1 # Test the /a, /d, etc regex modifiers
12 # Each case is a valid element of its hash key. Choose, where available, an
13 # ASCII-range, Latin-1 non-ASCII range, and above Latin1 range code point.
15 '\w' => [ ord("A"), 0xE2, 0x16B ], # Below expects these to all be alpha
16 '\d' => [ ord("0"), 0x0662 ],
17 '\s' => [ ord("\t"), 0xA0, 0x1680 ], # Below expects these to be [:blank:]
18 '[:cntrl:]' => [ 0x00, 0x88 ],
19 '[:graph:]' => [ ord("&"), 0xF7, 0x02C7 ], # Below expects these to be
21 '[:lower:]' => [ ord("g"), 0xE3, 0x0127 ],
22 '[:punct:]' => [ ord("!"), 0xBF, 0x055C ],
23 '[:upper:]' => [ ord("G"), 0xC3, 0x0126 ],
24 '[:xdigit:]' => [ ord("4"), 0xFF15 ],
27 $testcases{'[:digit:]'} = $testcases{'\d'};
28 $testcases{'[:alnum:]'} = $testcases{'\w'};
29 $testcases{'[:alpha:]'} = $testcases{'\w'};
30 $testcases{'[:blank:]'} = $testcases{'\s'};
31 $testcases{'[:print:]'} = $testcases{'[:graph:]'};
32 $testcases{'[:space:]'} = $testcases{'\s'};
33 $testcases{'[:word:]'} = $testcases{'\w'};
35 # For each possible character set...
36 foreach my $charset ("d", "u") {
39 foreach my $upgrade ("", 'utf8::upgrade($a); ') {
41 # reverse gets the, \w, \s, \d first.
42 for my $class (reverse sort keys %testcases) {
44 # The complement of \w is \W; of [:posix:] is [:^posix:]
45 my $complement = $class;
46 if ($complement !~ s/ ( \[: ) /$1^/x) {
47 $complement = uc($class);
51 foreach my $ord (@{$testcases{$class}}) {
52 my $char = display(chr($ord));
54 # > 255 already implies upgraded. Skip the ones that don't
55 # have an explicit upgradei. This shows more clearly in the
56 # output which tests are in utf8, or not.
57 next if $ord > 255 && ! $upgrade;
59 my $reason = ""; # Explanation output with each test
60 my $match = 1; # Calculated whether test regex should
63 # Everything always matches in ASCII, or under /u
64 if ($ord < 128 || $charset eq 'u') {
65 $reason = "\"$char\" is a $class under /$charset";
67 elsif ($charset eq "a") {
69 $reason = "\"$char\" is non-ASCII, which can't be a $class under /a";
72 $reason = "\"$char\" is a $class under /$charset";
74 elsif ($charset eq 'l') {
76 # We are using the C locale, which is essentially ASCII,
77 # but under utf8, the above-latin1 chars are treated as
79 $reason = "\"$char\" is not a $class in this locale under /l";
83 $reason = "\"$char\" is a $class in utf8 under /d";
86 $reason = "\"$char\" is latin1, which requires utf8 to be a $class under /d";
89 $reason = "; $reason" if $reason;
103 foreach my $bracketed (0, 1) {
108 # Adds an extra char to the character class to make sure
109 # that the class doesn't get optimized away.
110 $lb = ($bracketed) ? '[_' : "";
111 $rb = ($bracketed) ? ']' : "";
113 else { # [:posix:] must be inside outer [ ]
114 next if $class =~ /\[/;
117 my $length = 10; # For regexec.c regrepeat() cases by
118 # matching more than one item
119 # Test both class and its complement, and with one or more
120 # than one item to match.
