliases('$prop', '$value') returns undef for suppressed property $prop=$value");
754 next;
755 }
756
757 # Add in test for loose matching.
758 my $mod_value = "$extra_chars$value";
759
760 # If the value is a number, optionally negative, including a floating
761 # point or rational numer, it should be only strictly matched, so the
762 # loose matching should fail.
763 if ($value =~ / ^ -? \d+ (?: [\/.] \d+ )? $ /x) {
764 is(prop_value_aliases($mod_prop, $mod_value), undef, "prop_value_aliases('$mod_prop', '$mod_value') returns undef because '$mod_value' should be strictly matched");
765
766 # And reset so below tests just the strict matching.
767 $mod_value = $value;
768 }
769
770 if ($count == 0) {
771
772 @names_via_short = prop_value_aliases($mod_prop, $mod_value);
773
774 # If the 0th test fails, no sense in continuing with the others
775 last unless is($names_via_short[0], $value, "prop_value_aliases: In '$prop', '$value' is the short name for '$mod_value'");
776 $short_name = $value;
777 }
778 elsif ($count == 1) {
779
780 # Some properties have the same short and full name; no sense
781 # repeating the test if the same.
782 if ($value ne $short_name) {
783 my @names_via_full =
784 prop_value_aliases($mod_prop, $mod_value);
785 is_deeply(\@names_via_full, \@names_via_short, "In '$prop', prop_value_aliases() returns the same list for both '$short_name' and '$mod_value'");
786 }
787
788 # Tests scalar context
789 is(prop_value_aliases($prop, $short_name), $value, "'$value' is the long name for prop_value_aliases('$prop', '$short_name')");
790 }
791 else {
792 my @all_names = prop_value_aliases($mod_prop, $mod_value);
793 is_deeply(\@all_names, \@names_via_short, "In '$prop', prop_value_aliases() returns the same list for both '$short_name' and '$mod_value'");
e72b6605 794 ok((grep { &utf8::_loose_name(lc $_) eq &utf8::_loose_name(lc $value) } prop_value_aliases($prop, $short_name)), "'$value' is listed as an alias for prop_value_aliases('$prop', '$short_name')");
7ef25837
KW
795 }
796
e72b6605 797 $pva_tested{&utf8::_loose_name(lc $prop) . "=" . &utf8::_loose_name(lc $value)} = 1;
7ef25837
KW
798 $count++;
799 }
800}
801
802# And test as best we can, the non-official pva's that mktables generates.
803foreach my $hash (\%utf8::loose_to_file_of, \%utf8::stricter_to_file_of) {
804 foreach my $test (keys %$hash) {
805 next if exists $pva_tested{$test}; # Skip if already tested
806
807 my ($prop, $value) = split "=", $test;
808 next unless defined $value; # prop_value_aliases() requires an input
809 # 'value'
810 my $mod_value;
811 if ($hash == \%utf8::loose_to_file_of) {
812
813 # Add extra characters to test loose-match rhs value
814 $mod_value = "$extra_chars$value";
815 }
816 else { # Here value is strictly matched.
817
818 # Extra elements are added by mktables to this hash so that
819 # something like "age=6.0" has a synonym of "age=6". It's not
820 # clear to me (khw) if we should be encouraging those synonyms, so
821 # don't test for them.
822 next if $value !~ /\D/ && exists $hash->{"$prop=$value.0"};
823
824 # Verify that loose matching fails when only strict is called for.
825 next unless is(prop_value_aliases($prop, "$extra_chars$value"), undef,
826 "prop_value_aliases('$prop', '$extra_chars$value') returns undef since '$value' should be strictly matched"),
827
828 # Strict matching does allow for underscores between digits. Test
829 # for that.
830 $mod_value = $value;
831 while ($mod_value =~ s/(\d)(\d)/$1_$2/g) {}
832 }
833
834 # The lhs property is always loosely matched, so add in extra
835 # characters to test that.
836 my $mod_prop = "$extra_chars$prop";
837
838 if ($prop eq 'gc' && $value =~ /l[_&]$/) {
839 # These two names are special in that they don't appear in the
840 # returned list because they are discouraged from use. Verify
841 # that they return the same list as a non-discouraged version.
842 my @LC = prop_value_aliases('gc', 'lc');
843 my @l_ = prop_value_aliases($mod_prop, $mod_value);
844 is_deeply(\@l_, \@LC, "prop_value_aliases('$mod_prop', '$mod_value) returns the same list as prop_value_aliases('gc', 'lc')");
845 }
846 else {
e72b6605 847 ok((grep { &utf8::_loose_name(lc $_) eq &utf8::_loose_name(lc $value) }
7ef25837
KW
848 prop_value_aliases($mod_prop, $mod_value)),
849 "'$value' is listed as an alias for prop_value_aliases('$mod_prop', '$mod_value')");
850 }
851 }
852}
853
854undef %pva_tested;
855
681d705c
KW
856no warnings 'once'; # We use some values once from 'required' modules.
857
62b3b855
KW
858use Unicode::UCD qw(prop_invlist prop_invmap MAX_CP);
859
860# There were some problems with caching interfering with prop_invlist() vs
861# prop_invmap() on binary properties, and also between the 3 properties where
862# Perl used the same 'To' name as another property (see utf8_heavy.pl).
863# So, before testing all of prop_invlist(),
864# 1) call prop_invmap() to try both orders of these name issues. This uses
865# up two of the 3 properties; the third will be left so that invlist()
866# on it gets called before invmap()
867# 2) call prop_invmap() on a generic binary property, ahead of invlist().
868# This should test that the caching works in both directions.
869
870# These properties are not stable between Unicode versions, but the first few
871# elements are; just look at the first element to see if are getting the
872# distinction right. The general inversion map testing below will test the
873# whole thing.
