3 $::IS_ASCII = (ord("A") == 65) ? 1 : 0;
4 $::IS_EBCDIC = (ord("A") == 193) ? 1 : 0;
7 require Config; import Config;
8 if ($Config{'extensions'} !~ /\bStorable\b/) {
9 print "1..0 # Skip: Storable was not built; Unicode::UCD uses Storable\n";
15 local $SIG{__WARN__} = sub { push @warnings, @_ };
20 use Unicode::UCD qw(charinfo charprop charprops_all);
22 my $input_record_separator = 7; # Make sure Unicode::UCD isn't affected by
23 $/ = $input_record_separator; # setting this.
27 is(charinfo(0x110000), undef, "Verify charinfo() of non-unicode is undef");
28 is(charprop(0x110000, 'age'), "Unassigned", "Verify charprop(age) of non-unicode is Unassigned");
29 is(charprop(0x110000, 'in'), "Unassigned", "Verify charprop(in), a bipartite Perl extension, works");
30 is(charprop(0x110000, 'Any'), undef, "Verify charprop of non-bipartite Perl extension returns undef");
33 $charinfo = charinfo($cp); # Null is often problematic, so test it.
35 is($charinfo->{code}, "0000",
36 "Next tests are for charinfo and charprop; first NULL");
37 is($charinfo->{name}, "<control>");
38 is(charprop($cp, "name"), "");
40 # This gets a sl-type property returning a flattened list
41 is(charprop($cp, "name_alias"), "NULL: control,NUL: abbreviation");
43 is($charinfo->{category}, "Cc");
44 is(charprop($cp, "category"), "Control");
45 is($charinfo->{combining}, "0");
46 is(charprop($cp, "ccc"), "Not_Reordered");
47 is($charinfo->{bidi}, "BN");
48 is(charprop($cp, "bc"), "Boundary_Neutral");
49 is($charinfo->{decomposition}, "");
50 is(charprop($cp, "dm"), "\0");
51 is($charinfo->{decimal}, "");
52 is($charinfo->{digit}, "");
53 is($charinfo->{numeric}, "");
54 is(charprop($cp, "nv"), "NaN");
55 is($charinfo->{mirrored}, "N");
56 is(charprop($cp, "bidim"), "No");
57 is($charinfo->{unicode10}, "NULL");
58 is(charprop($cp, "na1"), "NULL");
59 is($charinfo->{comment}, "");
60 is(charprop($cp, "isc"), "");
61 is($charinfo->{upper}, "");
62 is(charprop($cp, "uc"), "\0");
63 is($charinfo->{lower}, "");
64 is(charprop($cp, "lc"), "\0");
65 is($charinfo->{title}, "");
66 is(charprop($cp, "tc"), "\0");
67 is($charinfo->{block}, "Basic Latin");
68 is(charprop($cp, "block"), "Basic_Latin");
69 is($charinfo->{script}, "Common");
70 is(charprop($cp, "script"), "Common");
72 $cp = utf8::unicode_to_native(0x41);
73 my $A_code = sprintf("%04X", ord("A"));
74 my $a_code = sprintf("%04X", ord("a"));
75 $charinfo = charinfo(utf8::unicode_to_native($cp));
77 is($charinfo->{code}, $A_code, "LATIN CAPITAL LETTER A");
78 is($charinfo->{name}, "LATIN CAPITAL LETTER A");
79 is(charprop($cp, 'name'), "LATIN CAPITAL LETTER A");
80 is($charinfo->{category}, "Lu");
81 is(charprop($cp, 'gc'), "Uppercase_Letter");
82 is($charinfo->{combining}, "0");
83 is(charprop($cp, 'ccc'), "Not_Reordered");
84 is($charinfo->{bidi}, "L");
85 is(charprop($cp, 'bc'), "Left_To_Right");
86 is($charinfo->{decomposition}, "");
87 is(charprop($cp, 'dm'), "A");
88 is($charinfo->{decimal}, "");
89 is($charinfo->{digit}, "");
90 is($charinfo->{numeric}, "");
91 is(charprop($cp, 'nv'), "NaN");
92 is($charinfo->{mirrored}, "N");
93 is(charprop($cp, 'bidim'), "No");
94 is($charinfo->{unicode10}, "");
95 is(charprop($cp, 'na1'), "");
96 is($charinfo->{comment}, "");
97 is(charprop($cp, 'isc'), "");
98 is($charinfo->{upper}, "");
99 is(charprop($cp, 'uc'), "A");
100 is($charinfo->{lower}, $a_code);
101 is(charprop($cp, 'lc'), "a");
102 is($charinfo->{title}, "");
103 is(charprop($cp, 'tc'), "A");
104 is($charinfo->{block}, "Basic Latin");
105 is(charprop($cp, 'block'), "Basic_Latin");
106 is($charinfo->{script}, "Latin");
107 is(charprop($cp, 'script'), "Latin");
110 $charinfo = charinfo($cp);
112 is($charinfo->{code}, "0100", "LATIN CAPITAL LETTER A WITH MACRON");
113 is($charinfo->{name}, "LATIN CAPITAL LETTER A WITH MACRON");
114 is(charprop($cp, 'name'), "LATIN CAPITAL LETTER A WITH MACRON");
115 is($charinfo->{category}, "Lu");
116 is(charprop($cp, 'gc'), "Uppercase_Letter");
117 is($charinfo->{combining}, "0");
118 is(charprop($cp, 'ccc'), "Not_Reordered");
119 is($charinfo->{bidi}, "L");
120 is(charprop($cp, 'bc'), "Left_To_Right");
121 is($charinfo->{decomposition}, "$A_code 0304");
122 is(charprop($cp, 'dm'), "A\x{0304}");
123 is($charinfo->{decimal}, "");
124 is($charinfo->{digit}, "");
125 is($charinfo->{numeric}, "");
126 is(charprop($cp, 'nv'), "NaN");
127 is($charinfo->{mirrored}, "N");
128 is(charprop($cp, 'bidim'), "No");
129 is($charinfo->{unicode10}, "LATIN CAPITAL LETTER A MACRON");
130 is(charprop($cp, 'na1'), "LATIN CAPITAL LETTER A MACRON");
131 is($charinfo->{comment}, "");
132 is(charprop($cp, 'isc'), "");
133 is($charinfo->{upper}, "");
134 is(charprop($cp, 'uc'), "\x{100}");
135 is($charinfo->{lower}, "0101");
136 is(charprop($cp, 'lc'), "\x{101}");
137 is($charinfo->{title}, "");
138 is(charprop($cp, 'tc'), "\x{100}");
139 is($charinfo->{block}, "Latin Extended-A");
140 is(charprop($cp, 'block'), "Latin_Extended_A");
141 is($charinfo->{script}, "Latin");
142 is(charprop($cp, 'script'), "Latin");
144 $cp = 0x590; # 0x0590 is in the Hebrew block but unused.
145 $charinfo = charinfo($cp);
147 is($charinfo->{code}, undef, "0x0590 - unused Hebrew");
148 is($charinfo->{name}, undef);
149 is(charprop($cp, 'name'), "");
150 is($charinfo->{category}, undef);
151 is(charprop($cp, 'gc'), "Unassigned");
152 is($charinfo->{combining}, undef);
153 is(charprop($cp, 'ccc'), "Not_Reordered");
154 is($charinfo->{bidi}, undef);
155 is(charprop($cp, 'bc'), "Right_To_Left");
156 is($charinfo->{decomposition}, undef);
157 is(charprop($cp, 'dm'), "\x{590}");
158 is($charinfo->{decimal}, undef);
159 is($charinfo->{digit}, undef);
160 is($charinfo->{numeric}, undef);
161 is(charprop($cp, 'nv'), "NaN");
162 is($charinfo->{mirrored}, undef);
163 is(charprop($cp, 'bidim'), "No");
164 is($charinfo->{unicode10}, undef);
165 is(charprop($cp, 'na1'), "");
166 is($charinfo->{comment}, undef);
167 is(charprop($cp, 'isc'), "");
168 is($charinfo->{upper}, undef);
169 is(charprop($cp, 'uc'), "\x{590}");
170 is($charinfo->{lower}, undef);
171 is(charprop($cp, 'lc'), "\x{590}");
172 is($charinfo->{title}, undef);
173 is(charprop($cp, 'tc'), "\x{590}");
174 is($charinfo->{block}, undef);
175 is(charprop($cp, 'block'), "Hebrew");
176 is($charinfo->{script}, undef);
177 is(charprop($cp, 'script'), "Unknown");
179 # 0x05d0 is in the Hebrew block and used.
182 $charinfo = charinfo($cp);
184 is($charinfo->{code}, "05D0", "05D0 - used Hebrew");
185 is($charinfo->{name}, "HEBREW LETTER ALEF");
186 is(charprop($cp, 'name'), "HEBREW LETTER ALEF");
187 is($charinfo->{category}, "Lo");
188 is(charprop($cp, 'gc'), "Other_Letter");
189 is($charinfo->{combining}, "0");
190 is(charprop($cp, 'ccc'), "Not_Reordered");
191 is($charinfo->{bidi}, "R");
192 is(charprop($cp, 'bc'), "Right_To_Left");
193 is($charinfo->{decomposition}, "");
194 is(charprop($cp, 'dm'), "\x{5d0}");
195 is($charinfo->{decimal}, "");
196 is($charinfo->{digit}, "");
197 is($charinfo->{numeric}, "");
198 is(charprop($cp, 'nv'), "NaN");
199 is($charinfo->{mirrored}, "N");
200 is(charprop($cp, 'bidim'), "No");
201 is($charinfo->{unicode10}, "");
202 is(charprop($cp, 'na1'), "");
203 is($charinfo->{comment}, "");
204 is(charprop($cp, 'isc'), "");
205 is($charinfo->{upper}, "");
206 is(charprop($cp, 'uc'), "\x{5d0}");
207 is($charinfo->{lower}, "");
208 is(charprop($cp, 'lc'), "\x{5d0}");
209 is($charinfo->{title}, "");
210 is(charprop($cp, 'tc'), "\x{5d0}");
211 is($charinfo->{block}, "Hebrew");
212 is(charprop($cp, 'block'), "Hebrew");
213 is($charinfo->{script}, "Hebrew");
214 is(charprop($cp, 'script'), "Hebrew");
216 # An open syllable in Hangul.
219 $charinfo = charinfo($cp);
221 is($charinfo->{code}, "AC00", "HANGUL SYLLABLE U+AC00");
222 is($charinfo->{name}, "HANGUL SYLLABLE GA");
223 is(charprop($cp, 'name'), "HANGUL SYLLABLE GA");
224 is($charinfo->{category}, "Lo");
225 is(charprop($cp, 'gc'), "Other_Letter");
226 is($charinfo->{combining}, "0");
227 is(charprop($cp, 'ccc'), "Not_Reordered");
228 is($charinfo->{bidi}, "L");
229 is(charprop($cp, 'bc'), "Left_To_Right");
230 is($charinfo->{decomposition}, "1100 1161");
231 is(charprop($cp, 'dm'), "\x{1100}\x{1161}");
232 is($charinfo->{decimal}, "");
233 is($charinfo->{digit}, "");
234 is($charinfo->{numeric}, "");
235 is(charprop($cp, 'nv'), "NaN");
236 is($charinfo->{mirrored}, "N");
237 is(charprop($cp, 'bidim'), "No");
238 is($charinfo->{unicode10}, "");
239 is(charprop($cp, 'na1'), "");
240 is($charinfo->{comment}, "");
241 is(charprop($cp, 'isc'), "");
242 is($charinfo->{upper}, "");
243 is(charprop($cp, 'uc'), "\x{AC00}");
244 is($charinfo->{lower}, "");
245 is(charprop($cp, 'lc'), "\x{AC00}");
246 is($charinfo->{title}, "");
247 is(charprop($cp, 'tc'), "\x{AC00}");
248 is($charinfo->{block}, "Hangul Syllables");
249 is(charprop($cp, 'block'), "Hangul_Syllables");
250 is($charinfo->{script}, "Hangul");
251 is(charprop($cp, 'script'), "Hangul");
253 # A closed syllable in Hangul.
