1#!./perl -w 2 3BEGIN { 4 chdir 't' if -d 't'; 5 @INC = '../lib'; 6 require './test.pl'; 7 eval 'require Config'; # assume defaults if this fails 8} 9 10use strict; 11use open qw(:utf8 :std); 12 13## 14## If the markers used are changed (search for "MARKER1" in regcomp.c), 15## update only these two regexs, and leave the {#} in the @death/@warning 16## arrays below. The {#} is a meta-marker -- it marks where the marker should 17## go. 18## 19## Returns empty string if that is what is expected. Otherwise, handles 20## either a scalar, turning it into a single element array; or a ref to an 21## array, adjusting each element. If called in array context, returns an 22## array, otherwise the join of all elements 23 24sub fixup_expect { 25 my $expect_ref = shift; 26 return if $expect_ref eq ""; 27 28 my @expect; 29 if (ref $expect_ref) { 30 @expect = @$expect_ref; 31 } 32 else { 33 @expect = $expect_ref; 34 } 35 36 foreach my $element (@expect) { 37 $element =~ s/{\#}/in regex; marked by <-- HERE in/; 38 $element =~ s/{\#}/ <-- HERE /; 39 $element .= " at "; 40 } 41 return wantarray ? @expect : join "", @expect; 42} 43 44## Because we don't "use utf8" in this file, we need to do some extra legwork 45## for the utf8 tests: Append 'use utf8' to the pattern, and mark the strings 46## to check against as UTF-8 47## 48## This also creates a second variant of the tests to check if the 49## latin1 error messages are working correctly. 50my $l1 = "\x{ef}"; 51my $utf8 = "\x{30cd}"; 52utf8::encode($utf8); 53 54sub mark_as_utf8 { 55 my @ret; 56 while ( my ($pat, $msg) = splice(@_, 0, 2) ) { 57 my $l1_pat = $pat =~ s/$utf8/$l1/gr; 58 my $l1_msg; 59 $pat = "use utf8; $pat"; 60 61 if (ref $msg) { 62 $l1_msg = [ map { s/$utf8/$l1/gr } @$msg ]; 63 @$msg = map { my $c = $_; utf8::decode($c); $c } @$msg; 64 } 65 else { 66 $l1_msg = $msg =~ s/$utf8/$l1/gr; 67 utf8::decode($msg); 68 } 69 push @ret, $pat => $msg; 70 push @ret, $l1_pat => $l1_msg unless $l1_pat =~ /#no latin1/; 71 } 72 return @ret; 73} 74 75my $inf_m1 = ($Config::Config{reg_infty} || 32767) - 1; 76my $inf_p1 = $inf_m1 + 2; 77 78## 79## Key-value pairs of code/error of code that should have fatal errors. 80## 81my @death = 82( 83 '/[[=foo=]]/' => 'POSIX syntax [= =] is reserved for future extensions {#} m/[[=foo=]{#}]/', 84 85 '/(?<= .*)/' => 'Variable length lookbehind not implemented in regex m/(?<= .*)/', 86 87 '/(?<= x{1000})/' => 'Lookbehind longer than 255 not implemented in regex m/(?<= x{1000})/', 88 89 '/(?@)/' => 'Sequence (?@...) not implemented {#} m/(?