874my $prop = "uc";
875my ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
d11155ec 876is($format, 'al', "prop_invmap() format of '$prop' is 'al'");
bf7fe2df 877is($missing, '0', "prop_invmap() missing of '$prop' is '0'");
62b3b855 878is($invlist_ref->[1], 0x61, "prop_invmap('$prop') list[1] is 0x61");
d11155ec 879is($invmap_ref->[1], 0x41, "prop_invmap('$prop') map[1] is 0x41");
62b3b855
KW
880
881$prop = "upper";
882($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
d11155ec
KW
883is($format, 's', "prop_invmap() format of '$prop' is 's");
884is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
62b3b855
KW
885is($invlist_ref->[1], 0x41, "prop_invmap('$prop') list[1] is 0x41");
886is($invmap_ref->[1], 'Y', "prop_invmap('$prop') map[1] is 'Y'");
887
888$prop = "lower";
889($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
d11155ec
KW
890is($format, 's', "prop_invmap() format of '$prop' is 's'");
891is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
62b3b855
KW
892is($invlist_ref->[1], 0x61, "prop_invmap('$prop') list[1] is 0x61");
893is($invmap_ref->[1], 'Y', "prop_invmap('$prop') map[1] is 'Y'");
894
895$prop = "lc";
896($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
d11155ec 897is($format, 'al', "prop_invmap() format of '$prop' is 'al'");
bf7fe2df 898is($missing, '0', "prop_invmap() missing of '$prop' is '0'");
62b3b855 899is($invlist_ref->[1], 0x41, "prop_invmap('$prop') list[1] is 0x41");
d11155ec 900is($invmap_ref->[1], 0x61, "prop_invmap('$prop') map[1] is 0x61");
62b3b855
KW
901
902# This property is stable and small, so can test all of it
903$prop = "ASCII_Hex_Digit";
904($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
905is($format, 's', "prop_invmap() format of '$prop' is 's'");
906is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
907is_deeply($invlist_ref, [ 0x0000, 0x0030, 0x003A, 0x0041,
908 0x0047, 0x0061, 0x0067, 0x110000 ],
909 "prop_invmap('$prop') code point list is correct");
910is_deeply($invmap_ref, [ 'N', 'Y', 'N', 'Y', 'N', 'Y', 'N', 'N' ] ,
911 "prop_invmap('$prop') map list is correct");
681d705c
KW
912
913is(prop_invlist("Unknown property"), undef, "prop_invlist(<Unknown property>) returns undef");
914is(prop_invlist(undef), undef, "prop_invlist(undef) returns undef");
915is(prop_invlist("Any"), 2, "prop_invlist('Any') returns the number of elements in scalar context");
916my @invlist = prop_invlist("Is_Any");
917is_deeply(\@invlist, [ 0, 0x110000 ], "prop_invlist works on 'Is_' prefixes");
918is(prop_invlist("Is_Is_Any"), undef, "prop_invlist('Is_Is_Any') returns <undef> since two is's");
919
920use Storable qw(dclone);
921
922is(prop_invlist("InKana"), undef, "prop_invlist(<user-defined property returns undef>)");
923
62b3b855 924# The way both the tests for invlist and invmap work is that they take the
681d705c
KW
925# lists returned by the functions and construct from them what the original
926# file should look like, which are then compared with the file. If they are
927# identical, the test passes. What this tests isn't that the results are
62b3b855 928# correct, but that invlist and invmap haven't introduced errors beyond what
681d705c
KW
929# are there in the files. As a small hedge against that, test some
930# prop_invlist() tables fully with the known correct result. We choose
931# ASCII_Hex_Digit again, as it is stable.
932@invlist = prop_invlist("AHex");
933is_deeply(\@invlist, [ 0x0030, 0x003A, 0x0041,
934 0x0047, 0x0061, 0x0067 ],
935 "prop_invlist('AHex') is exactly the expected set of points");
936@invlist = prop_invlist("AHex=f");
937is_deeply(\@invlist, [ 0x0000, 0x0030, 0x003A, 0x0041,
938 0x0047, 0x0061, 0x0067 ],
939 "prop_invlist('AHex=f') is exactly the expected set of points");
940
941sub fail_with_diff ($$$$) {
942 # For use below to output better messages
943 my ($prop, $official, $constructed, $tested_function_name) = @_;
944
945 is($constructed, $official, "$tested_function_name('$prop')");
946 diag("Comment out lines " . (__LINE__ - 1) . " through " . (__LINE__ + 1) . " in '$0' on Un*x-like systems to see just the differences. Uses the 'diff' first in your \$PATH");
947 return;
948
949 fail("$tested_function_name('$prop')");
950
951 require File::Temp;
952 my $off = File::Temp->new();
953 chomp $official;
954 print $off $official, "\n";
955 close $off || die "Can't close official";
956
957 chomp $constructed;
958 my $gend = File::Temp->new();
959 print $gend $constructed, "\n";
960 close $gend || die "Can't close gend";
961
962 my $diff = File::Temp->new();
963 system("diff $off $gend > $diff");
964
965 open my $fh, "<", $diff || die "Can't open $diff";
966 my @diffs = <$fh>;
967 diag("In the diff output below '<' marks lines from the filesystem tables;\n'>' are from $tested_function_name()");
968 diag(@diffs);
969}
970
971my %tested_invlist;
972
973# Look at everything we think that mktables tells us exists, both loose and
974# strict
975foreach my $set_of_tables (\%utf8::stricter_to_file_of, \%utf8::loose_to_file_of)
976{
977 foreach my $table (keys %$set_of_tables) {
978
979 my $mod_table;
980 my ($prop_only, $value) = split "=", $table;
981 if (defined $value) {
982
983 # If this is to be loose matched, add in characters to test that.
984 if ($set_of_tables == \%utf8::loose_to_file_of) {
985 $value = "$extra_chars$value";
986 }
987 else { # Strict match
988
989 # Verify that loose matching fails when only strict is called
990 # for.
991 next unless is(prop_invlist("$prop_only=$extra_chars$value"), undef, "prop_invlist('$prop_only=$extra_chars$value') returns undef since should be strictly matched");
992
993 # Strict matching does allow for underscores between digits.
994 # Test for that.
995 while ($value =~ s/(\d)(\d)/$1_$2/g) {}
996 }
997
998 # The property portion in compound form specifications always
999 # matches loosely
1000 $mod_table = "$extra_chars$prop_only = $value";
1001 }
1002 else { # Single-form.
1003
1004 # Like above, use looose if required, and insert underscores
1005 # between digits if strict.
1006 if ($set_of_tables == \%utf8::loose_to_file_of) {
1007 $mod_table = "$extra_chars$table";
1008 }
1009 else {
1010 $mod_table = $table;
1011 while ($mod_table =~ s/(\d)(\d)/$1_$2/g) {}
1012 }
1013 }
1014
1015 my @tested = prop_invlist($mod_table);
1016 if ($table =~ /^_/) {
1017 is(@tested, 0, "prop_invlist('$mod_table') returns an empty list since is internal-only");
1018 next;
1019 }
1020
1021 # If we have already tested a property that uses the same file, this
1022 # list should be identical to the one that was tested, and can bypass
1023 # everything else.
1024 my $file = $set_of_tables->{$table};
1025 if (exists $tested_invlist{$file}) {
1026 is_deeply(\@tested, $tested_invlist{$file}, "prop_invlist('$mod_table') gave same results as its name synonym");
1027 next;
1028 }
1029 $tested_invlist{$file} = dclone \@tested;
1030
1031 # A leading '!' in the file name means that it is to be inverted.
1032 my $invert = $file =~ s/^!//;
1033 my $official = do "unicore/lib/$file.pl";
1034
1035 # Get rid of any trailing space and comments in the file.