256 $charinfo = charinfo($cp);
258 is($charinfo->{code}, "AE00", "HANGUL SYLLABLE U+AE00");
259 is($charinfo->{name}, "HANGUL SYLLABLE GEUL");
260 is(charprop($cp, 'name'), "HANGUL SYLLABLE GEUL");
261 is($charinfo->{category}, "Lo");
262 is(charprop($cp, 'gc'), "Other_Letter");
263 is($charinfo->{combining}, "0");
264 is(charprop($cp, 'ccc'), "Not_Reordered");
265 is($charinfo->{bidi}, "L");
266 is(charprop($cp, 'bc'), "Left_To_Right");
267 is($charinfo->{decomposition}, "1100 1173 11AF");
268 is(charprop($cp, 'dm'), "\x{1100}\x{1173}\x{11AF}");
269 is($charinfo->{decimal}, "");
270 is($charinfo->{digit}, "");
271 is($charinfo->{numeric}, "");
272 is(charprop($cp, 'nv'), "NaN");
273 is($charinfo->{mirrored}, "N");
274 is(charprop($cp, 'bidim'), "No");
275 is($charinfo->{unicode10}, "");
276 is(charprop($cp, 'na1'), "");
277 is($charinfo->{comment}, "");
278 is(charprop($cp, 'isc'), "");
279 is($charinfo->{upper}, "");
280 is(charprop($cp, 'uc'), "\x{AE00}");
281 is($charinfo->{lower}, "");
282 is(charprop($cp, 'lc'), "\x{AE00}");
283 is($charinfo->{title}, "");
284 is(charprop($cp, 'tc'), "\x{AE00}");
285 is($charinfo->{block}, "Hangul Syllables");
286 is(charprop($cp, 'block'), "Hangul_Syllables");
287 is($charinfo->{script}, "Hangul");
288 is(charprop($cp, 'script'), "Hangul");
291 $charinfo = charinfo($cp);
293 is($charinfo->{code}, "1D400", "MATHEMATICAL BOLD CAPITAL A");
294 is($charinfo->{name}, "MATHEMATICAL BOLD CAPITAL A");
295 is(charprop($cp, 'name'), "MATHEMATICAL BOLD CAPITAL A");
296 is($charinfo->{category}, "Lu");
297 is(charprop($cp, 'gc'), "Uppercase_Letter");
298 is($charinfo->{combining}, "0");
299 is(charprop($cp, 'ccc'), "Not_Reordered");
300 is($charinfo->{bidi}, "L");
301 is(charprop($cp, 'bc'), "Left_To_Right");
302 is($charinfo->{decomposition}, "<font> $A_code");
303 is(charprop($cp, 'dm'), "A");
304 is($charinfo->{decimal}, "");
305 is($charinfo->{digit}, "");
306 is($charinfo->{numeric}, "");
307 is(charprop($cp, 'nv'), "NaN");
308 is($charinfo->{mirrored}, "N");
309 is(charprop($cp, 'bidim'), "No");
310 is($charinfo->{unicode10}, "");
311 is(charprop($cp, 'na1'), "");
312 is($charinfo->{comment}, "");
313 is(charprop($cp, 'isc'), "");
314 is($charinfo->{upper}, "");
315 is(charprop($cp, 'uc'), "\x{1D400}");
316 is($charinfo->{lower}, "");
317 is(charprop($cp, 'lc'), "\x{1D400}");
318 is($charinfo->{title}, "");
319 is(charprop($cp, 'tc'), "\x{1D400}");
320 is($charinfo->{block}, "Mathematical Alphanumeric Symbols");
321 is(charprop($cp, 'block'), "Mathematical_Alphanumeric_Symbols");
322 is($charinfo->{script}, "Common");
323 is(charprop($cp, 'script'), "Common");
325 $cp = 0x9FBA; #Bug 58428
326 $charinfo = charinfo(0x9FBA);
328 is($charinfo->{code}, "9FBA", "U+9FBA");
329 is($charinfo->{name}, "CJK UNIFIED IDEOGRAPH-9FBA");
330 is(charprop($cp, 'name'), "CJK UNIFIED IDEOGRAPH-9FBA");
331 is($charinfo->{category}, "Lo");
332 is(charprop($cp, 'gc'), "Other_Letter");
333 is($charinfo->{combining}, "0");
334 is(charprop($cp, 'ccc'), "Not_Reordered");
335 is($charinfo->{bidi}, "L");
336 is(charprop($cp, 'bc'), "Left_To_Right");
337 is($charinfo->{decomposition}, "");
338 is(charprop($cp, 'dm'), "\x{9FBA}");
339 is($charinfo->{decimal}, "");
340 is($charinfo->{digit}, "");
341 is($charinfo->{numeric}, "");
342 is(charprop($cp, 'nv'), "NaN");
343 is($charinfo->{mirrored}, "N");
344 is(charprop($cp, 'bidim'), "No");
345 is($charinfo->{unicode10}, "");
346 is(charprop($cp, 'na1'), "");
347 is($charinfo->{comment}, "");
348 is(charprop($cp, 'isc'), "");
349 is($charinfo->{upper}, "");
350 is(charprop($cp, 'uc'), "\x{9FBA}");
351 is($charinfo->{lower}, "");
352 is(charprop($cp, 'lc'), "\x{9FBA}");
353 is($charinfo->{title}, "");
354 is(charprop($cp, 'tc'), "\x{9FBA}");
355 is($charinfo->{block}, "CJK Unified Ideographs");
356 is(charprop($cp, 'block'), "CJK_Unified_Ideographs");
357 is($charinfo->{script}, "Han");
358 is(charprop($cp, 'script'), "Han");
360 use Unicode::UCD qw(charblock charscript);
362 # 0x0590 is in the Hebrew block but unused.
364 is(charblock(0x590), "Hebrew", "0x0590 - Hebrew unused charblock");
365 is(charscript(0x590), "Unknown", "0x0590 - Hebrew unused charscript");
366 is(charblock(0x1FFFF), "No_Block", "0x1FFFF - unused charblock");
368 my $fraction_3_4_code = sprintf("%04X", utf8::unicode_to_native(0xbe));
369 $cp = $fraction_3_4_code;
370 $charinfo = charinfo(hex $fraction_3_4_code);
372 is($charinfo->{code}, $fraction_3_4_code, "VULGAR FRACTION THREE QUARTERS");
373 is($charinfo->{name}, "VULGAR FRACTION THREE QUARTERS");
374 is(charprop($cp, 'name'), "VULGAR FRACTION THREE QUARTERS");
375 is($charinfo->{category}, "No");
376 is(charprop($cp, 'gc'), "Other_Number");
377 is($charinfo->{combining}, "0");
378 is(charprop($cp, 'ccc'), "Not_Reordered");
379 is($charinfo->{bidi}, "ON");
380 is(charprop($cp, 'bc'), "Other_Neutral");
381 is($charinfo->{decomposition}, "<fraction> "
382 . sprintf("%04X", ord "3")
384 . sprintf("%04X", ord "4"));
385 is(charprop($cp, 'dm'), "3\x{2044}4");
386 is($charinfo->{decimal}, "");
387 is($charinfo->{digit}, "");
388 is($charinfo->{numeric}, "3/4");
389 is(charprop($cp, 'nv'), "0.75");
390 is($charinfo->{mirrored}, "N");
391 is(charprop($cp, 'bidim'), "No");
392 is($charinfo->{unicode10}, "FRACTION THREE QUARTERS");
393 is(charprop($cp, 'na1'), "FRACTION THREE QUARTERS");
394 is($charinfo->{comment}, "");
395 is(charprop($cp, 'isc'), "");
396 is($charinfo->{upper}, "");
397 is(charprop($cp, 'uc'), "\x{be}");
398 is($charinfo->{lower}, "");
399 is(charprop($cp, 'lc'), "\x{be}");
400 is($charinfo->{title}, "");
401 is(charprop($cp, 'tc'), "\x{be}");
402 is($charinfo->{block}, "Latin-1 Supplement");
403 is(charprop($cp, 'block'), "Latin_1_Supplement");
404 is($charinfo->{script}, "Common");
405 is(charprop($cp, 'script'), "Common");
407 # This is to test a case where both simple and full lowercases exist and
410 $charinfo = charinfo($cp);
411 my $I_code = sprintf("%04X", ord("I"));
412 my $i_code = sprintf("%04X", ord("i"));
414 is($charinfo->{code}, "0130", "LATIN CAPITAL LETTER I WITH DOT ABOVE");
415 is($charinfo->{name}, "LATIN CAPITAL LETTER I WITH DOT ABOVE");
416 is(charprop($cp, 'name'), "LATIN CAPITAL LETTER I WITH DOT ABOVE");
417 is($charinfo->{category}, "Lu");
418 is(charprop($cp, 'gc'), "Uppercase_Letter");
419 is($charinfo->{combining}, "0");
420 is(charprop($cp, 'ccc'), "Not_Reordered");
421 is($charinfo->{bidi}, "L");
422 is(charprop($cp, 'bc'), "Left_To_Right");
423 is($charinfo->{decomposition}, "$I_code 0307");
424 is(charprop($cp, 'dm'), "I\x{0307}");
425 is($charinfo->{decimal}, "");
426 is($charinfo->{digit}, "");
427 is($charinfo->{numeric}, "");
428 is(charprop($cp, 'nv'), "NaN");
429 is($charinfo->{mirrored}, "N");
430 is(charprop($cp, 'bidim'), "No");
431 is($charinfo->{unicode10}, "LATIN CAPITAL LETTER I DOT");
432 is(charprop($cp, 'na1'), "LATIN CAPITAL LETTER I DOT");
433 is($charinfo->{comment}, "");
434 is(charprop($cp, 'isc'), "");
435 is($charinfo->{upper}, "");
436 is(charprop($cp, 'uc'), "\x{130}");
437 is($charinfo->{lower}, $i_code);
438 is(charprop($cp, 'lc'), "i\x{307}");
439 is($charinfo->{title}, "");
440 is(charprop($cp, 'tc'), "\x{130}");
441 is($charinfo->{block}, "Latin Extended-A");
442 is(charprop($cp, 'block'), "Latin_Extended_A");
443 is($charinfo->{script}, "Latin");
444 is(charprop($cp, 'script'), "Latin");
446 # This is to test a case where both simple and full uppercases exist and
449 $charinfo = charinfo($cp);
451 is($charinfo->{code}, "1F80", "GREEK SMALL LETTER ALPHA WITH PSILI AND YPOGEGRAMMENI");
452 is($charinfo->{name}, "GREEK SMALL LETTER ALPHA WITH PSILI AND YPOGEGRAMMENI");
453 is(charprop($cp, "name"), "GREEK SMALL LETTER ALPHA WITH PSILI AND YPOGEGRAMMENI");
454 is($charinfo->{category}, "Ll");
455 is(charprop($cp, "gc"), "Lowercase_Letter");
456 is($charinfo->{combining}, "0");
457 is(charprop($cp, "ccc"), "Not_Reordered");
458 is($charinfo->{bidi}, "L");
459 is(charprop($cp, "bc"), "Left_To_Right");
460 is($charinfo->{decomposition}, "1F00 0345");
461 is(charprop($cp, "dm"), "\x{1F00}\x{0345}");
462 is($charinfo->{decimal}, "");
463 is($charinfo->{digit}, "");
464 is($charinfo->{numeric}, "");
465 is(charprop($cp, "nv"), "NaN");
466 is($charinfo->{mirrored}, "N");
467 is(charprop($cp, "bidim"), "No");
468 is($charinfo->{unicode10}, "");
469 is(charprop($cp, "na1"), "");
470 is($charinfo->{comment}, "");
471 is(charprop($cp, "isc"), "");
472 is($charinfo->{upper}, "1F88");
473 is(charprop($cp, "uc"), "\x{1F08}\x{0399}");
474 is(charprop($cp, "suc"), "\x{1F88}");
475 is($charinfo->{lower}, "");
476 is(charprop($cp, "lc"), "\x{1F80}");
477 is($charinfo->{title}, "1F88");
478 is(charprop($cp, "tc"), "\x{1F88}");
479 is($charinfo->{block}, "Greek Extended");
480 is(charprop($cp, "block"), "Greek_Extended");
481 is($charinfo->{script}, "Greek");
482 is(charprop($cp, "script"), "Greek");
484 is(charprop(ord("A"), "foo"), undef,
485 "Verify charprop of unknown property returns <undef>");
487 # These were created from inspection of the code to exercise the branches
488 is(charprop(ord("("), "bpb"), ")",
489 "Verify charprop figures out that s-type properties can be char");
490 is(charprop(ord("9"), "nv"), 9,
491 "Verify charprop can adjust an ar-type property");
492 is(charprop(utf8::unicode_to_native(0xAD), "NFKC_Casefold"), "",
493 "Verify charprop can handle an \"\" in ae-type property");
495 my $mark_props_ref = charprops_all(0x300);
496 is($mark_props_ref->{'Bidi_Class'}, "Nonspacing_Mark",
497 "Next tests are charprops_all of 0x300");
498 is($mark_props_ref->{'Bidi_Mirrored'}, "No");
499 is($mark_props_ref->{'Canonical_Combining_Class'}, "Above");
500 is($mark_props_ref->{'Case_Folding'}, "\x{300}");
501 is($mark_props_ref->{'Decomposition_Mapping'}, "\x{300}");
502 is($mark_props_ref->{'Decomposition_Type'}, "None");
503 is($mark_props_ref->{'General_Category'}, "Nonspacing_Mark");
504 is($mark_props_ref->{'ISO_Comment'}, "");
505 is($mark_props_ref->{'Lowercase_Mapping'}, "\x{300}");
506 is($mark_props_ref->{'Name'}, "COMBINING GRAVE ACCENT");
507 is($mark_props_ref->{'Numeric_Type'}, "None");
508 is($mark_props_ref->{'Numeric_Value'}, "NaN");
509 is($mark_props_ref->{'Simple_Case_Folding'}, "\x{300}");
510 is($mark_props_ref->{'Simple_Lowercase_Mapping'}, "\x{300}");
511 is($mark_props_ref->{'Simple_Titlecase_Mapping'}, "\x{300}");
512 is($mark_props_ref->{'Simple_Uppercase_Mapping'}, "\x{300}");
513 is($mark_props_ref->{'Titlecase_Mapping'}, "\x{300}");
514 is($mark_props_ref->{'Unicode_1_Name'}, "NON-SPACING GRAVE");
515 is($mark_props_ref->{'Uppercase_Mapping'}, "\x{300}");
517 use Unicode::UCD qw(charblocks charscripts);
519 my $charblocks = charblocks();
521 ok(exists $charblocks->{Thai}, 'Thai charblock exists');
522 is($charblocks->{Thai}->[0]->[0], hex('0e00'));
523 ok(!exists $charblocks->{PigLatin}, 'PigLatin charblock does not exist');
525 my $charscripts = charscripts();
527 ok(exists $charscripts->{Armenian}, 'Armenian charscript exists');
528 is($charscripts->{Armenian}->[0]->[0], hex('0531'));
529 ok(!exists $charscripts->{PigLatin}, 'PigLatin charscript does not exist');
533 $charscript = charscript("12ab");
534 is($charscript, 'Ethiopic', 'Ethiopic charscript');
536 $charscript = charscript("0x12ab");
537 is($charscript, 'Ethiopic');
539 $charscript = charscript("U+12ab");
540 is($charscript, 'Ethiopic');
544 $ranges = charscript('Ogham');
545 is($ranges->[0]->[0], hex('1680'), 'Ogham charscript');
546 is($ranges->[0]->[1], hex('169C'));
548 use Unicode::UCD qw(charinrange);
550 $ranges = charscript('Cherokee');
551 ok(!charinrange($ranges, "139f"), 'Cherokee charscript');
552 ok( charinrange($ranges, "13a0"));
553 ok( charinrange($ranges, "13f4"));
554 ok(!charinrange($ranges, "13f5"));
556 use Unicode::UCD qw(general_categories);
558 my $gc = general_categories();
560 ok(exists $gc->{L}, 'has L');
561 is($gc->{L}, 'Letter', 'L is Letter');
562 is($gc->{Lu}, 'UppercaseLetter', 'Lu is UppercaseLetter');
564 use Unicode::UCD qw(bidi_types);
566 my $bt = bidi_types();
568 ok(exists $bt->{L}, 'has L');
569 is($bt->{L}, 'Left-to-Right', 'L is Left-to-Right');
570 is($bt->{AL}, 'Right-to-Left Arabic', 'AL is Right-to-Left Arabic');
572 # If this fails, then maybe one should look at the Unicode changes to see
573 # what else might need to be updated.