@{#})/', 90 91 '/(?{ 1/' => 'Missing right curly or square bracket', 92 93 '/(?(1x))/' => 'Switch condition not recognized {#} m/(?(1x{#}))/', 94 '/(?(1x(?#)))/'=> 'Switch condition not recognized {#} m/(?(1x{#}(?#)))/', 95 96 '/(?(1)x|y|z)/' => 'Switch (?(condition)... contains too many branches {#} m/(?(1)x|y|{#}z)/', 97 98 '/(?(x)y|x)/' => 'Unknown switch condition (?(...)) {#} m/(?(x{#})y|x)/', 99 100 '/(?/' => 'Sequence (? incomplete {#} m/(?{#}/', 101 102 '/(?;x/' => 'Sequence (?;...) not recognized {#} m/(?;{#}x/', 103 '/(?<;x/' => 'Group name must start with a non-digit word character {#} m/(?<;{#}x/', 104 '/(?\ix/' => 'Sequence (?\...) not recognized {#} m/(?\{#}ix/', 105 '/(?\mx/' => 'Sequence (?\...) not recognized {#} m/(?\{#}mx/', 106 '/(?\:x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}:x/', 107 '/(?\=x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}=x/', 108 '/(?\!x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}!x/', 109 '/(?\<=x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}<=x/', 110 '/(?\<!x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}<!x/', 111 '/(?\>x/' => 'Sequence (?\...) not recognized {#} m/(?\{#}>x/', 112 '/(?^-i:foo)/' => 'Sequence (?^-...) not recognized {#} m/(?^-{#}i:foo)/', 113 '/(?^-i)foo/' => 'Sequence (?^-...) not recognized {#} m/(?^-{#}i)foo/', 114 '/(?^d:foo)/' => 'Sequence (?^d...) not recognized {#} m/(?^d{#}:foo)/', 115 '/(?^d)foo/' => 'Sequence (?^d...) not recognized {#} m/(?^d{#})foo/', 116 '/(?^lu:foo)/' => 'Regexp modifiers "l" and "u" are mutually exclusive {#} m/(?^lu{#}:foo)/', 117 '/(?^lu)foo/' => 'Regexp modifiers "l" and "u" are mutually exclusive {#} m/(?^lu{#})foo/', 118'/(?da:foo)/' => 'Regexp modifiers "d" and "a" are mutually exclusive {#} m/(?da{#}:foo)/', 119'/(?lil:foo)/' => 'Regexp modifier "l" may not appear twice {#} m/(?lil{#}:foo)/', 120'/(?aaia:foo)/' => 'Regexp modifier "a" may appear a maximum of twice {#} m/(?aaia{#}:foo)/', 121'/(?i-l:foo)/' => 'Regexp modifier "l" may not appear after the "-" {#} m/(?i-l{#}:foo)/', 122'/a\b{cde/' => 'Use "\b\{" instead of "\b{" {#} m/a\{#}b{cde/', 123'/a\B{cde/' => 'Use "\B\{" instead of "\B{" {#} m/a\{#}B{cde/', 124 125 '/((x)/' => 'Unmatched ( {#} m/({#}(x)/', 126 127 "/x{$inf_p1}/" => "Quantifier in {,} bigger than $inf_m1 {#} m/x{{#}$inf_p1}/", 128 129 130 '/x**/' => 'Nested quantifiers {#} m/x**{#}/', 131 132 '/x[/' => 'Unmatched [ {#} m/x[{#}/', 133 134 '/*/', => 'Quantifier follows nothing {#} m/*{#}/', 135 136 '/\p{x/' => 'Missing right brace on \p{} {#} m/\p{{#}x/', 137 138 '/[\p{x]/' => 'Missing right brace on \p{} {#} m/[\p{{#}x]/', 139 140 '/(x)\2/' => 'Reference to nonexistent group {#} m/(x)\2{#}/', 141 142 '/\g/' => 'Unterminated \g... pattern {#} m/\g{#}/', 143 '/\g{1/' => 'Unterminated \g{...