1036 $official =~ s/\s*(#.*)?$//mg;
1037 chomp $official;
1038
1039 # If we are to test against an inverted file, it is easier to invert
1040 # our array than the file.
1041 # The file only is valid for Unicode code points, while the inversion
1042 # list is valid for all possible code points. Therefore, we must test
1043 # just the Unicode part against the file. Later we will test for
1044 # the non-Unicode part.
1045
1046 my $before_invert; # Saves the pre-inverted table.
1047 if ($invert) {
1048 $before_invert = dclone \@tested;
1049 if (@tested && $tested[0] == 0) {
1050 shift @tested;
1051 } else {
1052 unshift @tested, 0;
1053 }
1054 if (@tested && $tested[-1] == 0x110000) {
1055 pop @tested;
1056 }
1057 else {
1058 push @tested, 0x110000;
1059 }
1060 }
1061
1062 # Now construct a string from the list that should match the file.
1063 # The file gives ranges of code points with starting and ending values
1064 # in hex, like this:
1065 # 0041\t005A
1066 # 0061\t007A
1067 # 00AA
1068 # Our list has even numbered elements start ranges that are in the
1069 # list, and odd ones that aren't in the list. Therefore the odd
1070 # numbered ones are one beyond the end of the previous range, but
1071 # otherwise don't get reflected in the file.
1072 my $tested = "";
1073 my $i = 0;
1074 for (; $i < @tested - 1; $i += 2) {
1075 my $start = $tested[$i];
1076 my $end = $tested[$i+1] - 1;
1077 if ($start == $end) {
1078 $tested .= sprintf("%04X\n", $start);
1079 }
1080 else {
1081 $tested .= sprintf "%04X\t%04X\n", $start, $end;
1082 }
1083 }
1084
1085 # As mentioned earlier, the disk files only go up through Unicode,
1086 # whereas the prop_invlist() ones go as high as necessary. The
1087 # comparison is only valid through max Unicode.
1088 if ($i == @tested - 1 && $tested[$i] <= 0x10FFFF) {
1089 $tested .= sprintf("%04X\t10FFFF\n", $tested[$i]);
1090 }
1091 chomp $tested;
1092 if ($tested ne $official) {
1093 fail_with_diff($mod_table, $official, $tested, "prop_invlist");
1094 next;
1095 }
1096
1097 # Here, it matched the table. Now need to check for if it is correct
1098 # for beyond Unicode. First, calculate if is the default table or
1099 # not. This is the same algorithm as used internally in
1100 # prop_invlist(), so if it is wrong there, this test won't catch it.
1101 my $prop = lc $table;
1102 ($prop_only, $table) = split /\s*[:=]\s*/, $prop;
1103 if (defined $table) {
1104
1105 # May have optional prefixed 'is'
e72b6605 1106 $prop = &utf8::_loose_name($prop_only) =~ s/^is//r;
681d705c 1107 $prop = $utf8::loose_property_name_of{$prop};
e72b6605 1108 $prop .= "=" . &utf8::_loose_name($table);
681d705c
KW
1109 }
1110 else {
e72b6605 1111 $prop = &utf8::_loose_name($prop);
681d705c
KW
1112 }
1113 my $is_default = exists $Unicode::UCD::loose_defaults{$prop};
1114
1115 @tested = @$before_invert if $invert; # Use the original
1116 if (@tested % 2 == 0) {
1117
1118 # If there are an even number of elements, the final one starts a
1119 # range (going to infinity) of code points that are not in the
1120 # list.
1121 if ($is_default) {
1122 fail("prop_invlist('$mod_table')");
1123 diag("default table doesn't goto infinity");
1124 use Data::Dumper;
1125 diag Dumper \@tested;
1126 next;
1127 }
1128 }
1129 else {
1130 # An odd number of elements means the final one starts a range
1131 # (going to infinity of code points that are in the list.
1132 if (! $is_default) {
1133 fail("prop_invlist('$mod_table')");
1134 diag("non-default table needs to stop in the Unicode range");
1135 use Data::Dumper;
1136 diag Dumper \@tested;
1137 next;
1138 }
1139 }
1140
1141 pass("prop_invlist('$mod_table')");
1142 }
1143}
1144
62b3b855
KW
1145# Now test prop_invmap().
1146
1147@list = prop_invmap("Unknown property");
1148is (@list, 0, "prop_invmap(<Unknown property>) returns an empty list");
1149@list = prop_invmap(undef);
1150is (@list, 0, "prop_invmap(undef) returns an empty list");
1151ok (! eval "prop_invmap('gc')" && $@ ne "",
1152 "prop_invmap('gc') dies in scalar context");
1153@list = prop_invmap("_X_Begin");
1154is (@list, 0, "prop_invmap(<internal property>) returns an empty list");
1155@list = prop_invmap("InKana");
1156is(@list, 0, "prop_invmap(<user-defined property returns undef>)");
1157@list = prop_invmap("Perl_Decomposition_Mapping"), undef,
1158is(@list, 0, "prop_invmap('Perl_Decomposition_Mapping') returns <undef> since internal-Perl-only");
1159@list = prop_invmap("Perl_Charnames"), undef,
1160is(@list, 0, "prop_invmap('Perl_Charnames') returns <undef> since internal-Perl-only");
1161@list = prop_invmap("Is_Is_Any");
1162is(@list, 0, "prop_invmap('Is_Is_Any') returns <undef> since two is's");
1163
1164# The set of properties to test on has already been compiled into %props by
1165# the prop_aliases() tests.
1166
1167my %tested_invmaps;
1168
1169# Like prop_invlist(), prop_invmap() is tested by comparing the results
1170# returned by the function with the tables that mktables generates. Some of
1171# these tables are directly stored as files on disk, in either the unicore or
1172# unicore/To directories, and most should be listed in the mktables generated
1173# hash %utf8::loose_property_to_file_of, with a few additional ones that this
1174# handles specially. For these, the files are read in directly, massaged, and
1175# compared with what invmap() returns. The SPECIALS hash in some of these
1176# files overrides values in the main part of the file.
1177#
1178# The other properties are tested indirectly by generating all the possible
1179# inversion lists for the property, and seeing if those match the inversion
1180# lists returned by prop_invlist(), which has already been tested.