574 is(Unicode::UCD::UnicodeVersion, '7.0.0', 'UnicodeVersion');
576 use Unicode::UCD qw(compexcl);
578 ok(!compexcl(0x0100), 'compexcl');
579 ok(!compexcl(0xD801), 'compexcl of surrogate');
580 ok(!compexcl(0x110000), 'compexcl of non-Unicode code point');
581 ok( compexcl(0x0958));
583 use Unicode::UCD qw(casefold);
587 $casefold = casefold(utf8::unicode_to_native(0x41));
589 is($casefold->{code}, $A_code, 'casefold native(0x41) code');
590 is($casefold->{status}, 'C', 'casefold native(0x41) status');
591 is($casefold->{mapping}, $a_code, 'casefold native(0x41) mapping');
592 is($casefold->{full}, $a_code, 'casefold native(0x41) full');
593 is($casefold->{simple}, $a_code, 'casefold native(0x41) simple');
594 is($casefold->{turkic}, "", 'casefold native(0x41) turkic');
596 $casefold = casefold(utf8::unicode_to_native(0xdf));
597 my $sharp_s_code = sprintf("%04X", utf8::unicode_to_native(0xdf));
598 my $S_code = sprintf("%04X", ord "S");
599 my $s_code = sprintf("%04X", ord "s");
601 is($casefold->{code}, $sharp_s_code, 'casefold native(0xDF) code');
602 is($casefold->{status}, 'F', 'casefold native(0xDF) status');
603 is($casefold->{mapping}, "$s_code $s_code", 'casefold native(0xDF) mapping');
604 is($casefold->{full}, "$s_code $s_code", 'casefold native(0xDF) full');
605 is($casefold->{simple}, "", 'casefold native(0xDF) simple');
606 is($casefold->{turkic}, "", 'casefold native(0xDF) turkic');
608 # Do different tests depending on if version < 3.2, or not.
609 my $v_unicode_version = pack "C*", split /\./, Unicode::UCD::UnicodeVersion();
610 if ($v_unicode_version lt v3.2.0) {
611 $casefold = casefold(0x130);
613 is($casefold->{code}, '0130', 'casefold 0x130 code');
614 is($casefold->{status}, 'I' , 'casefold 0x130 status');
615 is($casefold->{mapping}, $i_code, 'casefold 0x130 mapping');
616 is($casefold->{full}, $i_code, 'casefold 0x130 full');
617 is($casefold->{simple}, $i_code, 'casefold 0x130 simple');
618 is($casefold->{turkic}, $i_code, 'casefold 0x130 turkic');
620 $casefold = casefold(0x131);
622 is($casefold->{code}, '0131', 'casefold 0x131 code');
623 is($casefold->{status}, 'I' , 'casefold 0x131 status');
624 is($casefold->{mapping}, $i_code, 'casefold 0x131 mapping');
625 is($casefold->{full}, $i_code, 'casefold 0x131 full');
626 is($casefold->{simple}, $i_code, 'casefold 0x131 simple');
627 is($casefold->{turkic}, $i_code, 'casefold 0x131 turkic');
629 $casefold = casefold(utf8::unicode_to_native(0x49));
631 is($casefold->{code}, $I_code, 'casefold native(0x49) code');
632 is($casefold->{status}, 'C' , 'casefold native(0x49) status');
633 is($casefold->{mapping}, $i_code, 'casefold native(0x49) mapping');
634 is($casefold->{full}, $i_code, 'casefold native(0x49) full');
635 is($casefold->{simple}, $i_code, 'casefold native(0x49) simple');
636 is($casefold->{turkic}, "0131", 'casefold native(0x49) turkic');
638 $casefold = casefold(0x130);
640 is($casefold->{code}, '0130', 'casefold 0x130 code');
641 is($casefold->{status}, 'F' , 'casefold 0x130 status');
642 is($casefold->{mapping}, "$i_code 0307", 'casefold 0x130 mapping');
643 is($casefold->{full}, "$i_code 0307", 'casefold 0x130 full');
644 is($casefold->{simple}, "", 'casefold 0x130 simple');
645 is($casefold->{turkic}, $i_code, 'casefold 0x130 turkic');
648 $casefold = casefold(0x1F88);
650 is($casefold->{code}, '1F88', 'casefold 0x1F88 code');
651 is($casefold->{status}, 'S' , 'casefold 0x1F88 status');
652 is($casefold->{mapping}, '1F80', 'casefold 0x1F88 mapping');
653 is($casefold->{full}, '1F00 03B9', 'casefold 0x1F88 full');
654 is($casefold->{simple}, '1F80', 'casefold 0x1F88 simple');
655 is($casefold->{turkic}, "", 'casefold 0x1F88 turkic');
657 ok(!casefold(utf8::unicode_to_native(0x20)));
659 use Unicode::UCD qw(casespec);
663 ok(!casespec(utf8::unicode_to_native(0x41)));
665 $casespec = casespec(utf8::unicode_to_native(0xdf));
667 ok($casespec->{code} eq $sharp_s_code &&
668 $casespec->{lower} eq $sharp_s_code &&
669 $casespec->{title} eq "$S_code $s_code" &&
670 $casespec->{upper} eq "$S_code $S_code" &&
671 !defined $casespec->{condition}, 'casespec native(0xDF)');
673 $casespec = casespec(0x307);
675 ok($casespec->{az}->{code} eq '0307' &&
676 !defined $casespec->{az}->{lower} &&
677 $casespec->{az}->{title} eq '0307' &&
678 $casespec->{az}->{upper} eq '0307' &&
679 $casespec->{az}->{condition} eq 'az After_I',
682 # perl #7305 UnicodeCD::compexcl is weird
684 for (1) {my $a=compexcl $_}
685 ok(1, 'compexcl read-only $_: perl #7305');
686 map {compexcl $_} %{{1=>2}};
687 ok(1, 'compexcl read-only hash: perl #7305');
689 is(Unicode::UCD::_getcode('123'), 123, "_getcode(123)");
690 is(Unicode::UCD::_getcode('0123'), 0x123, "_getcode(0123)");
691 is(Unicode::UCD::_getcode('0x123'), 0x123, "_getcode(0x123)");
692 is(Unicode::UCD::_getcode('0X123'), 0x123, "_getcode(0X123)");
693 is(Unicode::UCD::_getcode('U+123'), 0x123, "_getcode(U+123)");
694 is(Unicode::UCD::_getcode('u+123'), 0x123, "_getcode(u+123)");
695 is(Unicode::UCD::_getcode('U+1234'), 0x1234, "_getcode(U+1234)");
696 is(Unicode::UCD::_getcode('U+12345'), 0x12345, "_getcode(U+12345)");
697 is(Unicode::UCD::_getcode('123x'), undef, "_getcode(123x)");
698 is(Unicode::UCD::_getcode('x123'), undef, "_getcode(x123)");
699 is(Unicode::UCD::_getcode('0x123x'), undef, "_getcode(x123)");
700 is(Unicode::UCD::_getcode('U+123x'), undef, "_getcode(x123)");
703 my $r1 = charscript('Latin');
704 if (ok(defined $r1, "Found Latin script")) {
706 is($n1, 33, "number of ranges in Latin script (Unicode 7.0.0)") if $::IS_ASCII;
707 shift @$r1 while @$r1;
708 my $r2 = charscript('Latin');
709 is(@$r2, $n1, "modifying results should not mess up internal caches");
714 is(charinfo(0xdeadbeef), undef, "[perl #23273] warnings in Unicode::UCD");
717 use Unicode::UCD qw(namedseq);
719 is(namedseq("KATAKANA LETTER AINU P"), "\x{31F7}\x{309A}", "namedseq");
720 is(namedseq("KATAKANA LETTER AINU Q"), undef);
721 is(namedseq(), undef);
722 is(namedseq(qw(foo bar)), undef);
723 my @ns = namedseq("KATAKANA LETTER AINU P");
728 is($ns{"KATAKANA LETTER AINU P"}, "\x{31F7}\x{309A}");
732 use Unicode::UCD qw(num);
733 use charnames ":full";
735 is(num("0"), 0, 'Verify num("0") == 0');
736 is(num("98765"), 98765, 'Verify num("98765") == 98765');
737 ok(! defined num("98765\N{FULLWIDTH DIGIT FOUR}"), 'Verify num("98765\N{FULLWIDTH DIGIT FOUR}") isnt defined');
738 is(num("\N{NEW TAI LUE DIGIT TWO}"), 2, 'Verify num("\N{NEW TAI LUE DIGIT TWO}") == 2');
739 is(num("\N{NEW TAI LUE DIGIT ONE}"), 1, 'Verify num("\N{NEW TAI LUE DIGIT ONE}") == 1');
740 is(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');
741 ok(! 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');
742 is(num("\N{CHAM DIGIT ZERO}\N{CHAM DIGIT THREE}"), 3, 'Verify num("\N{CHAM DIGIT ZERO}\N{CHAM DIGIT THREE}") == 3');
743 ok(! defined num("\N{CHAM DIGIT ZERO}\N{JAVANESE DIGIT NINE}"), 'Verify num("\N{CHAM DIGIT ZERO}\N{JAVANESE DIGIT NINE}") isnt defined');
744 is(num("\N{SUPERSCRIPT TWO}"), 2, 'Verify num("\N{SUPERSCRIPT TWO} == 2');
745 is(num("\N{ETHIOPIC NUMBER TEN THOUSAND}"), 10000, 'Verify num("\N{ETHIOPIC NUMBER TEN THOUSAND}") == 10000');
746 is(num("\N{NORTH INDIC FRACTION ONE HALF}"), .5, 'Verify num("\N{NORTH INDIC FRACTION ONE HALF}") == .5');
747 is(num("\N{U+12448}"), 9, 'Verify num("\N{U+12448}") == 9');
748 is(num("\N{U+5146}"), 1000000000000, 'Verify num("\N{U+5146}") == 1000000000000');
750 # Create a user-defined property
756 use Unicode::UCD qw(prop_aliases);
758 is(prop_aliases(undef), undef, "prop_aliases(undef) returns <undef>");
759 is(prop_aliases("unknown property"), undef,
760 "prop_aliases(<unknown property>) returns <undef>");
761 is(prop_aliases("InKana"), undef,
762 "prop_aliases(<user-defined property>) returns <undef>");
763 is(prop_aliases("Perl_Decomposition_Mapping"), undef, "prop_aliases('Perl_Decomposition_Mapping') returns <undef> since internal-Perl-only");
764 is(prop_aliases("Perl_Charnames"), undef,
765 "prop_aliases('Perl_Charnames') returns <undef> since internal-Perl-only");
766 is(prop_aliases("isgc"), undef,
767 "prop_aliases('isgc') returns <undef> since is not covered Perl extension");
768 is(prop_aliases("Is_Is_Any"), undef,
769 "prop_aliases('Is_Is_Any') returns <undef> since two is's");
770 is(prop_aliases("ccc=vr"), undef,
771 "prop_aliases('ccc=vr') doesn't generate a warning");
773 require 'utf8_heavy.pl';
774 require "unicore/Heavy.pl";
776 # Keys are lists of properties. Values are defined if have been tested.