} pattern {#} m/\g{1{#}/', 144 145 'my $m = "\\\"; $m =~ $m', => 'Trailing \ in regex m/\/', 146 147 '/\x{1/' => 'Missing right brace on \x{} {#} m/\x{1{#}/', 148 '/\x{X/' => 'Missing right brace on \x{} {#} m/\x{{#}X/', 149 150 '/[\x{X]/' => 'Missing right brace on \x{} {#} m/[\x{{#}X]/', 151 '/[\x{A]/' => 'Missing right brace on \x{} {#} m/[\x{A{#}]/', 152 153 '/\o{1/' => 'Missing right brace on \o{ {#} m/\o{1{#}/', 154 '/\o{X/' => 'Missing right brace on \o{ {#} m/\o{{#}X/', 155 156 '/[\o{X]/' => 'Missing right brace on \o{ {#} m/[\o{{#}X]/', 157 '/[\o{7]/' => 'Missing right brace on \o{ {#} m/[\o{7{#}]/', 158 159 '/[[:barf:]]/' => 'POSIX class [:barf:] unknown {#} m/[[:barf:]{#}]/', 160 161 '/[[=barf=]]/' => 'POSIX syntax [= =] is reserved for future extensions {#} m/[[=barf=]{#}]/', 162 163 '/[[.barf.]]/' => 'POSIX syntax [. .] is reserved for future extensions {#} m/[[.barf.]{#}]/', 164 165 '/[z-a]/' => 'Invalid [] range "z-a" {#} m/[z-a{#}]/', 166 167 '/\p/' => 'Empty \p{} {#} m/\p{#}/', 168 169 '/\P{}/' => 'Empty \P{} {#} m/\P{{#}}/', 170 '/(?[[[:word]]])/' => "Unmatched ':' in POSIX class {#} m/(?[[[:word{#}]]])/", 171 '/(?[[:word]])/' => "Unmatched ':' in POSIX class {#} m/(?[[:word{#}]])/", 172 '/(?[[[:digit: ])/' => "Unmatched '[' in POSIX class {#} m/(?[[[:digit:{#} ])/", 173 '/(?[[:digit: ])/' => "Unmatched '[' in POSIX class {#} m/(?[[:digit:{#} ])/", 174 '/(?[[[::]]])/' => "POSIX class [::] unknown {#} m/(?[[[::]{#}]])/", 175 '/(?[[[:w:]]])/' => "POSIX class [:w:] unknown {#} m/(?[[[:w:]{#}]])/", 176 '/(?[[:w:]])/' => "POSIX class [:w:] unknown {#} m/(?[[:w:]{#}])/", 177 '/(?[a])/' => 'Unexpected character {#} m/(?[a{#}])/', 178 '/(?[\t])/l' => '(?[...]) not valid in locale {#} m/(?[{#}\t])/', 179 '/(?[ + \t ])/' => 'Unexpected binary operator \'+\' with no preceding operand {#} m/(?[ +{#} \t ])/', 180 '/(?[ \cK - ( + \t ) ])/' => 'Unexpected binary operator \'+\' with no preceding operand {#} m/(?[ \cK - ( +{#} \t ) ])/', 181 '/(?[ \cK ( \t ) ])/' => 'Unexpected \'(\' with no preceding operator {#} m/(?[ \cK ({#} \t ) ])/', 182 '/(?[ \cK \t ])/' => 'Operand with no preceding operator {#} m/(?[ \cK \t{#} ])/', 183 '/(?[ \0004 ])/' => 'Need exactly 3 octal digits {#} m/(?[ \0004 {#}])/', 184 '/(?[ \05 ])/' => 'Need exactly 3 octal digits {#} m/(?