1181
1182PROPERTY:
1183foreach my $prop (keys %props) {
e72b6605 1184 my $loose_prop = &utf8::_loose_name(lc $prop);
62b3b855
KW
1185 my $suppressed = grep { $_ eq $loose_prop }
1186 @Unicode::UCD::suppressed_properties;
1187
1188 # Find the short and full names that this property goes by
1189 my ($name, $full_name) = prop_aliases($prop);
1190 if (! $name) {
1191 if (! $suppressed) {
1192 fail("prop_invmap('$prop')");
1193 diag("is unknown to prop_aliases(), and we need it in order to test prop_invmap");
1194 }
1195 next PROPERTY;
1196 }
1197
1198 # Normalize the short name, as it is stored in the hashes under the
1199 # normalized version.
e72b6605 1200 $name = &utf8::_loose_name(lc $name);
62b3b855
KW
1201
1202 # Add in the characters that are supposed to be ignored to test loose
1203 # matching, which the tested function applies to all properties
1204 my $mod_prop = "$extra_chars$prop";
1205
1206 my ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($mod_prop);
1207 my $return_ref = [ $invlist_ref, $invmap_ref, $format, $missing ];
1208
1209 # If have already tested this property under a different name, merely
1210 # compare the return from now with the saved one from before.
1211 if (exists $tested_invmaps{$name}) {
1212 is_deeply($return_ref, $tested_invmaps{$name}, "prop_invmap('$mod_prop') gave same results as its synonym, '$name'");
1213 next PROPERTY;
1214 }
1215 $tested_invmaps{$name} = dclone $return_ref;
1216
1217 # If prop_invmap() returned nothing, is ok iff is a property whose file is
1218 # not generated.
1219 if ($suppressed) {
1220 if (defined $format) {
1221 fail("prop_invmap('$mod_prop')");
1222 diag("did not return undef for suppressed property $prop");
1223 }
1224 next PROPERTY;
1225 }
1226 elsif (!defined $format) {
1227 fail("prop_invmap('$mod_prop')");
1228 diag("'$prop' is unknown to prop_invmap()");
1229 next PROPERTY;
1230 }
1231
1232 # The two parallel arrays must have the same number of elements.
1233 if (@$invlist_ref != @$invmap_ref) {
1234 fail("prop_invmap('$mod_prop')");
1235 diag("invlist has "
1236 . scalar @$invlist_ref
1237 . " while invmap has "
1238 . scalar @$invmap_ref
1239 . " elements");
1240 next PROPERTY;
1241 }
1242
1243 # The last element must be for the above-Unicode code points, and must be
1244 # for the default value.
1245 if ($invlist_ref->[-1] != 0x110000) {
1246 fail("prop_invmap('$mod_prop')");
1247 diag("The last inversion list element is not 0x110000");
1248 next PROPERTY;
1249 }
1250 if ($invmap_ref->[-1] ne $missing) {
1251 fail("prop_invmap('$mod_prop')");
1252 diag("The last inversion list element is '$invmap_ref->[-1]', and should be '$missing'");
1253 next PROPERTY;
1254 }
1255
1256 if ($name eq 'bmg') { # This one has an atypical $missing
1257 if ($missing ne "") {
1258 fail("prop_invmap('$mod_prop')");
1259 diag("The missings should be \"\"; got '$missing'");
1260 next PROPERTY;
1261 }
1262 }
4f143a72 1263 elsif ($format =~ /^ a (?!r) /x) {
b0b13ada
KW
1264 if ($full_name eq 'Perl_Decimal_Digit') {
1265 if ($missing ne "") {
1266 fail("prop_invmap('$mod_prop')");
1267 diag("The missings should be \"\"; got '$missing'");
1268 next PROPERTY;
1269 }
1270 }
1271 elsif ($missing ne "0") {
bf7fe2df
KW
1272 fail("prop_invmap('$mod_prop')");
1273 diag("The missings should be '0'; got '$missing'");
1274 next PROPERTY;
1275 }
1276 }
62b3b855
KW
1277 elsif ($missing =~ /[<>]/) {
1278 fail("prop_invmap('$mod_prop')");
1279 diag("The missings should NOT be something with <...>'");
1280 next PROPERTY;
1281
1282 # I don't want to hard code in what all the missings should be, so
1283 # those don't get fully tested.
1284 }
1285
1286 # Certain properties don't have their own files, but must be constructed
1287 # using proxies.
1288 my $proxy_prop = $name;
1289 if ($full_name eq 'Present_In') {
1290 $proxy_prop = "age"; # The maps for these two props are identical
1291 }
1292 elsif ($full_name eq 'Simple_Case_Folding'
1293 || $full_name =~ /Simple_ (.) .*? case_Mapping /x)
1294 {
1295 if ($full_name eq 'Simple_Case_Folding') {
1296 $proxy_prop = 'cf';
1297 }
1298 else {
1299 # We captured the U, L, or T, leading to uc, lc, or tc.
1300 $proxy_prop = lc $1 . "c";
1301 }
d11155ec 1302 if ($format ne "a") {
62b3b855 1303 fail("prop_invmap('$mod_prop')");
d11155ec 1304 diag("The format should be 'a'; got '$format'");
62b3b855
KW
1305 next PROPERTY;
1306 }
1307 }
1308
1309 my $base_file;
1310 my $official;
1311
1312 # Handle the properties that have full disk files for them (except the
1313 # Name property which is structurally enough different that it is handled
1314 # separately below.)
1315 if ($name ne 'na'
1316 && ($name eq 'blk'
1317 || defined
1318 ($base_file = $utf8::loose_property_to_file_of{$proxy_prop})
1319 || exists $utf8::loose_to_file_of{$proxy_prop}
1320 || $name eq "dm"))
1321 {
1322 # In the above, blk is done unconditionally, as we need to test that
1323 # the old-style block names are returned, even if mktables has
1324 # generated a file for the new-style; the test for dm comes afterward,
1325 # so that if a file has been generated for it explicitly, we use that
1326 # file (which is valid, unlike blk) instead of the combo
1327 # Decomposition.pl files.
1328 my $file;
1329 my $is_binary = 0;
1330 if ($name eq 'blk') {
1331
1332 # The blk property is special. The original file with old block
1333 # names is retained, and the default is to not write out a
1334 # new-name file. What we do is get the old names into a data
1335 # structure, and from that create what the new file would look
1336 # like. $base_file is needed to be defined, just to avoid a
1337 # message below.
1338 $base_file = "This is a dummy name";
1339 my $blocks_ref = charblocks();
1340 $official = "";
1341 for my $range (sort { $a->[0][0] <=> $b->[0][0] }
1342 values %$blocks_ref)
1343 {
1344 # Translate the charblocks() data structure to what the file
1345 # would like.
1346 $official .= sprintf"%04X\t%04X\t%s\n",
1347 $range->[0][0],
1348 $range->[0][1],
1349 $range->[0][2];
1350 }
1351 }
1352 else {
d11155ec 1353 $base_file = "Decomposition" if $format eq 'ad';
62b3b855
KW
1354
1355 # Above leaves $base_file undefined only if it came from the hash
1356 # below. This should happen only when it is a binary property
1357 # (and are accessing via a single-form name, like 'In_Latin1'),
1358 # and so it is stored in a different directory than the To ones.