779 # To test for loose matching, add in the characters that are ignored there.
780 my $extra_chars = "-_ ";
782 # The one internal property we accept
783 $props{'Perl_Decimal_Digit'} = 1;
784 my @list = prop_aliases("perldecimaldigit");
786 [ "Perl_Decimal_Digit",
788 ], "prop_aliases('perldecimaldigit') returns Perl_Decimal_Digit as both short and full names");
790 # Get the official Unicode property name synonyms and test them.
793 skip "PropertyAliases.txt is not in this Unicode version", 1 if $v_unicode_version lt v3.2.0;
794 open my $props, "<", "../lib/unicore/PropertyAliases.txt"
795 or die "Can't open Unicode PropertyAliases.txt";
798 s/\s*#.*//; # Remove comments
799 next if /^\s* $/x; # Ignore empty and comment lines
802 local $/ = $input_record_separator;
803 my $count = 0; # 0th field in line is short name; 1th is long name
807 foreach my $alias (split /\s*;\s*/) { # Fields are separated by
809 # Add in the characters that are supposed to be ignored, to test loose
810 # matching, which the tested function does on all inputs.
811 my $mod_name = "$extra_chars$alias";
813 my $loose = &utf8::_loose_name(lc $alias);
815 # Indicate we have tested this.
818 my @all_names = prop_aliases($mod_name);
819 if (grep { $_ eq $loose } @Unicode::UCD::suppressed_properties) {
820 is(@all_names, 0, "prop_aliases('$mod_name') returns undef since $alias is not installed");
823 elsif (! @all_names) {
824 fail("prop_aliases('$mod_name')");
825 diag("'$alias' is unknown to prop_aliases()");
829 if ($count == 0) { # Is short name
831 @names_via_short = prop_aliases($mod_name);
833 # If the 0th test fails, no sense in continuing with the others
834 last unless is($names_via_short[0], $alias,
835 "prop_aliases: '$alias' is the short name for '$mod_name'");
836 $short_name = $alias;
838 elsif ($count == 1) { # Is full name
840 # Some properties have the same short and full name; no sense
841 # repeating the test if the same.
842 if ($alias ne $short_name) {
843 my @names_via_full = prop_aliases($mod_name);
844 is_deeply(\@names_via_full, \@names_via_short, "prop_aliases() returns the same list for both '$short_name' and '$mod_name'");
847 # Tests scalar context
848 is(prop_aliases($short_name), $alias,
849 "prop_aliases: '$alias' is the long name for '$short_name'");
851 else { # Is another alias
852 is_deeply(\@all_names, \@names_via_short, "prop_aliases() returns the same list for both '$short_name' and '$mod_name'");
853 ok((grep { $_ =~ /^$alias$/i } @all_names),
854 "prop_aliases: '$alias' is listed as an alias for '$mod_name'");
860 } # End of SKIP block
862 # Now test anything we can find that wasn't covered by the tests of the
863 # official properties. We have no way of knowing if mktables omitted a Perl
864 # extension or not, but we do the best we can from its generated lists
866 foreach my $alias (sort keys %utf8::loose_to_file_of) {
867 next if $alias =~ /=/;
868 my $lc_name = lc $alias;
869 my $loose = &utf8::_loose_name($lc_name);
870 next if exists $props{$loose}; # Skip if already tested
872 my $mod_name = "$extra_chars$alias"; # Tests loose matching
873 my @aliases = prop_aliases($mod_name);
874 my $found_it = grep { &utf8::_loose_name(lc $_) eq $lc_name } @aliases;
876 pass("prop_aliases: '$lc_name' is listed as an alias for '$mod_name'");
878 elsif ($lc_name =~ /l[_&]$/) {
880 # These two names are special in that they don't appear in the
881 # returned list because they are discouraged from use. Verify
882 # that they return the same list as a non-discouraged version.
883 my @LC = prop_aliases('Is_LC');
884 is_deeply(\@aliases, \@LC, "prop_aliases: '$lc_name' returns the same list as 'Is_LC'");
887 my $stripped = $lc_name =~ s/^is//;
889 # Could be that the input includes a prefix 'is', which is rarely
890 # returned as an alias, so having successfully stripped it off above,
893 $found_it = grep { &utf8::_loose_name(lc $_) eq $lc_name } @aliases;
896 # If that didn't work, it could be that it's a block, which is always
897 # returned with a leading 'In_' to avoid ambiguity. Try comparing
898 # with that stripped off.
900 $found_it = grep { &utf8::_loose_name(s/^In_(.*)/\L$1/r) eq $lc_name }
902 # Could check that is a real block, but tests for invmap will
903 # likely pickup any errors, since this will be tested there.
904 $lc_name = "in$lc_name" if $found_it; # Change for message below
906 my $message = "prop_aliases: '$lc_name' is listed as an alias for '$mod_name'";
907 ($found_it) ? pass($message) : fail($message);
911 # Some of the Perl extensions should always be built; make sure they have the
912 # correct full name, etc.
913 for my $prop (qw(Alnum Blank Cntrl Digit Graph Print Word XDigit)) {
914 my @expected = ( $prop, "XPosix$prop" );
915 my @got = prop_aliases($prop);
917 is_deeply(\@got, \@expected, "Got expected aliases for $prop");
921 foreach my $alias (keys %utf8::stricter_to_file_of) {
922 if ($alias =~ /=/) { # Only test one case where there is an equals
923 next if $done_equals;
926 my $lc_name = lc $alias;
927 my @list = prop_aliases($alias);
928 if ($alias =~ /^_/) {
929 is(@list, 0, "prop_aliases: '$lc_name' returns an empty list since it is internal_only");
931 elsif ($alias =~ /=/) {
932 is(@list, 0, "prop_aliases: '$lc_name' returns an empty list since is illegal property name");
935 ok((grep { lc $_ eq $lc_name } @list),
936 "prop_aliases: '$lc_name' is listed as an alias for '$alias'");
940 use Unicode::UCD qw(prop_values prop_value_aliases);
942 is(prop_value_aliases("unknown property", "unknown value"), undef,
943 "prop_value_aliases(<unknown property>, <unknown value>) returns <undef>");
944 is(prop_value_aliases(undef, undef), undef,
945 "prop_value_aliases(undef, undef) returns <undef>");
946 is((prop_value_aliases("na", "A")), "A", "test that prop_value_aliases returns its input for properties that don't have synonyms");
947 is(prop_value_aliases("isgc", "C"), undef, "prop_value_aliases('isgc', 'C') returns <undef> since is not covered Perl extension");
948 is(prop_value_aliases("gc", "isC"), undef, "prop_value_aliases('gc', 'isC') returns <undef> since is not covered Perl extension");
949 is(prop_value_aliases("Any", "None"), undef, "prop_value_aliases('Any', 'None') returns <undef> since is Perl extension and 'None' is not valid");
950 is(prop_value_aliases("lc", "A"), "A", "prop_value_aliases('lc', 'A') returns its input, as docs say it does");
952 # We have no way of knowing if mktables omitted a Perl extension that it
953 # shouldn't have, but we can check if it omitted an official Unicode property
954 # name synonym. And for those, we can check if the short and full names are
957 my %pva_tested; # List of things already tested.
960 skip "PropValueAliases.txt is not in this Unicode version", 1 if $v_unicode_version lt v3.2.0;
961 open my $propvalues, "<", "../lib/unicore/PropValueAliases.txt"
962 or die "Can't open Unicode PropValueAliases.txt";
965 # Each examined line in the file is for a single value for a property. We
966 # accumulate all the values for each property using these two variables.
968 my @this_prop_values;
970 while (<$propvalues>) {
971 s/\s*#.*//; # Remove comments
972 next if /^\s* $/x; # Ignore empty and comment lines
974 local $/ = $input_record_separator;
976 # Fix typo in official input file
977 s/CCC133/CCC132/g if $v_unicode_version eq v6.1.0;
979 my @fields = split /\s*;\s*/; # Fields are separated by semi-colons
980 my $prop = shift @fields; # 0th field is the property,
982 # When changing properties, we examine the accumulated values for the old
983 # one to see if our function that returns them matches.
984 if ($prev_prop ne $prop) {
985 if ($prev_prop ne "") { # Skip for the first time through
986 my @ucd_function_values = prop_values($prev_prop);
987 @ucd_function_values = () unless @ucd_function_values;
989 # This perl extension doesn't appear in the official file
990 push @this_prop_values, "Non_Canon" if $prev_prop eq 'dt';
992 my @file_values = undef;
993 @file_values = sort { lc($a =~ s/_//gr) cmp lc($b =~ s/_//gr) }
994 @this_prop_values if @this_prop_values;
995 is_deeply(\@ucd_function_values, \@file_values,
996 "prop_values('$prev_prop') returns correct list of values");
999 undef @this_prop_values;
1002 my $count = 0; # 0th field in line (after shifting off the property) is
1003 # short name; 1th is long name
1005 my @names_via_short; # Saves the values between iterations
1007 # The property on the lhs of the = is always loosely matched. Add in
1008 # characters that are ignored under loose matching to test that
1009 my $mod_prop = "$extra_chars$prop";
1011 if ($fields[0] eq 'n/a') { # See comments in input file, essentially
1012 # means full name and short name are identical
1013 $fields[0] = $fields[1];
1015 elsif ($fields[0] ne $fields[1]
1016 && &utf8::_loose_name(lc $fields[0])
1017 eq &utf8::_loose_name(lc $fields[1])
1018 && $fields[1] !~ /[[:upper:]]/)
1020 # Also, there is a bug in the file in which "n/a" is omitted, and
1021 # the two fields are identical except for case, and the full name
1022 # is all lower case. Copy the "short" name unto the full one to
1023 # give it some upper case.
1025 $fields[1] = $fields[0];
1028 # The ccc property in the file is special; has an extra numeric field
1029 # (0th), which should go at the end, since we use the next two fields as
1030 # the short and full names, respectively. See comments in input file.
1031 splice (@fields, 0, 0, splice(@fields, 1, 2)) if $prop eq 'ccc';
1033 my $loose_prop = &utf8::_loose_name(lc $prop);
1034 my $suppressed = grep { $_ eq $loose_prop }
1035 @Unicode::UCD::suppressed_properties;
1036 push @this_prop_values, $fields[0] unless $suppressed;
1037 foreach my $value (@fields) {
1039 is(prop_value_aliases($prop, $value), undef, "prop_value_aliases('$prop', '$value') returns undef for suppressed property $prop");
1042 elsif (grep { $_ eq ("$loose_prop=" . &utf8::_loose_name(lc $value)) } @Unicode::UCD::suppressed_properties) {
1043 is(prop_value_aliases($prop, $value), undef, "prop_value_aliases('$prop', '$value') returns undef for suppressed property $prop=$value");
1047 # Add in test for loose matching.
1048 my $mod_value = "$extra_chars$value";
1050 # If the value is a number, optionally negative, including a floating
1051 # point or rational numer, it should be only strictly matched, so the
1052 # loose matching should fail.
1053 if ($value =~ / ^ -? \d+ (?: [\/.] \d+ )? $ /x) {
1054 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");
1056 # And reset so below tests just the strict matching.
1057 $mod_value = $value;
1062 @names_via_short = prop_value_aliases($mod_prop, $mod_value);
1064 # If the 0th test fails, no sense in continuing with the others
1065 last unless is($names_via_short[0], $value, "prop_value_aliases: In '$prop', '$value' is the short name for '$mod_value'");
1066 $short_name = $value;
1068 elsif ($count == 1) {
1070 # Some properties have the same short and full name; no sense
1071 # repeating the test if the same.
1072 if ($value ne $short_name) {
1073 my @names_via_full =
1074 prop_value_aliases($mod_prop, $mod_value);
1075 is_deeply(\@names_via_full, \@names_via_short, "In '$prop', prop_value_aliases() returns the same list for both '$short_name' and '$mod_value'");
1078 # Tests scalar context
1079 is(prop_value_aliases($prop, $short_name), $value, "'$value' is the long name for prop_value_aliases('$prop', '$short_name')");
1082 my @all_names = prop_value_aliases($mod_prop, $mod_value);
1083 is_deeply(\@all_names, \@names_via_short, "In '$prop', prop_value_aliases() returns the same list for both '$short_name' and '$mod_value'");
1084 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')");
1087 $pva_tested{&utf8::_loose_name(lc $prop) . "=" . &utf8::_loose_name(lc $value)} = 1;
1091 } # End of SKIP block
1093 # And test as best we can, the non-official pva's that mktables generates.