122 qq[my \$a = "$char"; $upgrade\$a $op qr/ (?$charset: $lb$class$rb ) /x],
123 qq[my \$a = "$char"; $upgrade\$a $neg_op qr/ (?$charset: $lb$complement$rb ) /x],
124 qq[my \$a = "$char" x $length; $upgrade\$a $op qr/ (?$charset: $lb$class$rb\{$length} ) /x],
125 qq[my \$a = "$char" x $length; $upgrade\$a $neg_op qr/ (?$charset: $lb$complement$rb\{$length} ) /x],
127 ok (eval $eval, $eval . $reason);
131 next if $class ne '\w';
133 # Test \b, \B at beginning and end of string
135 qq[my \$a = "$char"; $upgrade\$a $op qr/ (?$charset: ^ \\b . ) /x],
136 qq[my \$a = "$char"; $upgrade\$a $op qr/ (?$charset: . \\b \$) /x],
137 qq[my \$a = "$char"; $upgrade\$a $neg_op qr/(?$charset: ^ \\B . ) /x],
138 qq[my \$a = "$char"; $upgrade\$a $neg_op qr/(?$charset: . \\B \$ ) /x],
140 ok (eval $eval, $eval . $reason);
143 # Test \b, \B adjacent to a non-word char, both before it and
144 # after. We test with ASCII, Latin1 and Unicode non-word chars
145 foreach my $space_ord (@{$testcases{'\s'}}) {
147 # Useless to try to test non-utf8 when the ord itself
149 next if $space_ord > 255 && ! $upgrade;
151 my $space = display(chr $space_ord);
154 qq[my \$a = "$space$char"; $upgrade\$a $op qr/ (?$charset: . \\b . ) /x],
155 qq[my \$a = "$char$space"; $upgrade\$a $op qr/ (?$charset: . \\b . ) /x],
156 qq[my \$a = "$space$char"; $upgrade\$a $neg_op qr/ (?$charset: . \\B . ) /x],
157 qq[my \$a = "$char$space"; $upgrade\$a $neg_op qr/ (?$charset: . \\B . ) /x],
159 ok (eval $eval, $eval . $reason . "; \"$space\" is not a \\w");
163 # Test \b, \B in the middle of two nominally word chars, but
164 # one or both may be considered non-word depending on range
166 foreach my $other_ord (@{$testcases{'\w'}}) {
167 next if $other_ord > 255 && ! $upgrade;
168 my $other = display(chr $other_ord);
170 # Determine if the other char is a word char in current
172 my $other_is_word = 1;
173 my $other_reason = "\"$other\" is a $class under /$charset";
177 || ($other_ord < 256 && ($charset eq 'l' || ! $upgrade))))
180 $other_reason = "\"$other\" is not a $class under /$charset";
182 my $both_reason = $reason;
183 $both_reason .= "; $other_reason" if $other_ord != $ord;
185 # If both are the same wordness, then \b will fail; \B
187 if ($match == $other_is_word) {
197 qq[my \$a = "$other$char"; $upgrade\$a $op qr/ (?$charset: $other \\b $char ) /x],
198 qq[my \$a = "$char$other"; $upgrade\$a $op qr/ (?$charset: $char \\b $other ) /x],
199 qq[my \$a = "$other$char"; $upgrade\$a $neg_op qr/ (?$charset: $other \\B $char ) /x],
200 qq[my \$a = "$char$other"; $upgrade\$a $neg_op qr/ (?$charset: $char \\B $other ) /x],
202 ok (eval $eval, $eval . $both_reason);
205 next if $other_ord == $ord;
207 # These start with the \b or \B. They are included, based
208 # on source code analysis, to force the testing of the FBC
209 # (find_by_class) portions of regexec.c.
211 qq[my \$a = "$other$char"; $upgrade\$a $op qr/ (?$charset: \\b $char ) /x],
212 qq[my \$a = "$char$other"; $upgrade\$a $op qr/ (?$charset: \\b $other ) /x],
213 qq[my \$a = "$other$char"; $upgrade\$a $neg_op qr/ (?$charset: \\B $char ) /x],
214 qq[my \$a = "$char$other"; $upgrade\$a $neg_op qr/ (?$charset: \\B $other ) /x],
216 ok (eval $eval, $eval . $both_reason);
219 } # End of each test case in a class
220 } # End of \w, \s, ...
221 } # End of utf8 upgraded or not
224 plan(curr_test() - 1);