1359 # XXX Currently, the only cases where it is complemented are the
1360 # ones that have no code points. And it works out for these that
1361 # 1) complementing them, and then 2) adding or subtracting the
1362 # initial 0 and final 110000 cancel each other out. But further
1363 # work would be needed in the unlikely event that an inverted
1364 # property comes along without these characteristics
1365 if (!defined $base_file) {
1366 $base_file = $utf8::loose_to_file_of{$proxy_prop};
1367 $is_binary = ($base_file =~ s/^!//) ? -1 : 1;
1368 $base_file = "lib/$base_file";
1369 }
1370
1371 # Read in the file
62b3b855
KW
1372 $file = "unicore/$base_file.pl";
1373 $official = do $file;
1374
1375 # Get rid of any trailing space and comments in the file.
1376 $official =~ s/\s*(#.*)?$//mg;
1377
d11155ec 1378 if ($format eq 'ad') {
bea2c146
KW
1379 my @official = split /\n/, $official;
1380 $official = "";
1381 foreach my $line (@official) {
1382 my ($start, $end, $value)
1383 = $line =~ / ^ (.+?) \t (.*?) \t (.+?)
1384 \s* ( \# .* )? $ /x;
1385 # Decomposition.pl also has the <compatible> types in it,
1386 # which should be removed.
1387 $value =~ s/<.*?> //;
1388 $official .= "$start\t\t$value\n";
1389
1390 # If this is a multi-char range, we turn it into as many
1391 # single character ranges as necessary. This makes things
1392 # easier below.
1393 if ($end ne "") {
1394 for my $i (hex($start) + 1 .. hex $end) {
1395 $official .= sprintf "%04X\t\t%s\n", $i, $value;
1396 }
1397 }
1398 }
1399 }
62b3b855
KW
1400 }
1401 chomp $official;
1402
1403 # If there are any special elements, get a reference to them.
bf7fe2df
KW
1404 my $swash_name = $utf8::file_to_swash_name{$base_file};
1405 my $specials_ref;
1406 if ($swash_name) {
1407 $specials_ref = $utf8::SwashInfo{$swash_name}{'specials_name'};
62b3b855
KW
1408 if ($specials_ref) {
1409
1410 # Convert from the name to the actual reference.
1411 no strict 'refs';
1412 $specials_ref = \%{$specials_ref};
1413 }
1414 }
1415
1416 # Certain of the proxy properties have to be adjusted to match the
1417 # real ones.
ae1bcb1f 1418 if ($full_name =~ /^(Case_Folding|(Lower|Title|Upper)case_Mapping)/) {
62b3b855
KW
1419
1420 # Here we have either
1421 # 1) Case_Folding; or
1422 # 2) a proxy that is a full mapping, which means that what the
1423 # real property is is the equivalent simple mapping.
1424 # In both cases, the file will have a standard list containing
1425 # simple mappings (to a single code point), and a specials hash
1426 # which contains all the mappings that are to multiple code
1427 # points. First, extract a list containing all the file's simple
1428 # mappings.
1429 my @list;
1430 for (split "\n", $official) {
1431 my ($start, $end, $value) = / ^ (.+?) \t (.*?) \t (.+?)
bea2c146 1432 \s* ( \# .* )? $ /x;
62b3b855 1433 $end = $start if $end eq "";
bf7fe2df 1434 push @list, [ hex $start, hex $end, $value ];
62b3b855
KW
1435 }
1436
ae1bcb1f 1437 # For these mappings, the file contains all the simple mappings,
62b3b855 1438 # including the ones that are overridden by the specials. These
ae1bcb1f 1439 # need to be removed as the list is for just the full ones.
62b3b855
KW
1440
1441 # Go through any special mappings one by one. They are packed.
1442 my $i = 0;
1443 foreach my $utf8_cp (sort keys %$specials_ref) {
1444 my $cp = unpack("C0U", $utf8_cp);
1445
62b3b855
KW
1446 # Find the spot in the @list of simple mappings that this
1447 # special applies to; uses a linear search.
1448 while ($i < @list -1 ) {
bea2c146 1449 last if $cp <= $list[$i][1];
62b3b855
KW
1450 $i++;
1451 }
1452
bea2c146
KW
1453 # Here $i is such that it points to the first range which ends
1454 # at or above cp, and hence is the only range that could
1455 # possibly contain it.
1456
1457 # If not in this range, no range contains it: nothing to
1458 # remove.
1459 next if $cp < $list[$i][0];
1460
1461 # Otherwise, remove the existing entry. If it is the first
1462 # element of the range...
1463 if ($cp == $list[$i][0]) {
62b3b855 1464
bea2c146
KW
1465 # ... and there are other elements in the range, just shorten
1466 # the range to exclude this code point.
1467 if ($list[$i][1] > $list[$i][0]) {
1468 $list[$i][0]++;
1469 }
62b3b855 1470
bea2c146
KW
1471 # ... but if it is the only element in the range, remove
1472 # it entirely.
1473 else {
1474 splice @list, $i, 1;
1475 }
1476 }
1477 else { # Is somewhere in the middle of the range
1478 # Split the range into two, excluding this one in the
1479 # middle
1480 splice @list, $i, 1,
1481 [ $list[$i][0], $cp - 1, $list[$i][2] ],
1482 [ $cp + 1, $list[$i][1], $list[$i][2] ];
1483 }
62b3b855
KW
1484 }
1485
1486 # Here, have gone through all the specials, modifying @list as
1487 # needed. Turn it back into what the file should look like.
bf7fe2df
KW
1488 $official = "";
1489 for my $element (@list) {
1490 $official .= "\n" if $official;
1491 if ($element->[1] == $element->[0]) {
1492 $official .= sprintf "%04X\t\t%s", $element->[0], $element->[2];
1493 }
1494 else {
1495 $official .= sprintf "%04X\t%04X\t%s", $element->[0], $element->[1], $element->[2];
1496 }
1497 }
62b3b855 1498 }
ae1bcb1f
KW
1499 elsif ($full_name =~ /Simple_(Case_Folding|(Lower|Title|Upper)case_Mapping)/)
1500 {
62b3b855 1501
ae1bcb1f 1502 # These properties have everything in the regular array, and the
62b3b855 1503 # specials are superfluous.
ae1bcb1f 1504 undef $specials_ref;
62b3b855 1505 }
bf7fe2df
KW
1506 elsif ($name eq 'bmg') {
1507
1508 # For this property, the file is output using hex notation for the
1509 # map, with all ranges equal to length 1. Convert from hex to
1510 # decimal.