1094 foreach my $hash (\%utf8::loose_to_file_of, \%utf8::stricter_to_file_of) {
1095 foreach my $test (sort keys %$hash) {
1096 next if exists $pva_tested{$test}; # Skip if already tested
1098 my ($prop, $value) = split "=", $test;
1099 next unless defined $value; # prop_value_aliases() requires an input
1102 if ($hash == \%utf8::loose_to_file_of) {
1104 # Add extra characters to test loose-match rhs value
1105 $mod_value = "$extra_chars$value";
1107 else { # Here value is strictly matched.
1109 # Extra elements are added by mktables to this hash so that
1110 # something like "age=6.0" has a synonym of "age=6". It's not
1111 # clear to me (khw) if we should be encouraging those synonyms, so
1112 # don't test for them.
1113 next if $value !~ /\D/ && exists $hash->{"$prop=$value.0"};
1115 # Verify that loose matching fails when only strict is called for.
1116 next unless is(prop_value_aliases($prop, "$extra_chars$value"), undef,
1117 "prop_value_aliases('$prop', '$extra_chars$value') returns undef since '$value' should be strictly matched"),
1119 # Strict matching does allow for underscores between digits. Test
1121 $mod_value = $value;
1122 while ($mod_value =~ s/(\d)(\d)/$1_$2/g) {}
1125 # The lhs property is always loosely matched, so add in extra
1126 # characters to test that.
1127 my $mod_prop = "$extra_chars$prop";
1129 if ($prop eq 'gc' && $value =~ /l[_&]$/) {
1130 # These two names are special in that they don't appear in the
1131 # returned list because they are discouraged from use. Verify
1132 # that they return the same list as a non-discouraged version.
1133 my @LC = prop_value_aliases('gc', 'lc');
1134 my @l_ = prop_value_aliases($mod_prop, $mod_value);
1135 is_deeply(\@l_, \@LC, "prop_value_aliases('$mod_prop', '$mod_value) returns the same list as prop_value_aliases('gc', 'lc')");
1138 ok((grep { &utf8::_loose_name(lc $_) eq &utf8::_loose_name(lc $value) }
1139 prop_value_aliases($mod_prop, $mod_value)),
1140 "'$value' is listed as an alias for prop_value_aliases('$mod_prop', '$mod_value')");
1147 no warnings 'once'; # We use some values once from 'required' modules.
1149 use Unicode::UCD qw(prop_invlist prop_invmap MAX_CP);
1151 # There were some problems with caching interfering with prop_invlist() vs
1152 # prop_invmap() on binary properties, and also between the 3 properties where
1153 # Perl used the same 'To' name as another property (see utf8_heavy.pl).
1154 # So, before testing all of prop_invlist(),
1155 # 1) call prop_invmap() to try both orders of these name issues. This uses
1156 # up two of the 3 properties; the third will be left so that invlist()
1157 # on it gets called before invmap()
1158 # 2) call prop_invmap() on a generic binary property, ahead of invlist().
1159 # This should test that the caching works in both directions.
1161 # These properties are not stable between Unicode versions, but the first few
1162 # elements are; just look at the first element to see if are getting the
1163 # distinction right. The general inversion map testing below will test the
1167 my ($invlist_ref, $invmap_ref, $format, $missing);
1168 if ($::IS_ASCII) { # On EBCDIC, other things will come first, and can vary
1169 # according to code page
1171 ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
1172 is($format, 'al', "prop_invmap() format of '$prop' is 'al'");
1173 is($missing, '0', "prop_invmap() missing of '$prop' is '0'");
1174 is($invlist_ref->[1], 0x61, "prop_invmap('$prop') list[1] is 0x61");
1175 is($invmap_ref->[1], 0x41, "prop_invmap('$prop') map[1] is 0x41");
1178 ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
1179 is($format, 's', "prop_invmap() format of '$prop' is 's");
1180 is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
1181 is($invlist_ref->[1], 0x41, "prop_invmap('$prop') list[1] is 0x41");
1182 is($invmap_ref->[1], 'Y', "prop_invmap('$prop') map[1] is 'Y'");
1185 ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
1186 is($format, 's', "prop_invmap() format of '$prop' is 's'");
1187 is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
1188 is($invlist_ref->[1], 0x61, "prop_invmap('$prop') list[1] is 0x61");
1189 is($invmap_ref->[1], 'Y', "prop_invmap('$prop') map[1] is 'Y'");
1192 ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
1193 is($format, 'al', "prop_invmap() format of '$prop' is 'al'");
1194 is($missing, '0', "prop_invmap() missing of '$prop' is '0'");
1195 is($invlist_ref->[1], 0x41, "prop_invmap('$prop') list[1] is 0x41");
1196 is($invmap_ref->[1], 0x61, "prop_invmap('$prop') map[1] is 0x61");
1199 # This property is stable and small, so can test all of it
1200 $prop = "ASCII_Hex_Digit";
1201 ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($prop);
1202 is($format, 's', "prop_invmap() format of '$prop' is 's'");
1203 is($missing, 'N', "prop_invmap() missing of '$prop' is 'N'");
1205 is_deeply($invlist_ref, [ 0x0000, 0x0030, 0x003A,
1207 0x0061, 0x0067, 0x110000
1209 "prop_invmap('$prop') code point list is correct");
1211 elsif ($::IS_EBCDIC) {
1212 is_deeply($invlist_ref, [
1213 utf8::unicode_to_native(0x0000),
1214 utf8::unicode_to_native(0x0061), utf8::unicode_to_native(0x0066) + 1,
1215 utf8::unicode_to_native(0x0041), utf8::unicode_to_native(0x0046) + 1,
1216 utf8::unicode_to_native(0x0030), utf8::unicode_to_native(0x0039) + 1,
1217 utf8::unicode_to_native(0x110000)
1219 "prop_invmap('$prop') code point list is correct");
1221 is_deeply($invmap_ref, [ 'N', 'Y', 'N', 'Y', 'N', 'Y', 'N', 'N' ] ,
1222 "prop_invmap('$prop') map list is correct");
1224 is(prop_invlist("Unknown property"), undef, "prop_invlist(<Unknown property>) returns undef");
1225 is(prop_invlist(undef), undef, "prop_invlist(undef) returns undef");
1226 is(prop_invlist("Any"), 2, "prop_invlist('Any') returns the number of elements in scalar context");
1227 my @invlist = prop_invlist("Is_Any");
1228 is_deeply(\@invlist, [ 0, 0x110000 ], "prop_invlist works on 'Is_' prefixes");
1229 is(prop_invlist("Is_Is_Any"), undef, "prop_invlist('Is_Is_Any') returns <undef> since two is's");
1231 use Storable qw(dclone);
1233 is(prop_invlist("InKana"), undef, "prop_invlist(<user-defined property returns undef>)");
1235 # The way both the tests for invlist and invmap work is that they take the
1236 # lists returned by the functions and construct from them what the original
1237 # file should look like, which are then compared with the file. If they are
1238 # identical, the test passes. What this tests isn't that the results are
1239 # correct, but that invlist and invmap haven't introduced errors beyond what
1240 # are there in the files. As a small hedge against that, test some
1241 # prop_invlist() tables fully with the known correct result. We choose
1242 # ASCII_Hex_Digit again, as it is stable.
1244 @invlist = prop_invlist("AHex");
1245 is_deeply(\@invlist, [ 0x0030, 0x003A, 0x0041,
1246 0x0047, 0x0061, 0x0067 ],
1247 "prop_invlist('AHex') is exactly the expected set of points");
1248 @invlist = prop_invlist("AHex=f");
1249 is_deeply(\@invlist, [ 0x0000, 0x0030, 0x003A, 0x0041,
1250 0x0047, 0x0061, 0x0067 ],
1251 "prop_invlist('AHex=f') is exactly the expected set of points");
1253 elsif ($::IS_EBCDIC) { # Relies on the ranges 0-9, a-f, and A-F each being
1255 @invlist = prop_invlist("AHex");
1256 is_deeply(\@invlist, [
1257 utf8::unicode_to_native(0x0061), utf8::unicode_to_native(0x0066) + 1,
1258 utf8::unicode_to_native(0x0041), utf8::unicode_to_native(0x0046) + 1,
1259 utf8::unicode_to_native(0x0030), utf8::unicode_to_native(0x0039) + 1,
1261 "prop_invlist('AHex') is exactly the expected set of points");
1262 @invlist = prop_invlist("AHex=f");
1263 is_deeply(\@invlist, [
1264 utf8::unicode_to_native(0x0000),
1265 utf8::unicode_to_native(0x0061),
1266 utf8::unicode_to_native(0x0066) + 1,
1267 utf8::unicode_to_native(0x0041),
1268 utf8::unicode_to_native(0x0046) + 1,
1269 utf8::unicode_to_native(0x0030),
1270 utf8::unicode_to_native(0x0039) + 1,
1272 "prop_invlist('AHex=f') is exactly the expected set of points");
1275 sub fail_with_diff ($$$$) {
1276 # For use below to output better messages
1277 my ($prop, $official, $constructed, $tested_function_name) = @_;
1279 is($constructed, $official, "$tested_function_name('$prop')");
1280 diag("Comment out lines " . (__LINE__ - 1) . " through " . (__LINE__ + 1) . " in '$0' on Un*x-like systems to see just the differences. Uses the 'diff' first in your \$PATH");
1283 fail("$tested_function_name('$prop')");
1286 my $off = File::Temp->new();
1289 print $off $official, "\n";
1290 close $off || die "Can't close official";
1293 my $gend = File::Temp->new();
1294 print $gend $constructed, "\n";
1295 close $gend || die "Can't close gend";
1297 my $diff = File::Temp->new();
1298 system("diff $off $gend > $diff");
1300 open my $fh, "<", $diff || die "Can't open $diff";
1302 diag("In the diff output below '<' marks lines from the filesystem tables;\n'>' are from $tested_function_name()");
1308 # Look at everything we think that mktables tells us exists, both loose and
1310 foreach my $set_of_tables (\%utf8::stricter_to_file_of, \%utf8::loose_to_file_of)
1312 foreach my $table (sort keys %$set_of_tables) {
1315 my ($prop_only, $value) = split "=", $table;
1316 if (defined $value) {
1318 # If this is to be loose matched, add in characters to test that.
1319 if ($set_of_tables == \%utf8::loose_to_file_of) {
1320 $value = "$extra_chars$value";
1322 else { # Strict match
1324 # Verify that loose matching fails when only strict is called
1326 next unless is(prop_invlist("$prop_only=$extra_chars$value"), undef, "prop_invlist('$prop_only=$extra_chars$value') returns undef since should be strictly matched");
1328 # Strict matching does allow for underscores between digits.
1330 while ($value =~ s/(\d)(\d)/$1_$2/g) {}
1333 # The property portion in compound form specifications always
1335 $mod_table = "$extra_chars$prop_only = $value";
1337 else { # Single-form.
1339 # Like above, use loose if required, and insert underscores
1340 # between digits if strict.
1341 if ($set_of_tables == \%utf8::loose_to_file_of) {
1342 $mod_table = "$extra_chars$table";
1345 $mod_table = $table;
1346 while ($mod_table =~ s/(\d)(\d)/$1_$2/g) {}
1350 my @tested = prop_invlist($mod_table);
1351 if ($table =~ /^_/) {
1352 is(@tested, 0, "prop_invlist('$mod_table') returns an empty list since is internal-only");
1356 # If we have already tested a property that uses the same file, this
1357 # list should be identical to the one that was tested, and can bypass
1359 my $file = $set_of_tables->{$table};
1360 if (exists $tested_invlist{$file}) {
1361 is_deeply(\@tested, $tested_invlist{$file}, "prop_invlist('$mod_table') gave same results as its name synonym");
1364 $tested_invlist{$file} = dclone \@tested;
1366 # A '!' in the file name means that it is to be inverted.
1367 my $invert = $file =~ s/!//;
1370 # If the file's directory is '#', it is a special case where the
1371 # contents are in-lined with semi-colons meaning new-lines, instead of
1372 # it being an actual file to read. The file is an index in to the
1373 # array of the definitions
1374 if ($file =~ s!^#/!!) {
1375 $official = $utf8::inline_definitions[$file];
1378 $official = do "unicore/lib/$file.pl";
1381 # Get rid of any trailing space and comments in the file.
1382 $official =~ s/\s*(#.*)?$//mg;
1385 $/ = $input_record_separator;
1387 # If we are to test against an inverted file, it is easier to invert
1388 # our array than the file.
1390 if (@tested && $tested[0] == 0) {
1397 # Now construct a string from the list that should match the file.