[ \05 {#}])/', 185 '/(?[ \o{1038} ])/' => 'Non-octal character {#} m/(?[ \o{1038{#}} ])/', 186 '/(?[ \o{} ])/' => 'Number with no digits {#} m/(?[ \o{}{#} ])/', 187 '/(?[ \x{defg} ])/' => 'Non-hex character {#} m/(?[ \x{defg{#}} ])/', 188 '/(?[ \xabcdef ])/' => 'Use \\x{...} for more than two hex characters {#} m/(?[ \xabc{#}def ])/', 189 '/(?[ \x{} ])/' => 'Number with no digits {#} m/(?[ \x{}{#} ])/', 190 '/(?[ \cK + ) ])/' => 'Unexpected \')\' {#} m/(?[ \cK + ){#} ])/', 191 '/(?[ \cK + ])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[ \cK + {#}])/', 192 '/(?[ \p{foo} ])/' => 'Property \'foo\' is unknown {#} m/(?[ \p{foo}{#} ])/', 193 '/(?[ \p{ foo = bar } ])/' => 'Property \'foo = bar\' is unknown {#} m/(?[ \p{ foo = bar }{#} ])/', 194 '/(?[ \8 ])/' => 'Unrecognized escape \8 in character class {#} m/(?[ \8{#} ])/', 195 '/(?[ \t ]/' => 'Syntax error in (?[...]) in regex m/(?[ \t ]/', 196 '/(?[ [ \t ]/' => 'Syntax error in (?[...]) in regex m/(?[ [ \t ]/', 197 '/(?[ \t ] ]/' => 'Syntax error in (?[...]) in regex m/(?[ \t ] ]/', 198 '/(?[ [ ] ]/' => 'Syntax error in (?[...]) in regex m/(?[ [ ] ]/', 199 '/(?[ \t + \e # This was supposed to be a comment ])/' => 'Syntax error in (?[...]) in regex m/(?[ \t + \e # This was supposed to be a comment ])/', 200 '/(?[ ])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[ {#}])/', 201 'm/(?[[a-\d]])/' => 'False [] range "a-\d" {#} m/(?[[a-\d{#}]])/', 202 'm/(?[[\w-x]])/' => 'False [] range "\w-" {#} m/(?[[\w-{#}x]])/', 203 'm/(?[[a-\pM]])/' => 'False [] range "a-\pM" {#} m/(?[[a-\pM{#}]])/', 204 'm/(?[[\pM-x]])/' => 'False [] range "\pM-" {#} m/(?[[\pM-{#}x]])/', 205 'm/(?[[\N{LATIN CAPITAL LETTER A WITH MACRON AND GRAVE}]])/' => '\N{} in character class restricted to one character {#} m/(?[[\N{U+100.300{#}}]])/', 206 'm/(?[ \p{Digit} & (?(?[ \p{Thai} | \p{Lao} ]))])/' => 'Sequence (?(...) not recognized {#} m/(?[ \p{Digit} & (?({#}?[ \p{Thai} | \p{Lao} ]))])/', 207 'm/(?[ \p{Digit} & (?:(?[ \p{Thai} | \p{Lao} ]))])/' => 'Expecting \'(?flags:(?[...\' {#} m/(?[ \p{Digit} & (?{#}:(?[ \p{Thai} | \p{Lao} ]))])/', 208 'm/\o{/' => 'Missing right brace on \o{ {#} m/\o{{#}/', 209 'm/\o/' => 'Missing braces on \o{} {#} m/\o{#}/', 210 'm/\o{}/' => 'Number with no digits {#} m/\o{}{#}/', 211 'm/[\o{]/' => 'Missing right brace on \o{ {#} m/[\o{{#}]/', 212 'm/[\o]/' => 'Missing braces on \o{} {#} m/[\o{#}]/', 213 'm/[\o{}]/' => 'Number with no digits {#} m/[\o{}{#}]/', 214 'm/(?^-i:foo)/' => 'Sequence (?^-...) not recognized {#} m/(?