1511 my @lines = split "\n", $official;
1512 foreach my $line (@lines) {
1513 my ($code_point, $map) = split "\t\t", $line;
1514 $line = $code_point . "\t\t" . hex $map;
1515 }
1516 $official = join "\n", @lines;
1517 }
62b3b855
KW
1518
1519 # Here, in $official, we have what the file looks like, or should like
1520 # if we've had to fix it up. Now take the invmap() output and reverse
1521 # engineer from that what the file should look like. Each iteration
1522 # appends the next line to the running string.
1523 my $tested_map = "";
1524
1525 # Create a copy of the file's specials hash. (It has been undef'd if
1526 # we know it isn't relevant to this property, so if it exists, it's an
1527 # error or is relevant). As we go along, we delete from that copy.
1528 # If a delete fails, or something is left over after we are done,
1529 # it's an error
1530 my %specials = %$specials_ref if $specials_ref;
1531
1532 # The extra -1 is because the final element has been tested above to
1533 # be for anything above Unicode. The file doesn't go that high.
bea2c146 1534 for (my $i = 0; $i < @$invlist_ref - 1; $i++) {
62b3b855
KW
1535
1536 # If the map element is a reference, have to stringify it (but
1537 # don't do so if the format doesn't allow references, so that an
1538 # improper format will generate an error.
1539 if (ref $invmap_ref->[$i]
d11155ec 1540 && ($format eq 'ad' || $format =~ /^ . l /x))
62b3b855 1541 {
3b6a8189 1542 # The stringification depends on the format.
62b3b855 1543 if ($format eq 'sl') {
3b6a8189
KW
1544
1545 # At the time of this writing, there are two types of 'sl'
1546 # format One, in Name_Alias, has multiple separate entries
1547 # for each code point; the other, in Script_Extension, is space
1548 # separated. Assume the latter for non-Name_Alias.
1549 if ($full_name ne 'Name_Alias') {
1550 $invmap_ref->[$i] = join " ", @{$invmap_ref->[$i]};
1551 }
1552 else {
1553 # For Name_Alias, we emulate the file. Entries with
1554 # just one value don't need any changes, but we
1555 # convert the list entries into a series of lines for
1556 # the file, starting with the first name. The
1557 # succeeding entries are on separate lines, with the
1558 # code point repeated for each one and then two tabs,
1559 # then the value. Code at the end of the loop will
1560 # set up the first line with its code point and two
1561 # tabs before the value, just as it does for every
1562 # other property; thus the special handling of the
1563 # first line.
1564 if (ref $invmap_ref->[$i]) {
1565 my $hex_cp = sprintf("%04X", $invlist_ref->[$i]);
1566 my $concatenated = $invmap_ref->[$i][0];
1567 for (my $j = 1; $j < @{$invmap_ref->[$i]}; $j++) {
1568 $concatenated .= "\n$hex_cp\t\t" . $invmap_ref->[$i][$j];
1569 }
1570 $invmap_ref->[$i] = $concatenated;
1571 }
1572 }
62b3b855 1573 }
d11155ec 1574 elsif ($format =~ / ^ al e? $/x) {
62b3b855 1575
d11155ec 1576 # For a al property, the stringified result should be in
62b3b855
KW
1577 # the specials hash. The key is the packed code point,
1578 # and the value is the packed map.
1579 my $value;
1580 if (! defined ($value = delete $specials{pack("C0U", $invlist_ref->[$i]) })) {
1581 fail("prop_invmap('$mod_prop')");
1582 diag(sprintf "There was no specials element for %04X", $invlist_ref->[$i]);
1583 next PROPERTY;
1584 }
1585 my $packed = pack "U*", @{$invmap_ref->[$i]};
1586 if ($value ne $packed) {
1587 fail("prop_invmap('$mod_prop')");
1588 diag(sprintf "For %04X, expected the mapping to be '$packed', but got '$value'");
1589 next PROPERTY;
1590 }
1591
1592 # As this doesn't get tested when we later compare with
1593 # the actual file, it could be out of order and we
1594 # wouldn't know it.
1595 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1596 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
1597 {
1598 fail("prop_invmap('$mod_prop')");
1599 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1600 next PROPERTY;
1601 }
1602 next;
1603 }
d11155ec 1604 elsif ($format eq 'ad') {
62b3b855
KW
1605
1606 # The decomposition mapping file has the code points as
1607 # a string of space-separated hex constants.
1608 $invmap_ref->[$i] = join " ", map { sprintf "%04X", $_ } @{$invmap_ref->[$i]};
1609 }
1610 else {
1611 fail("prop_invmap('$mod_prop')");
1612 diag("Can't handle format '$format'");
1613 next PROPERTY;
1614 }
1615 }
d11155ec 1616 elsif ($format eq 'ad' || $format eq 'ale') {
34132297 1617
d11155ec
KW
1618 # The numerics in the returned map are stored as adjusted
1619 # decimal integers. The defaults are 0, and don't appear in
1620 # $official, and are excluded later, but the elements must be
1621 # converted back to their hex values before comparing with
34132297 1622 # $official, as these files, for backwards compatibility, are
d11155ec 1623 # not stored as adjusted. (There currently is only one ale
34132297 1624 # property, nfkccf. If that changed this would also have to.)
bea2c146
KW
1625 if ($invmap_ref->[$i] =~ / ^ -? \d+ $ /x
1626 && $invmap_ref->[$i] != 0)
1627 {
d11155ec
KW
1628 my $next = $invmap_ref->[$i] + 1;
1629 $invmap_ref->[$i] = sprintf("%04X", $invmap_ref->[$i]);
1630
1631 # If there are other elements in this range they need to
1632 # be adjusted; they must individually be re-mapped. Do
1633 # this by splicing in a new element into the list and the
1634 # map containing the remainder of the range. Next time
1635 # through we will look at that (possibly splicing again
1636 # until the whole range is processed).
bea2c146
KW
1637 if ($invlist_ref->[$i+1] > $invlist_ref->[$i] + 1) {
1638 splice @$invlist_ref, $i+1, 0,
1639 $invlist_ref->[$i] + 1;
d11155ec 1640 splice @$invmap_ref, $i+1, 0, $next;
bea2c146
KW
1641 }
1642 }
d11155ec 1643 if ($format eq 'ale' && $invmap_ref->[$i] eq "") {
62b3b855 1644
d11155ec 1645 # ale properties have maps to the empty string that also
4066e594
KW
1646 # should be in the specials hash, with the key the packed
1647 # code point, and the map just empty.
1648 my $value;
1649 if (! defined ($value = delete $specials{pack("C0U", $invlist_ref->[$i]) })) {
1650 fail("prop_invmap('$mod_prop')");
1651 diag(sprintf "There was no specials element for %04X", $invlist_ref->[$i]);
1652 next PROPERTY;
1653 }
1654 if ($value ne "") {
1655 fail("prop_invmap('$mod_prop')");
1656 diag(sprintf "For %04X, expected the mapping to be \"\", but got '$value'", $invlist_ref->[$i]);
1657 next PROPERTY;
1658 }
62b3b855 1659
4066e594
KW
1660 # As this doesn't get tested when we later compare with
1661 # the actual file, it could be out of order and we
1662 # wouldn't know it.