1398 # The file is inversion list format code points, like this:
1404 # The V indicates it's an inversion list, and is followed immediately
1405 # by the number of elements (lines) that follow giving its contents.
1406 # The list has even numbered elements (0th, 2nd, ...) start ranges
1407 # that are in the list, and odd ones that aren't in the list.
1408 # Therefore the odd numbered ones are one beyond the end of the
1409 # previous range, but otherwise don't get reflected in the file.
1410 my $tested = join "\n", ("V" . scalar @tested), @tested;
1413 $/ = $input_record_separator;
1414 if ($tested ne $official) {
1415 fail_with_diff($mod_table, $official, $tested, "prop_invlist");
1419 pass("prop_invlist('$mod_table')");
1423 # Now test prop_invmap().
1425 @list = prop_invmap("Unknown property");
1426 is (@list, 0, "prop_invmap(<Unknown property>) returns an empty list");
1427 @list = prop_invmap(undef);
1428 is (@list, 0, "prop_invmap(undef) returns an empty list");
1429 ok (! eval "prop_invmap('gc')" && $@ ne "",
1430 "prop_invmap('gc') dies in scalar context");
1431 @list = prop_invmap("_X_Begin");
1432 is (@list, 0, "prop_invmap(<internal property>) returns an empty list");
1433 @list = prop_invmap("InKana");
1434 is(@list, 0, "prop_invmap(<user-defined property returns undef>)");
1435 @list = prop_invmap("Perl_Decomposition_Mapping"), undef,
1436 is(@list, 0, "prop_invmap('Perl_Decomposition_Mapping') returns <undef> since internal-Perl-only");
1437 @list = prop_invmap("Perl_Charnames"), undef,
1438 is(@list, 0, "prop_invmap('Perl_Charnames') returns <undef> since internal-Perl-only");
1439 @list = prop_invmap("Is_Is_Any");
1440 is(@list, 0, "prop_invmap('Is_Is_Any') returns <undef> since two is's");
1442 # The files for these properties are not used by Perl, but are retained for
1443 # backwards compatibility with applications that read them directly, with
1444 # comments in them that their use is deprecated. Until such time as we remove
1445 # them completely, we test that they exist, are correct, and that their
1446 # formats haven't changed. This hash contains the info needed to test them as
1447 # if they were regular properties. 'replaced_by' gives the equivalent
1448 # property now used by Perl.
1449 my %legacy_props = (
1450 Legacy_Case_Folding => { replaced_by => 'cf',
1452 swash_name => 'ToFold'
1454 Legacy_Lowercase_Mapping => { replaced_by => 'lc',
1456 swash_name => 'ToLower'
1458 Legacy_Titlecase_Mapping => { replaced_by => 'tc',
1460 swash_name => 'ToTitle'
1462 Legacy_Uppercase_Mapping => { replaced_by => 'uc',
1464 swash_name => 'ToUpper'
1466 Legacy_Perl_Decimal_Digit => { replaced_by => 'Perl_Decimal_Digit',
1468 swash_name => 'ToDigit'
1472 foreach my $legacy_prop (keys %legacy_props) {
1473 @list = prop_invmap($legacy_prop);
1474 is(@list, 0, "'$legacy_prop' is unknown to prop_invmap");
1477 # The files for these properties shouldn't have their formats changed in case
1478 # applications use them (though such use is deprecated).
1479 my @legacy_file_format = (keys %legacy_props,
1480 qw( Bidi_Mirroring_Glyph
1485 # The set of properties to test on has already been compiled into %props by
1486 # the prop_aliases() tests.
1490 # Like prop_invlist(), prop_invmap() is tested by comparing the results
1491 # returned by the function with the tables that mktables generates. Some of
1492 # these tables are directly stored as files on disk, in either the unicore or
1493 # unicore/To directories, and most should be listed in the mktables generated
1494 # hash %utf8::loose_property_to_file_of, with a few additional ones that this
1495 # handles specially. For these, the files are read in directly, massaged, and
1496 # compared with what invmap() returns. The SPECIALS hash in some of these
1497 # files overrides values in the main part of the file.
1499 # The other properties are tested indirectly by generating all the possible
1500 # inversion lists for the property, and seeing if those match the inversion
1501 # lists returned by prop_invlist(), which has already been tested.
1504 foreach my $prop (sort(keys %props), sort keys %legacy_props) {
1506 my $loose_prop = &utf8::_loose_name(lc $prop);
1507 my $suppressed = grep { $_ eq $loose_prop }
1508 @Unicode::UCD::suppressed_properties;
1510 my $actual_lookup_prop;
1511 my $display_prop; # The property name that is displayed, as opposed
1512 # to the one that is actually used.
1514 # Find the short and full names that this property goes by
1515 my ($name, $full_name) = prop_aliases($prop);
1518 # Here, Perl doesn't know about this property. It could be a
1519 # suppressed one, or a legacy one.
1520 if (grep { $prop eq $_ } keys %legacy_props) {
1522 # For legacy properties, we look up the modern equivalent
1523 # property instead; later massaging the results to look like the
1524 # known format of the legacy property. We add info about the
1525 # legacy property to the data structures for the rest of the
1526 # properties; this is to avoid more special cases for the legacies
1528 $full_name = $name = $prop;
1529 $actual_lookup_prop = $legacy_props{$prop}->{'replaced_by'};
1530 my $base_file = $legacy_props{$prop}->{'file'};
1532 # This legacy property is otherwise unknown to Perl; so shouldn't
1533 # have any information about it already.
1534 ok(! exists $utf8::loose_property_to_file_of{$loose_prop},
1535 "There isn't a hash entry for file lookup of $prop");
1536 $utf8::loose_property_to_file_of{$loose_prop} = $base_file;
1538 ok(! exists $utf8::file_to_swash_name{$loose_prop},
1539 "There isn't a hash entry for swash lookup of $prop");
1540 $utf8::file_to_swash_name{$base_file}
1541 = $legacy_props{$prop}->{'swash_name'};
1542 $display_prop = $prop;
1546 if (! $suppressed) {
1547 fail("prop_invmap('$prop')");
1548 diag("is unknown to prop_aliases(), and we need it in order to test prop_invmap");
1554 # Normalize the short name, as it is stored in the hashes under the
1555 # normalized version.
1556 $name = &utf8::_loose_name(lc $name);
1558 # Add in the characters that are supposed to be ignored to test loose
1559 # matching, which the tested function applies to all properties
1560 $display_prop = "$extra_chars$prop" unless $display_prop;
1561 $actual_lookup_prop = $display_prop unless $actual_lookup_prop;
1563 my ($invlist_ref, $invmap_ref, $format, $missing) = prop_invmap($actual_lookup_prop);
1564 my $return_ref = [ $invlist_ref, $invmap_ref, $format, $missing ];
1567 # The legacy property files all are expanded out so that each range is 1
1568 # element long. That isn't true of the modern equivalent we use to check
1569 # those files for correctness against. So take the output of the proxy
1570 # and expand it to match the legacy file.
1574 for my $i (0 .. @$invlist_ref - 1 - 1) {
1575 if (ref $invmap_ref->[$i] || $invmap_ref->[$i] eq $missing) {
1577 # No adjustments should be done for the default mapping and
1578 # the multi-char ones.
1579 push @expanded_list, $invlist_ref->[$i];
1580 push @expanded_map, $invmap_ref->[$i];
1584 # Expand the range into separate elements for each item.
1586 for my $j ($invlist_ref->[$i] .. $invlist_ref->[$i+1] -1) {
1587 push @expanded_list, $j;
1588 push @expanded_map, $invmap_ref->[$i] + $offset;
1590 # The 'ae' format is for Legacy_Perl_Decimal_Digit; the
1591 # other 4 are kept with leading zeros in the file, so
1593 $expanded_map[-1] = sprintf("%04X", $expanded_map[-1])
1600 # Final element is taken as is. The map should always be to the
1601 # default value, so don't do a sprintf like we did above.
1602 push @expanded_list, $invlist_ref->[-1];
1603 push @expanded_map, $invmap_ref->[-1];
1605 $invlist_ref = \@expanded_list;
1606 $invmap_ref = \@expanded_map;
1609 # If have already tested this property under a different name, merely
1610 # compare the return from now with the saved one from before.
1611 if (exists $tested_invmaps{$name}) {
1612 is_deeply($return_ref, $tested_invmaps{$name}, "prop_invmap('$display_prop') gave same results as its synonym, '$name'");
1615 $tested_invmaps{$name} = dclone $return_ref;
1617 # If prop_invmap() returned nothing, is ok iff is a property whose file is
1620 if (defined $format) {
1621 fail("prop_invmap('$display_prop')");
1622 diag("did not return undef for suppressed property $prop");
1626 elsif (!defined $format) {
1627 fail("prop_invmap('$display_prop')");
1628 diag("'$prop' is unknown to prop_invmap()");
1632 # The two parallel arrays must have the same number of elements.
1633 if (@$invlist_ref != @$invmap_ref) {
1634 fail("prop_invmap('$display_prop')");
1636 . scalar @$invlist_ref
1637 . " while invmap has "
1638 . scalar @$invmap_ref
1643 # The last element must be for the above-Unicode code points, and must be
1644 # for the default value.
1645 if ($invlist_ref->[-1] != 0x110000) {
1646 fail("prop_invmap('$display_prop')");
1647 diag("The last inversion list element is not 0x110000");
1651 my $upper_limit_subtract;
1653 # prop_invmap() adds an extra element not present in the disk files for
1654 # the above-Unicode code points. For almost all properties, that will be
1655 # to $missing. In that case we don't look further at it when comparing
1656 # with the disk files.
1657 if ($invmap_ref->[-1] eq $missing) {
1658 $upper_limit_subtract = 1;
1660 elsif ($invmap_ref->[-1] eq 'Y' && ! grep { $_ !~ /[YN]/ } @$invmap_ref) {
1662 # But that's not true for a few binary properties like 'Unassigned'
1663 # that are Perl extensions (in this case for Gc=Unassigned) which
1664 # match above-Unicode code points (hence the 'Y' in the test above).
1665 # For properties where it isn't $missing, we're going to want to look
1666 # at the whole thing when comparing with the disk file.
1667 $upper_limit_subtract = 0;
1669 # In those properties like 'Unassigned, the final element should be
1670 # just a repetition of the next-to-last element, and won't be in the
1671 # disk file, so remove it for the comparison. Otherwise, we will
1672 # compare the whole of the array with the whole of the disk file.
1673 if ($invlist_ref->[-2] <= 0x10FFFF && $invmap_ref->[-2] eq 'Y') {
1679 fail("prop_invmap('$display_prop')");
1680 diag("The last inversion list element is '$invmap_ref->[-1]', and should be '$missing'");
1684 if ($name eq 'bmg') { # This one has an atypical $missing
1685 if ($missing ne "") {
1686 fail("prop_invmap('$display_prop')");
1687 diag("The missings should be \"\"; got '$missing'");
1691 elsif ($format =~ /^ a (?!r) /x) {
1692 if ($full_name eq 'Perl_Decimal_Digit') {
1693 if ($missing ne "") {
1694 fail("prop_invmap('$display_prop')");
1695 diag("The missings should be \"\"; got '$missing'");
1699 elsif ($missing ne "0" && ! grep { $prop eq $_ } keys %legacy_props) {
1700 fail("prop_invmap('$display_prop')");
1701 diag("The missings should be '0'; got '$missing'");
1705 elsif ($missing =~ /[<>]/) {
1706 fail("prop_invmap('$display_prop')");
1707 diag("The missings should NOT be something with <...>'");
1710 # I don't want to hard code in what all the missings should be, so
1711 # those don't get fully tested.
1714 # Certain properties don't have their own files, but must be constructed
1716 my $proxy_prop = $name;
1717 if ($full_name eq 'Present_In') {
1718 $proxy_prop = "age"; # The maps for these two props are identical
1720 elsif ($full_name eq 'Simple_Case_Folding'
1721 || $full_name =~ /Simple_ (.) .*? case_Mapping /x)
1723 if ($full_name eq 'Simple_Case_Folding') {
1727 # We captured the U, L, or T, leading to uc, lc, or tc.
1728 $proxy_prop = lc $1 . "c";
1730 if ($format ne "a") {
1731 fail("prop_invmap('$display_prop')");
1732 diag("The format should be 'a'; got '$format'");
1737 if ($format !~ / ^ (?: a [der]? | ale? | n | sl? ) $ /x) {
1738 fail("prop_invmap('$display_prop')");
1739 diag("Unknown format '$format'");
1746 # Handle the properties that have full disk files for them (except the
1747 # Name property which is structurally enough different that it is handled
1748 # separately below.)
1752 ($base_file = $utf8::loose_property_to_file_of{$proxy_prop})
1753 || exists $utf8::loose_to_file_of{$proxy_prop}
1756 # In the above, blk is done unconditionally, as we need to test that
1757 # the old-style block names are returned, even if mktables has
1758 # generated a file for the new-style; the test for dm comes afterward,
1759 # so that if a file has been generated for it explicitly, we use that
1760 # file (which is valid, unlike blk) instead of the combo
1761 # Decomposition.pl files.