^-{#}i:foo)/', 215 'm/\87/' => 'Reference to nonexistent group {#} m/\87{#}/', 216 'm/a\87/' => 'Reference to nonexistent group {#} m/a\87{#}/', 217 'm/a\97/' => 'Reference to nonexistent group {#} m/a\97{#}/', 218 'm/(*DOOF)/' => 'Unknown verb pattern \'DOOF\' {#} m/(*DOOF){#}/', 219 'm/(?&a/' => 'Sequence (?&... not terminated {#} m/(?&a{#}/', 220 'm/(?P=/' => 'Sequence ?P=... not terminated {#} m/(?P={#}/', 221 "m/(?'/" => "Sequence (?'... not terminated {#} m/(?'{#}/", 222 "m/(?</" => "Sequence (?<... not terminated {#} m/(?<{#}/", 223 'm/(?&/' => 'Sequence (?&... not terminated {#} m/(?&{#}/', 224 'm/(?(</' => 'Sequence (?(<... not terminated {#} m/(?(<{#}/', 225 "m/(?('/" => "Sequence (?('... not terminated {#} m/(?('{#}/", 226 'm/\g{/' => 'Sequence \g{... not terminated {#} m/\g{{#}/', 227 'm/\k</' => 'Sequence \k<... not terminated {#} m/\k<{#}/', 228 'm/\cß/' => "Character following \"\\c\" must be printable ASCII", 229); 230 231my @death_utf8 = mark_as_utf8( 232 '/ネ[[=ネ=]]ネ/' => 'POSIX syntax [= =] is reserved for future extensions {#} m/ネ[[=ネ=]{#}]ネ/', 233 '/ネ(?<= .*)/' => 'Variable length lookbehind not implemented in regex m/ネ(?<= .*)/', 234 235 '/(?<= ネ{1000})/' => 'Lookbehind longer than 255 not implemented in regex m/(?<= ネ{1000})/', 236 237 '/ネ(?ネ)ネ/' => 'Sequence (?ネ...) not recognized {#} m/ネ(?ネ{#})ネ/', 238 239 '/ネ(?(1ネ))ネ/' => 'Switch condition not recognized {#} m/ネ(?(1ネ{#}))ネ/', 240 241 '/(?(1)ネ|y|ヌ)/' => 'Switch (?(condition)... contains too many branches {#} m/(?(1)ネ|y|{#}ヌ)/', 242 243 '/(?(ネ)y|ネ)/' => 'Unknown switch condition (?(...)) {#} m/(?(ネ{#})y|ネ)/', 244 245 '/ネ(?/' => 'Sequence (? incomplete {#} m/ネ(?{#}/', 246 247 '/ネ(?;ネ/' => 'Sequence (?;...) not recognized {#} m/ネ(?;{#}ネ/', 248 '/ネ(?<;ネ/' => 'Group name must start with a non-digit word character {#} m/ネ(?<;{#}ネ/', 249 '/ネ(?\ixネ/' => 'Sequence (?\...) not recognized {#} m/ネ(?\{#}ixネ/', 250 '/ネ(?^lu:ネ)/' => 'Regexp modifiers "l" and "u" are mutually exclusive {#} m/ネ(?^lu{#}:ネ)/', 251'/ネ(?lil:ネ)/' => 'Regexp modifier "l" may not appear twice {#} m/ネ(?lil{#}:ネ)/', 252'/ネ(?aaia:ネ)/' => 'Regexp modifier "a" may appear a maximum of twice {#} m/ネ(?aaia{#}:ネ)/', 253'/ネ(?i-l:ネ)/' => 'Regexp modifier "l" may not appear after the "-" {#} m/ネ(?i-l{#}:ネ)/', 254 255 '/ネ((ネ)/' => 'Unmatched ( {#} m/ネ({#}(ネ)/', 256 257 "/ネ{$inf_p1}ネ/" => "Quantifier in {,} bigger than $inf_m1 {#} m/ネ{{#}$inf_p1}ネ/", 258 259 260 '/ネ**ネ/' => 'Nested quantifiers {#} m/ネ**{#}ネ/', 261 262 '/ネ[ネ/' => 'Unmatched [ {#} m/ネ[{#}ネ/', 263 264 '/*ネ/', => 'Quantifier follows nothing {#} m/*{#}ネ/', 265 266 '/ネ\p{ネ/' => 'Missing right brace on \p{} {#} m/ネ\p{{#}ネ/', 267 268 '/(ネ)\2ネ/' => 'Reference to nonexistent group {#} m/(ネ)\2{#}ネ/', 269 270 '/\g{ネ/; #no latin1' => 'Sequence \g{... not terminated {#} m/\g{ネ{#}/', 271 272 'my $m = "ネ\\\"; $m =~ $m', => 'Trailing \ in regex m/ネ\/', 273 274 '/\x{ネ/' => 'Missing right brace on \x{} {#} m/\x{{#}ネ/', 275 '/ネ[\x{ネ]ネ/' => 'Missing right brace on \x{} {#} m/ネ[\x{{#}ネ]ネ/', 276 '/ネ[\x{ネ]/' => 'Missing right brace on \x{} {#} m/ネ[\x{{#}ネ]/', 277 278 '/ネ\o{ネ/' => 'Missing right brace on \o{ {#} m/ネ\o{{#}ネ/', 279 '/ネ[[:ネ:]]ネ/' => 'POSIX class [:ネ:] unknown {#} m/ネ[[:ネ:]{#}]ネ/', 280 281 '/ネ[[=ネ=]]ネ/' => 'POSIX syntax [= =] is reserved for future extensions {#} m/ネ[[=ネ=]{#}]ネ/', 282 283 '/ネ[[.ネ.]]ネ/' => 'POSIX syntax [. .] is reserved for future extensions {#} m/ネ[[.ネ.]{#}]ネ/', 284 285 '/[ネ-a]ネ/' => 'Invalid [] range "ネ-a" {#} m/[ネ-a{#}]ネ/', 286 287 '/ネ\p{}ネ/' => 'Empty \p{} {#} m/ネ\p{{#}}ネ/', 288 289 '/ネ(?[[[:ネ]]])ネ/' => "Unmatched ':' in POSIX class {#} m/ネ(?[[[:ネ{#}]]])ネ/", 290 '/ネ(?[[[:ネ: ])ネ/' => "Unmatched '[' in POSIX class {#} m/ネ(?[[[:ネ:{#} ])ネ/", 291 '/ネ(?[[[::]]])ネ/' => "POSIX class [::] unknown {#} m/ネ(?[[[::]{#}]])ネ/", 292 '/ネ(?[[[:ネ:]]])ネ/' => "POSIX class [:ネ:] unknown {#} m/ネ(?[[[:ネ:]{#}]])ネ/", 293 '/ネ(?[[:ネ:]])ネ/' => "POSIX class [:ネ:] unknown {#} m/ネ(?[[:ネ:]{#}])ネ/", 294 '/ネ(?[ネ])ネ/' => 'Unexpected character {#} m/ネ(?[ネ{#}])ネ/', 295 '/ネ(?[ネ])/l' => '(?[...]) not valid in locale {#} m/ネ(?[{#}ネ])/', 296 '/ネ(?[ + [ネ] ])/' => 'Unexpected binary operator \'+\' with no preceding operand {#} m/ネ(?[ +{#} [ネ] ])/', 297 '/ネ(?[ \cK - ( + [ネ] ) ])/' => 'Unexpected binary operator \'+\' with no preceding operand {#} m/ネ(?[ \cK - ( +{#} [ネ] ) ])/', 298 '/ネ(?[ \cK ( [ネ] ) ])/' => 'Unexpected \'(\' with no preceding operator {#} m/ネ(?[ \cK ({#} [ネ] ) ])/', 299 '/ネ(?[ \cK [ネ] ])ネ/' => 'Operand with no preceding operator {#} m/ネ(?[ \cK [ネ{#}] ])ネ/', 300 '/ネ(?[ \0004 ])ネ/' => 'Need exactly 3 octal digits {#} m/ネ(?[ \0004 {#}])ネ/', 301 '/(?[ \o{ネ} ])ネ/' => 'Non-octal character {#} m/(?[ \o{ネ{#}} ])ネ/', 302 '/ネ(?[ \o{} ])ネ/' => 'Number with no digits {#} m/ネ(?[ \o{}{#} ])ネ/', 303 '/(?[ \x{ネ} ])ネ/' => 'Non-hex character {#} m/(?[ \x{ネ{#}} ])ネ/', 304 '/(?[ \p{ネ} ])/' => 'Property \'ネ\' is unknown {#} m/(?[ \p{ネ}{#} ])/', 305 '/(?[ \p{ ネ = bar } ])/' => 'Property \'ネ = bar\' is unknown {#} m/(?[ \p{ ネ = bar }{#} ])/', 306 '/ネ(?[ \t ]/' => 'Syntax error in (?[...]) in regex m/ネ(?[ \t ]/', 307 '/(?[ \t + \e # ネ This was supposed to be a comment ])/' => 'Syntax error in (?