1663 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1664 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
1665 {
1666 fail("prop_invmap('$mod_prop')");
1667 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1668 next PROPERTY;
1669 }
1670 next;
34132297 1671 }
62b3b855
KW
1672 }
1673 elsif ($is_binary) { # These binary files don't have an explicit Y
1674 $invmap_ref->[$i] =~ s/Y//;
1675 }
1676
1677 # The file doesn't include entries that map to $missing, so don't
1678 # include it in the built-up string. But make sure that it is in
1679 # the correct order in the input.
1680 if ($invmap_ref->[$i] eq $missing) {
1681 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1682 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
1683 {
1684 fail("prop_invmap('$mod_prop')");
1685 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1686 next PROPERTY;
1687 }
1688 next;
1689 }
1690
7f131b96
KW
1691 # The ad property has one entry which isn't in the file.
1692 # Ignore it, but make sure it is in order.
1693 if ($format eq 'ad'
1694 && $invmap_ref->[$i] eq '<hangul syllable>'
1695 && $invlist_ref->[$i] == 0xAC00)
1696 {
1697 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1698 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
62b3b855 1699 {
7f131b96
KW
1700 fail("prop_invmap('$mod_prop')");
1701 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1702 next PROPERTY;
62b3b855 1703 }
7f131b96
KW
1704 next;
1705 }
62b3b855
KW
1706
1707 # Finally have figured out what the map column in the file should
1708 # be. Append the line to the running string.
1709 my $start = $invlist_ref->[$i];
1710 my $end = $invlist_ref->[$i+1] - 1;
1711 $end = ($start == $end) ? "" : sprintf("%04X", $end);
1712 if ($invmap_ref->[$i] ne "") {
1713 $tested_map .= sprintf "%04X\t%s\t%s\n", $start, $end, $invmap_ref->[$i];
1714 }
1715 elsif ($end ne "") {
1716 $tested_map .= sprintf "%04X\t%s\n", $start, $end;
1717 }
1718 else {
1719 $tested_map .= sprintf "%04X\n", $start;
1720 }
1721 } # End of looping over all elements.
1722
1723 # Here are done with generating what the file should look like
1724
1725 chomp $tested_map;
1726
1727 # And compare.
1728 if ($tested_map ne $official) {
1729 fail_with_diff($mod_prop, $official, $tested_map, "prop_invmap");
1730 next PROPERTY;
1731 }
1732
1733 # There shouldn't be any specials unaccounted for.
1734 if (keys %specials) {
1735 fail("prop_invmap('$mod_prop')");
1736 diag("Unexpected specials: " . join ", ", keys %specials);
1737 next PROPERTY;
1738 }
1739 }
1740 elsif ($format eq 'n') {
1741
1742 # Handle the Name property similar to the above. But the file is
1743 # sufficiently different that it is more convenient to make a special
3b6a8189
KW
1744 # case for it. It is a combination of the Name, Unicode1_Name, and
1745 # Name_Alias properties, and named sequences. We need to remove all
1746 # but the Name in order to do the comparison.
62b3b855
KW
1747
1748 if ($missing ne "") {
1749 fail("prop_invmap('$mod_prop')");
1750 diag("The missings should be \"\"; got \"missing\"");
1751 next PROPERTY;
1752 }
1753
1754 $official = do "unicore/Name.pl";
1755
1756 # Get rid of the named sequences portion of the file. These don't
1757 # have a tab before the first blank on a line.
1758 $official =~ s/ ^ [^\t]+ \ .*? \n //xmg;
1759
1760 # And get rid of the controls. These are named in the file, but
3b6a8189
KW
1761 # shouldn't be in the property. This gets rid of the two ranges in
1762 # one fell swoop, and also all the Unicode1_Name values that may not
1763 # be in Name_Alias.
62b3b855
KW
1764 $official =~ s/ 00000 \t .* 0001F .*? \n//xs;
1765 $official =~ s/ 0007F \t .* 0009F .*? \n//xs;
1766
3b6a8189
KW
1767 # And remove the aliases. We read in the Name_Alias property, and go
1768 # through them one by one.
1769 my ($aliases_code_points, $aliases_maps, undef, undef)
1770 = &prop_invmap('Name_Alias');
1771 for (my $i = 0; $i < @$aliases_code_points; $i++) {
1772 my $code_point = $aliases_code_points->[$i];
1773
1774 # Already removed these above.
1775 next if $code_point <= 0x1F
1776 || ($code_point >= 0x7F && $code_point <= 0x9F);
1777
1778 my $hex_code_point = sprintf "%05X", $code_point;
1779
1780 # Convert to a list if not already to make the following loop
1781 # control uniform.
1782 $aliases_maps->[$i] = [ $aliases_maps->[$i] ]
1783 if ! ref $aliases_maps->[$i];
1784
1785 # Remove each alias for this code point from the file
1786 foreach my $alias (@{$aliases_maps->[$i]}) {
1787
1788 # Remove the alias type from the entry, retaining just the name.
1789 $alias =~ s/:.*//;
1790
1791 $alias = quotemeta($alias);
1792 $official =~ s/$hex_code_point \t $alias \n //x;
62b3b855
KW
1793 }
1794 }
62b3b855
KW
1795 chomp $official;
1796
1797 # Here have adjusted the file. We also have to adjust the returned
1798 # inversion map by checking and deleting all the lines in it that
1799 # won't be in the file. These are the lines that have generated
1800 # things, like <hangul syllable>.
1801 my $tested_map = ""; # Current running string
1802 my @code_point_in_names =
1803 @Unicode::UCD::code_points_ending_in_code_point;
1804
1805 for my $i (0 .. @$invlist_ref - 1 - 1) {
1806 my $start = $invlist_ref->[$i];
1807 my $end = $invlist_ref->[$i+1] - 1;
1808 if ($invmap_ref->[$i] eq $missing) {
1809 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1810 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
1811 {
1812 fail("prop_invmap('$mod_prop')");
1813 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1814 next PROPERTY;
1815 }
1816 next;
1817 }
1818 if ($invmap_ref->[$i] =~ / (.*) ( < .*? > )/x) {
1819 my $name = $1;
1820 my $type = $2;
1821 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
1822 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
1823 {
1824 fail("prop_invmap('$mod_prop')");
1825 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1826 next PROPERTY;
1827 }
1828 if ($type eq "<hangul syllable>") {
1829 if ($name ne "") {
1830 fail("prop_invmap('$mod_prop')");
1831 diag("Unexpected text in $invmap_ref->[$i]");
1832 next PROPERTY;
1833 }
1834 if ($start != 0xAC00) {
1835 fail("prop_invmap('$mod_prop')");
1836 diag(sprintf("<hangul syllables> should begin at 0xAC00, got %04X", $start));
1837 next PROPERTY;
1838 }
1839 if ($end != $start + 11172 - 1) {
1840 fail("prop_invmap('$mod_prop')");
1841 diag(sprintf("<hangul syllables> should end at %04X, got %04X", $start + 11172 -1, $end));
1842 next PROPERTY;
1843 }
1844 }
1845 elsif ($type ne "<code point>") {
1846 fail("prop_invmap('$mod_prop')");
1847 diag("Unexpected text '$type' in $invmap_ref->[$i]");
1848 next PROPERTY;
1849 }
1850 else {
1851
1852 # Look through the array of names that end in code points,
1853 # and look for this start and end. If not found is an
1854 # error. If found, delete it, and at the end, make sure
1855 # have deleted everything.