1764 if ($name eq 'blk') {
1766 # The blk property is special. The original file with old block
1767 # names is retained, and the default (on ASCII platforms) is to
1768 # not write out a new-name file. What we do is get the old names
1769 # into a data structure, and from that create what the new file
1770 # would look like. $base_file is needed to be defined, just to
1771 # avoid a message below.
1772 $base_file = "This is a dummy name";
1773 my $blocks_ref = charblocks();
1776 # On EBCDIC, the first two blocks can each contain multiple
1777 # ranges. We create a new version with each of these
1778 # flattened, so have one level. ($index is used as a dummy
1782 foreach my $block (values %$blocks_ref) {
1783 foreach my $range (@$block) {
1784 $new_blocks{$index++}[0] = $range;
1787 $blocks_ref = \%new_blocks;
1790 for my $range (sort { $a->[0][0] <=> $b->[0][0] }
1791 values %$blocks_ref)
1793 # Translate the charblocks() data structure to what the file
1794 # would look like. (The sub range is for EBCDIC platforms
1795 # where Latin1 and ASCII are intermixed.)
1796 if ($range->[0][0] == $range->[0][1]) {
1797 $official .= sprintf("%X\t\t%s\n",
1802 $official .= sprintf("%X\t%X\t%s\n",
1810 $base_file = "Decomposition" if $format eq 'ad';
1812 # Above leaves $base_file undefined only if it came from the hash
1813 # below. This should happen only when it is a binary property
1814 # (and are accessing via a single-form name, like 'In_Latin1'),
1815 # and so it is stored in a different directory than the To ones.
1816 # XXX Currently, the only cases where it is complemented are the
1817 # ones that have no code points. And it works out for these that
1818 # 1) complementing them, and then 2) adding or subtracting the
1819 # initial 0 and final 110000 cancel each other out. But further
1820 # work would be needed in the unlikely event that an inverted
1821 # property comes along without these characteristics
1822 if (!defined $base_file) {
1823 $base_file = $utf8::loose_to_file_of{$proxy_prop};
1824 $is_binary = ($base_file =~ s/!//) ? -1 : 1;
1825 $base_file = "lib/$base_file" unless $base_file =~ m!^#/!;
1828 # Read in the file. If the file's directory is '#', it is a
1829 # special case where the contents are in-lined with semi-colons
1830 # meaning new-lines, instead of it being an actual file to read.
1831 if ($base_file =~ s!^#/!!) {
1832 $official = $utf8::inline_definitions[$base_file];
1835 $official = do "unicore/$base_file.pl";
1838 # Get rid of any trailing space and comments in the file.
1839 $official =~ s/\s*(#.*)?$//mg;
1841 if ($format eq 'ad') {
1842 my @official = split /\n/, $official;
1844 foreach my $line (@official) {
1845 my ($start, $end, $value)
1846 = $line =~ / ^ (.+?) \t (.*?) \t (.+?)
1847 \s* ( \# .* )? $ /x;
1848 # Decomposition.pl also has the <compatible> types in it,
1849 # which should be removed.
1850 $value =~ s/<.*?> //;
1851 $official .= "$start\t\t$value\n";
1853 # If this is a multi-char range, we turn it into as many
1854 # single character ranges as necessary. This makes things
1857 for my $i (hex($start) + 1 .. hex $end) {
1858 $official .= sprintf "%X\t\t%s\n", $i, $value;
1866 $/ = $input_record_separator;
1868 # Get the format for the file, and if there are any special elements,
1869 # get a reference to them.
1870 my $swash_name = $utf8::file_to_swash_name{$base_file};
1872 my $file_format; # The 'format' given inside the file
1874 $specials_ref = $utf8::SwashInfo{$swash_name}{'specials_name'};
1875 if ($specials_ref) {
1877 # Convert from the name to the actual reference.
1879 $specials_ref = \%{$specials_ref};
1882 $file_format = $utf8::SwashInfo{$swash_name}{'format'};
1885 # Leading zeros used to be used with the values in the files that give,
1886 # ranges, but these have been mostly stripped off, except for some
1887 # files whose formats should not change in any way.
1888 my $file_range_format = (grep { $full_name eq $_ } @legacy_file_format)
1891 # Currently this property still has leading zeroes in the mapped-to
1892 # values, but otherwise, those values follow the same rules as the
1894 my $file_map_format = ($full_name eq 'Decomposition_Mapping')
1896 : $file_range_format;
1898 # Certain of the proxy properties have to be adjusted to match the
1901 =~ /^(Legacy_)?(Case_Folding|(Lower|Title|Upper)case_Mapping)/)
1904 # Here we have either
1905 # 1) Case_Folding; or
1906 # 2) a proxy that is a full mapping, which means that what the
1907 # real property is is the equivalent simple mapping.
1908 # In both cases, the file will have a standard list containing
1909 # simple mappings (to a single code point), and a specials hash
1910 # which contains all the mappings that are to multiple code
1911 # points. First, extract a list containing all the file's simple
1914 for (split "\n", $official) {
1915 my ($start, $end, $value) = / ^ (.+?) \t (.*?) \t (.+?)
1916 \s* ( \# .* )? $ /x;
1917 $end = $start if $end eq "";
1918 push @list, [ hex $start, hex $end, hex $value ];
1921 # For these mappings, the file contains all the simple mappings,
1922 # including the ones that are overridden by the specials. These
1923 # need to be removed as the list is for just the full ones.
1925 # Go through any special mappings one by one. The keys are the
1926 # UTF-8 representation of code points.
1928 foreach my $utf8_cp (sort keys %$specials_ref) {
1933 # Find the spot in the @list of simple mappings that this
1934 # special applies to; uses a linear search.
1935 while ($i < @list -1 ) {
1936 last if $cp <= $list[$i][1];
1940 # Here $i is such that it points to the first range which ends
1941 # at or above cp, and hence is the only range that could
1942 # possibly contain it.
1944 # If not in this range, no range contains it: nothing to
1946 next if $cp < $list[$i][0];
1948 # Otherwise, remove the existing entry. If it is the first
1949 # element of the range...
1950 if ($cp == $list[$i][0]) {
1952 # ... and there are other elements in the range, just
1953 # shorten the range to exclude this code point.
1954 if ($list[$i][1] > $list[$i][0]) {
1958 # ... but if it is the only element in the range, remove
1961 splice @list, $i, 1;
1964 else { # Is somewhere in the middle of the range
1965 # Split the range into two, excluding this one in the
1967 splice @list, $i, 1,
1968 [ $list[$i][0], $cp - 1, $list[$i][2] ],
1969 [ $cp + 1, $list[$i][1], $list[$i][2] ];
1973 # Here, have gone through all the specials, modifying @list as
1974 # needed. Turn it back into what the file should look like.
1976 for my $element (@list) {
1977 $official .= "\n" if $official;
1978 if ($element->[1] == $element->[0]) {
1980 .= sprintf "$file_range_format\t\t$file_map_format",
1981 $element->[0], $element->[2];
1984 $official .= sprintf "$file_range_format\t$file_range_format\t$file_map_format",
1992 =~ / ^ Simple_(Case_Folding|(Lower|Title|Upper)case_Mapping) $ /x)
1995 # These properties have everything in the regular array, and the
1996 # specials are superfluous.
1997 undef $specials_ref;
1999 elsif ($format !~ /^a/ && defined $file_format && $file_format eq 'x') {
2001 # For these properties the file is output using hex notation for the
2002 # map. Convert from hex to decimal.
2003 my @lines = split "\n", $official;
2004 foreach my $line (@lines) {
2005 my ($lower, $upper, $map) = split "\t", $line;
2006 $line = "$lower\t$upper\t" . hex $map;
2008 $official = join "\n", @lines;
2011 # Here, in $official, we have what the file looks like, or should like
2012 # if we've had to fix it up. Now take the invmap() output and reverse
2013 # engineer from that what the file should look like. Each iteration
2014 # appends the next line to the running string.
2015 my $tested_map = "";
2017 # For use with files for binary properties only, which are stored in
2018 # inversion list format. This counts the number of data lines in the
2020 my $binary_count = 0;
2022 # Create a copy of the file's specials hash. (It has been undef'd if
2023 # we know it isn't relevant to this property, so if it exists, it's an
2024 # error or is relevant). As we go along, we delete from that copy.
2025 # If a delete fails, or something is left over after we are done,
2027 my %specials = %$specials_ref if $specials_ref;
2029 # The extra -$upper_limit_subtract is because the final element may
2030 # have been tested above to be for anything above Unicode, in which
2031 # case the file may not go that high.
2032 for (my $i = 0; $i < @$invlist_ref - $upper_limit_subtract; $i++) {
2034 # If the map element is a reference, have to stringify it (but
2035 # don't do so if the format doesn't allow references, so that an
2036 # improper format will generate an error.
2037 if (ref $invmap_ref->[$i]
2038 && ($format eq 'ad' || $format =~ /^ . l /x))
2040 # The stringification depends on the format.
2041 if ($format eq 'sl') {
2043 # At the time of this writing, there are two types of 'sl'
2044 # format One, in Name_Alias, has multiple separate
2045 # entries for each code point; the other, in
2046 # Script_Extension, is space separated. Assume the latter
2047 # for non-Name_Alias.
2048 if ($full_name ne 'Name_Alias') {
2049 $invmap_ref->[$i] = join " ", @{$invmap_ref->[$i]};
2052 # For Name_Alias, we emulate the file. Entries with
2053 # just one value don't need any changes, but we
2054 # convert the list entries into a series of lines for
2055 # the file, starting with the first name. The
2056 # succeeding entries are on separate lines, with the
2057 # code point repeated for each one and then two tabs,
2058 # then the value. Code at the end of the loop will
2059 # set up the first line with its code point and two
2060 # tabs before the value, just as it does for every
2061 # other property; thus the special handling of the
2063 if (ref $invmap_ref->[$i]) {
2064 my $hex_cp = sprintf("%X", $invlist_ref->[$i]);
2065 my $concatenated = $invmap_ref->[$i][0];
2066 for (my $j = 1; $j < @{$invmap_ref->[$i]}; $j++) {
2067 $concatenated .= "\n$hex_cp\t\t"
2068 . $invmap_ref->[$i][$j];
2070 $invmap_ref->[$i] = $concatenated;
2074 elsif ($format =~ / ^ al e? $/x) {
2076 # For an al property, the stringified result should be in
2077 # the specials hash. The key is the utf8 bytes of the
2078 # code point, and the value is its map as a utf-8 string.
2080 my $key = chr $invlist_ref->[$i];
2082 if (! defined ($value = delete $specials{$key})) {
2083 fail("prop_invmap('$display_prop')");
2084 diag(sprintf "There was no specials element for %04X", $invlist_ref->[$i]);
2087 my $packed = pack "W*", @{$invmap_ref->[$i]};
2088 utf8::upgrade($packed);
2089 if ($value ne $packed) {
2090 fail("prop_invmap('$display_prop')");
2091 diag(sprintf "For %04X, expected the mapping to be "
2092 . "'$packed', but got '$value'", $invlist_ref->[$i]);
2096 # As this doesn't get tested when we later compare with
2097 # the actual file, it could be out of order and we
2099 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2100 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2102 fail("prop_invmap('$display_prop')");
2103 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2108 elsif ($format eq 'ad') {
2110 # The decomposition mapping file has the code points as
2111 # a string of space-separated hex constants.
2112 $invmap_ref->[$i] = join " ", map { sprintf "%04X", $_ }
2113 @{$invmap_ref->[$i]};
2116 fail("prop_invmap('$display_prop')");
2117 diag("Can't handle format '$format'");
2120 } # Otherwise, the map is to a simple scalar
2121 elsif (defined $file_format && $file_format eq 'ax') {
2122 # These maps are in hex
2123 $invmap_ref->[$i] = sprintf("%X", $invmap_ref->[$i]);
2125 elsif ($format eq 'ad' || $format eq 'ale') {
2127 # The numerics in the returned map are stored as adjusted
2128 # decimal integers. The defaults are 0, and don't appear in
2129 # $official, and are excluded later, but the elements must be
2130 # converted back to their hex values before comparing with
2131 # $official, as these files, for backwards compatibility, are
2132 # not stored as adjusted. (There currently is only one ale
2133 # property, nfkccf. If that changed this would also have to.)
2134 if ($invmap_ref->[$i] =~ / ^ -? \d+ $ /x
2135 && $invmap_ref->[$i] != 0)
2137 my $next = $invmap_ref->[$i] + 1;
2138 $invmap_ref->[$i] = sprintf($file_map_format,
2141 # If there are other elements in this range they need to
2142 # be adjusted; they must individually be re-mapped. Do
2143 # this by splicing in a new element into the list and the
2144 # map containing the remainder of the range. Next time
2145 # through we will look at that (possibly splicing again
2146 # until the whole range is processed).