[...]) in regex m/(?[ \t + \e # ネ This was supposed to be a comment ])/', 308 'm/(*ネ)ネ/' => q<Unknown verb pattern 'ネ' {#} m/(*ネ){#}ネ/>, 309 '/\cネ/' => "Character following \"\\c\" must be printable ASCII", 310); 311push @death, @death_utf8; 312 313# Tests involving a user-defined charnames translator are in pat_advanced.t 314 315# In the following arrays of warnings, the value can be an array of things to 316# expect. If the empty string, it means no warning should be raised. 317 318## 319## Key-value pairs of code/error of code that should have non-fatal regexp warnings. 320## 321my @warning = ( 322 'm/\b*/' => '\b* matches null string many times {#} m/\b*{#}/', 323 'm/[:blank:]/' => 'POSIX syntax [: :] belongs inside character classes {#} m/[:blank:]{#}/', 324 325 "m'[\\y]'" => 'Unrecognized escape \y in character class passed through {#} m/[\y{#}]/', 326 327 'm/[a-\d]/' => 'False [] range "a-\d" {#} m/[a-\d{#}]/', 328 'm/[\w-x]/' => 'False [] range "\w-" {#} m/[\w-{#}x]/', 329 'm/[a-\pM]/' => 'False [] range "a-\pM" {#} m/[a-\pM{#}]/', 330 'm/[\pM-x]/' => 'False [] range "\pM-" {#} m/[\pM-{#}x]/', 331 "m'\\y'" => 'Unrecognized escape \y passed through {#} m/\y{#}/', 332 '/x{3,1}/' => 'Quantifier {n,m} with n > m can\'t match {#} m/x{3,1}{#}/', 333 '/\08/' => '\'\08\' resolved to \'\o{0}8\' {#} m/\08{#}/', 334 '/\018/' => '\'\018\' resolved to \'\o{1}8\' {#} m/\018{#}/', 335 '/[\08]/' => '\'\08\' resolved to \'\o{0}8\' {#} m/[\08{#}]/', 336 '/[\018]/' => '\'\018\' resolved to \'\o{1}8\' {#} m/[\018{#}]/', 337 '/(?=a)*/' => '(?=a)* matches null string many times {#} m/(?=a)*{#}/', 338 'my $x = \'\m\'; qr/a$x/' => 'Unrecognized escape \m passed through {#} m/a\m{#}/', 339 '/\q/' => 'Unrecognized escape \q passed through {#} m/\q{#}/', 340 '/\q{/' => 'Unrecognized escape \q{ passed through {#} m/\q{{#}/', 341 '/(?=a){1,3}/' => 'Quantifier unexpected on zero-length expression {#} m/(?=a){1,3}{#}/', 342 '/(a|b)(?=a){3}/' => 'Quantifier unexpected on zero-length expression {#} m/(a|b)(?=a){3}{#}/', 343 '/\_/' => "", 344 '/[\_\0]/' => "", 345 '/[\07]/' => "", 346 '/[\006]/' => "", 347 '/[\0005]/' => "", 348 '/[\8\9]/' => ['Unrecognized escape \8 in character class passed through {#} m/[\8{#}\9]/', 349 'Unrecognized escape \9 in character class passed through {#} m/[\8\9{#}]/', 350 ], 351 '/[:alpha:]/' => 'POSIX syntax [: :] belongs inside character classes {#} m/[:alpha:]{#}/', 352 '/[:zog:]/' => 'POSIX syntax [: :] belongs inside character classes {#} m/[:zog:]{#}/', 353 '/[.zog.]