1856 for my $i (0 .. @code_point_in_names - 1) {
1857 my $hash = $code_point_in_names[$i];
1858 if ($hash->{'low'} == $start
1859 && $hash->{'high'} == $end
1860 && "$hash->{'name'}-" eq $name)
1861 {
1862 splice @code_point_in_names, $i, 1;
1863 last;
1864 }
1865 else {
1866 fail("prop_invmap('$mod_prop')");
1867 diag("Unexpected code-point-in-name line '$invmap_ref->[$i]'");
1868 next PROPERTY;
1869 }
1870 }
1871 }
1872
1873 next;
1874 }
1875
1876 # Have adjusted the map, as needed. Append to running string.
1877 $end = ($start == $end) ? "" : sprintf("%05X", $end);
1878 $tested_map .= sprintf "%05X\t%s\n", $start, $invmap_ref->[$i];
1879 }
1880
1881 # Finished creating the string from the inversion map. Can compare
1882 # with what the file is.
1883 chomp $tested_map;
1884 if ($tested_map ne $official) {
1885 fail_with_diff($mod_prop, $official, $tested_map, "prop_invmap");
1886 next PROPERTY;
1887 }
1888 if (@code_point_in_names) {
1889 fail("prop_invmap('$mod_prop')");
1890 use Data::Dumper;
1891 diag("Missing code-point-in-name line(s)" . Dumper \@code_point_in_names);
1892 next PROPERTY;
1893 }
1894 }
4f143a72 1895 elsif ($format eq 's') {
62b3b855
KW
1896
1897 # Here the map is not more or less directly from a file stored on
1898 # disk. We try a different tack. These should all be properties that
1899 # have just a few possible values (most of them are binary). We go
1900 # through the map list, sorting each range into buckets, one for each
1901 # map value. Thus for binary properties there will be a bucket for Y
1902 # and one for N. The buckets are inversion lists. We compare each
1903 # constructed inversion list with what we would get for it using
1904 # prop_invlist(), which has already been tested. If they all match,
1905 # the whole map must have matched.
1906 my %maps;
1907 my $previous_map;
1908
1909 # (The extra -1 is to not look at the final element in the loop, which
1910 # we know is the one that starts just beyond Unicode and goes to
1911 # infinity.)
1912 for my $i (0 .. @$invlist_ref - 1 - 1) {
1913 my $range_start = $invlist_ref->[$i];
1914
1915 # Because we are sorting into buckets, things could be
1916 # out-of-order here, and still be in the correct order in the
1917 # bucket, and hence wouldn't show up as an error; so have to
1918 # check.
1919 if (($i > 0 && $range_start <= $invlist_ref->[$i-1])
1920 || $range_start >= $invlist_ref->[$i+1])
1921 {
1922 fail("prop_invmap('$mod_prop')");
1923 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
1924 next PROPERTY;
1925 }
1926
1927 # This new range closes out the range started in the previous
1928 # iteration.
1929 push @{$maps{$previous_map}}, $range_start if defined $previous_map;
1930
1931 # And starts a range which will be closed in the next iteration.
1932 $previous_map = $invmap_ref->[$i];
1933 push @{$maps{$previous_map}}, $range_start;
1934 }
1935
1936 # The range we just started hasn't been closed, and we didn't look at
1937 # the final element of the loop. If that range is for the default
1938 # value, it shouldn't be closed, as it is to extend to infinity. But
1939 # otherwise, it should end at the final Unicode code point, and the
1940 # list that maps to the default value should have another element that
1941 # does go to infinity for every above Unicode code point.
1942
1943 if (@$invlist_ref > 1) {
1944 my $penultimate_map = $invmap_ref->[-2];
1945 if ($penultimate_map ne $missing) {
1946
1947 # The -1th element contains the first non-Unicode code point.
1948 push @{$maps{$penultimate_map}}, $invlist_ref->[-1];
1949 push @{$maps{$missing}}, $invlist_ref->[-1];
1950 }
1951 }
1952
1953 # Here, we have the buckets (inversion lists) all constructed. Go
1954 # through each and verify that matches what prop_invlist() returns.
1955 # We could use is_deeply() for the comparison, but would get multiple
1956 # messages for each $prop.
1957 foreach my $map (keys %maps) {
1958 my @off_invlist = prop_invlist("$prop = $map");
1959 my $min = (@off_invlist >= @{$maps{$map}})
1960 ? @off_invlist
1961 : @{$maps{$map}};
1962 for my $i (0 .. $min- 1) {
1963 if ($i > @off_invlist - 1) {
1964 fail("prop_invmap('$mod_prop')");
1965 diag("There is no element [$i] for $prop=$map from prop_invlist(), while [$i] in the implicit one constructed from prop_invmap() is '$maps{$map}[$i]'");
1966 next PROPERTY;
1967 }
1968 elsif ($i > @{$maps{$map}} - 1) {
1969 fail("prop_invmap('$mod_prop')");
1970 diag("There is no element [$i] from the implicit $prop=$map constructed from prop_invmap(), while [$i] in the one from prop_invlist() is '$off_invlist[$i]'");
1971 next PROPERTY;
1972 }
1973 elsif ($maps{$map}[$i] ne $off_invlist[$i]) {
1974 fail("prop_invmap('$mod_prop')");
1975 diag("Element [$i] of the implicit $prop=$map constructed from prop_invmap() is '$maps{$map}[$i]', and the one from prop_invlist() is '$off_invlist[$i]'");
1976 next PROPERTY;
1977 }
1978 }
1979 }
1980 }
1981 else { # Don't know this property nor format.
1982
1983 fail("prop_invmap('$mod_prop')");
1984 diag("Unknown format '$format'");
1985 }
1986
1987 pass("prop_invmap('$mod_prop')");
1988}
1989
eaebe4db 1990done_testing();