2147 if ($invlist_ref->[$i+1] > $invlist_ref->[$i] + 1) {
2148 splice @$invlist_ref, $i+1, 0,
2149 $invlist_ref->[$i] + 1;
2150 splice @$invmap_ref, $i+1, 0, $next;
2153 if ($format eq 'ale' && $invmap_ref->[$i] eq "") {
2155 # ale properties have maps to the empty string that also
2156 # should be in the specials hash, with the key the utf8
2157 # bytes representing the code point, and the map just empty.
2159 my $key = chr $invlist_ref->[$i];
2161 if (! defined ($value = delete $specials{$key})) {
2162 fail("prop_invmap('$display_prop')");
2163 diag(sprintf "There was no specials element for %04X", $invlist_ref->[$i]);
2167 fail("prop_invmap('$display_prop')");
2168 diag(sprintf "For %04X, expected the mapping to be \"\", but got '$value'", $invlist_ref->[$i]);
2172 # As this doesn't get tested when we later compare with
2173 # the actual file, it could be out of order and we
2175 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2176 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2178 fail("prop_invmap('$display_prop')");
2179 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2185 elsif ($is_binary) { # These binary files don't have an explicit Y
2186 $invmap_ref->[$i] =~ s/Y//;
2189 # The file doesn't include entries that map to $missing, so don't
2190 # include it in the built-up string. But make sure that it is in
2191 # the correct order in the input.
2192 if ($invmap_ref->[$i] eq $missing) {
2193 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2194 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2196 fail("prop_invmap('$display_prop')");
2197 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2203 # The ad property has one entry which isn't in the file.
2204 # Ignore it, but make sure it is in order.
2206 && $invmap_ref->[$i] eq '<hangul syllable>'
2207 && $invlist_ref->[$i] == 0xAC00)
2209 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2210 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2212 fail("prop_invmap('$display_prop')");
2213 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2219 # Finally have figured out what the map column in the file should
2220 # be. Append the line to the running string.
2221 my $start = $invlist_ref->[$i];
2222 my $end = (defined $invlist_ref->[$i+1])
2223 ? $invlist_ref->[$i+1] - 1
2224 : $Unicode::UCD::MAX_CP;
2227 # Files for binary properties are in inversion list format,
2229 $tested_map .= "$start\n";
2232 # If the final value is infinity, no line for it exists.
2233 if ($end < $Unicode::UCD::MAX_CP) {
2234 $tested_map .= ($end + 1) . "\n";
2239 $end = ($start == $end) ? "" : sprintf($file_range_format, $end);
2240 if ($invmap_ref->[$i] ne "") {
2241 $tested_map .= sprintf "$file_range_format\t%s\t%s\n",
2242 $start, $end, $invmap_ref->[$i];
2244 elsif ($end ne "") {
2245 $tested_map .= sprintf "$file_range_format\t%s\n",
2249 $tested_map .= sprintf "$file_range_format\n", $start;
2252 } # End of looping over all elements.
2254 # Binary property files begin with a line count line.
2255 $tested_map = "V$binary_count\n$tested_map" if $binary_count;
2257 # Here are done with generating what the file should look like
2261 $/ = $input_record_separator;
2264 if ($tested_map ne $official) {
2265 fail_with_diff($display_prop, $official, $tested_map, "prop_invmap");
2269 # There shouldn't be any specials unaccounted for.
2270 if (keys %specials) {
2271 fail("prop_invmap('$display_prop')");
2272 diag("Unexpected specials: " . join ", ", keys %specials);
2276 elsif ($format eq 'n') {
2278 # Handle the Name property similar to the above. But the file is
2279 # sufficiently different that it is more convenient to make a special
2280 # case for it. It is a combination of the Name, Unicode1_Name, and
2281 # Name_Alias properties, and named sequences. We need to remove all
2282 # but the Name in order to do the comparison.
2284 if ($missing ne "") {
2285 fail("prop_invmap('$display_prop')");
2286 diag("The missings should be \"\"; got \"missing\"");
2290 $official = do "unicore/Name.pl";
2292 # Get rid of the named sequences portion of the file. These don't
2293 # have a tab before the first blank on a line.
2294 $official =~ s/ ^ [^\t]+ \ .*? \n //xmg;
2296 # And get rid of the controls. These are named in the file, but
2297 # shouldn't be in the property. This gets rid of the two ranges in
2298 # one fell swoop, and also all the Unicode1_Name values that may not
2301 $official =~ s/ 00000 \t .* 0001F .*? \n//xs;
2302 $official =~ s/ 0007F \t .* 0009F .*? \n//xs;
2304 elsif ($::IS_EBCDIC) { # Won't work for POSIX-BC
2305 $official =~ s/ 00000 \t .* 0003F .*? \n//xs;
2306 $official =~ s/ 000FF \t .* 000FF .*? \n//xs;
2309 # And remove the aliases. We read in the Name_Alias property, and go
2310 # through them one by one.
2311 my ($aliases_code_points, $aliases_maps, undef, undef)
2312 = &prop_invmap('Name_Alias');
2313 for (my $i = 0; $i < @$aliases_code_points; $i++) {
2314 my $code_point = $aliases_code_points->[$i];
2316 # Already removed these above.
2317 next if $code_point <= 0x1F
2318 || ($code_point >= 0x7F && $code_point <= 0x9F);
2320 my $hex_code_point = sprintf "%05X", $code_point;
2322 # Convert to a list if not already to make the following loop
2324 $aliases_maps->[$i] = [ $aliases_maps->[$i] ]
2325 if ! ref $aliases_maps->[$i];
2327 # Remove each alias for this code point from the file
2328 foreach my $alias (@{$aliases_maps->[$i]}) {
2330 # Remove the alias type from the entry, retaining just the name.
2333 $alias = quotemeta($alias);
2334 $official =~ s/$hex_code_point \t $alias \n //x;
2339 $/ = $input_record_separator;
2341 # Here have adjusted the file. We also have to adjust the returned
2342 # inversion map by checking and deleting all the lines in it that
2343 # won't be in the file. These are the lines that have generated
2344 # things, like <hangul syllable>.
2345 my $tested_map = ""; # Current running string
2346 my @code_point_in_names =
2347 @Unicode::UCD::code_points_ending_in_code_point;
2349 for my $i (0 .. @$invlist_ref - 1 - $upper_limit_subtract) {
2350 my $start = $invlist_ref->[$i];
2351 my $end = $invlist_ref->[$i+1] - 1;
2352 if ($invmap_ref->[$i] eq $missing) {
2353 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2354 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2356 fail("prop_invmap('$display_prop')");
2357 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2362 if ($invmap_ref->[$i] =~ / (.*) ( < .*? > )/x) {
2365 if (($i > 0 && $invlist_ref->[$i] <= $invlist_ref->[$i-1])
2366 || $invlist_ref->[$i] >= $invlist_ref->[$i+1])
2368 fail("prop_invmap('$display_prop')");
2369 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2372 if ($type eq "<hangul syllable>") {
2374 fail("prop_invmap('$display_prop')");
2375 diag("Unexpected text in $invmap_ref->[$i]");
2378 if ($start != 0xAC00) {
2379 fail("prop_invmap('$display_prop')");
2380 diag(sprintf("<hangul syllables> should begin at 0xAC00, got %04X", $start));
2383 if ($end != $start + 11172 - 1) {
2384 fail("prop_invmap('$display_prop')");
2385 diag(sprintf("<hangul syllables> should end at %04X, got %04X", $start + 11172 -1, $end));
2389 elsif ($type ne "<code point>") {
2390 fail("prop_invmap('$display_prop')");
2391 diag("Unexpected text '$type' in $invmap_ref->[$i]");
2396 # Look through the array of names that end in code points,
2397 # and look for this start and end. If not found is an
2398 # error. If found, delete it, and at the end, make sure
2399 # have deleted everything.
2400 for my $i (0 .. @code_point_in_names - 1) {
2401 my $hash = $code_point_in_names[$i];
2402 if ($hash->{'low'} == $start
2403 && $hash->{'high'} == $end
2404 && "$hash->{'name'}-" eq $name)
2406 splice @code_point_in_names, $i, 1;
2410 fail("prop_invmap('$display_prop')");
2411 diag("Unexpected code-point-in-name line '$invmap_ref->[$i]'");
2420 # Have adjusted the map, as needed. Append to running string.
2421 $end = ($start == $end) ? "" : sprintf("%05X", $end);
2422 $tested_map .= sprintf "%05X\t%s\n", $start, $invmap_ref->[$i];
2425 # Finished creating the string from the inversion map. Can compare
2426 # with what the file is.
2429 $/ = $input_record_separator;
2430 if ($tested_map ne $official) {
2431 fail_with_diff($display_prop, $official, $tested_map, "prop_invmap");
2434 if (@code_point_in_names) {
2435 fail("prop_invmap('$display_prop')");
2437 diag("Missing code-point-in-name line(s)" . Dumper \@code_point_in_names);
2441 elsif ($format eq 's') {
2443 # Here the map is not more or less directly from a file stored on
2444 # disk. We try a different tack. These should all be properties that
2445 # have just a few possible values (most of them are binary). We go
2446 # through the map list, sorting each range into buckets, one for each
2447 # map value. Thus for binary properties there will be a bucket for Y
2448 # and one for N. The buckets are inversion lists. We compare each
2449 # constructed inversion list with what we would get for it using
2450 # prop_invlist(), which has already been tested. If they all match,
2451 # the whole map must have matched.
2455 for my $i (0 .. @$invlist_ref - 1 - $upper_limit_subtract) {
2456 my $range_start = $invlist_ref->[$i];
2458 # Because we are sorting into buckets, things could be
2459 # out-of-order here, and still be in the correct order in the
2460 # bucket, and hence wouldn't show up as an error; so have to
2462 if (($i > 0 && $range_start <= $invlist_ref->[$i-1])
2463 || $range_start >= $invlist_ref->[$i+1])
2465 fail("prop_invmap('$display_prop')");
2466 diag(sprintf "Range beginning at %04X is out-of-order.", $invlist_ref->[$i]);
2470 # This new range closes out the range started in the previous
2472 push @{$maps{$previous_map}}, $range_start if defined $previous_map;
2474 # And starts a range which will be closed in the next iteration.
2475 $previous_map = $invmap_ref->[$i];
2476 push @{$maps{$previous_map}}, $range_start;
2479 # The range we just started hasn't been closed, and we didn't look at
2480 # the final element of the loop. If that range is for the default
2481 # value, it shouldn't be closed, as it is to extend to infinity. But
2482 # otherwise, it should end at the final Unicode code point, and the
2483 # list that maps to the default value should have another element that
2484 # does go to infinity for every above Unicode code point.
2486 if (@$invlist_ref > 1) {
2487 my $penultimate_map = $invmap_ref->[-2];
2488 if ($penultimate_map ne $missing) {
2490 # The -1th element contains the first non-Unicode code point.
2491 push @{$maps{$penultimate_map}}, $invlist_ref->[-1];
2492 push @{$maps{$missing}}, $invlist_ref->[-1];
2496 # Here, we have the buckets (inversion lists) all constructed. Go
2497 # through each and verify that matches what prop_invlist() returns.
2498 # We could use is_deeply() for the comparison, but would get multiple
2499 # messages for each $prop.
2500 foreach my $map (sort keys %maps) {
2501 my @off_invlist = prop_invlist("$prop = $map");
2502 my $min = (@off_invlist >= @{$maps{$map}})
2505 for my $i (0 .. $min- 1) {
2506 if ($i > @off_invlist - 1) {
2507 fail("prop_invmap('$display_prop')");
2508 diag("There is no element [$i] for $prop=$map from prop_invlist(), while [$i] in the implicit one constructed from prop_invmap() is '$maps{$map}[$i]'");
2511 elsif ($i > @{$maps{$map}} - 1) {
2512 fail("prop_invmap('$display_prop')");
2513 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[$i]'");
2516 elsif ($maps{$map}[$i] ne $off_invlist[$i]) {
2517 fail("prop_invmap('$display_prop')");
2518 diag("Element [$i] of the implicit $prop=$map constructed from prop_invmap() is '$maps{$map}[$i]', and the one from prop_invlist() is '$off_invlist[$i]'");
2524 else { # Don't know this property nor format.
2526 fail("prop_invmap('$display_prop')");
2527 diag("Unknown property '$display_prop' or format '$format'");
2531 pass("prop_invmap('$display_prop')");
2534 # A few tests of search_invlist
2535 use Unicode::UCD qw(search_invlist);
2537 my ($scripts_ranges_ref, $scripts_map_ref) = prop_invmap("Script");
2538 my $index = search_invlist($scripts_ranges_ref, 0x390);
2539 is($scripts_map_ref->[$index], "Greek", "U+0390 is Greek");
2540 my @alpha_invlist = prop_invlist("Alpha");
2541 is(search_invlist(\@alpha_invlist, ord("\t")), undef, "search_invlist returns undef for code points before first one on the list");
2543 ok($/ eq $input_record_separator, "The record separator didn't get overridden");
2545 if (! ok(@warnings == 0, "No warnings were generated")) {
2546 diag(join "\n", "The warnings are:", @warnings);