/' => 'POSIX syntax [. .] belongs inside character classes {#} m/[.zog.]{#}/', 354 '/[a-b]/' => "", 355 '/[a-\d]/' => 'False [] range "a-\d" {#} m/[a-\d{#}]/', 356 '/[\d-b]/' => 'False [] range "\d-" {#} m/[\d-{#}b]/', 357 '/[\s-\d]/' => 'False [] range "\s-" {#} m/[\s-{#}\d]/', 358 '/[\d-\s]/' => 'False [] range "\d-" {#} m/[\d-{#}\s]/', 359 '/[a-[:digit:]]/' => 'False [] range "a-[:digit:]" {#} m/[a-[:digit:]{#}]/', 360 '/[[:digit:]-b]/' => 'False [] range "[:digit:]-" {#} m/[[:digit:]-{#}b]/', 361 '/[[:alpha:]-[:digit:]]/' => 'False [] range "[:alpha:]-" {#} m/[[:alpha:]-{#}[:digit:]]/', 362 '/[[:digit:]-[:alpha:]]/' => 'False [] range "[:digit:]-" {#} m/[[:digit:]-{#}[:alpha:]]/', 363 '/[a\zb]/' => 'Unrecognized escape \z in character class passed through {#} m/[a\z{#}b]/', 364 '/(?c)/' => 'Useless (?c) - use /gc modifier {#} m/(?c{#})/', 365 '/(?-c)/' => 'Useless (?-c) - don\'t use /gc modifier {#} m/(?-c{#})/', 366 '/(?g)/' => 'Useless (?g) - use /g modifier {#} m/(?g{#})/', 367 '/(?-g)/' => 'Useless (?-g) - don\'t use /g modifier {#} m/(?-g{#})/', 368 '/(?o)/' => 'Useless (?o) - use /o modifier {#} m/(?o{#})/', 369 '/(?-o)/' => 'Useless (?-o) - don\'t use /o modifier {#} m/(?-o{#})/', 370 '/(?g-o)/' => [ 'Useless (?g) - use /g modifier {#} m/(?g{#}-o)/', 371 'Useless (?-o) - don\'t use /o modifier {#} m/(?g-o{#})/', 372 ], 373 '/(?g-c)/' => [ 'Useless (?g) - use /g modifier {#} m/(?g{#}-c)/', 374 'Useless (?-c) - don\'t use /gc modifier {#} m/(?g-c{#})/', 375 ], 376 # (?c) means (?g) error won't be thrown 377 '/(?o-cg)/' => [ 'Useless (?o) - use /o modifier {#} m/(?o{#}-cg)/', 378 'Useless (?-c) - don\'t use /gc modifier {#} m/(?o-c{#}g)/', 379 ], 380 '/(?ogc)/' => [ 'Useless (?o) - use /o modifier {#} m/(?o{#}gc)/', 381 'Useless (?g) - use /g modifier {#} m/(?og{#}c)/', 382 'Useless (?c) - use /gc modifier {#} m/(?ogc{#})/', 383 ], 384 '/a{1,1}?/' => 'Useless use of greediness modifier \'?\' {#} m/a{1,1}?{#}/', 385); 386 387my @warnings_utf8 = mark_as_utf8( 388 'm/ネ\b*ネ/' => '\b* matches null string many times {#} m/ネ\b*{#}ネ/', 389 '/(?=ネ)*/' => '(?=ネ)* matches null string many times {#} m/(?=ネ)*{#}/', 390 'm/ネ[:foo:]ネ/' => 'POSIX syntax [: :] belongs inside character classes {#} m/ネ[:foo:]{#}ネ/', 391 "m'ネ[\\y]ネ'" => 'Unrecognized escape \y in character class passed through {#} m/ネ[\y{#}]ネ/', 392 'm/ネ[ネ-\d]ネ/' => 'False [] range "ネ-\d" {#} m/ネ[ネ-\d{#}]ネ/', 393 'm/ネ[\w-ネ]ネ/' => 'False [] range "\w-" {#} m/ネ[\w-{#}ネ]ネ/', 394 'm/ネ[ネ-\pM]ネ/' => 'False [] range "ネ-\pM" {#} m/ネ[ネ-\pM{#}]ネ/', 395 '/ネ[ネ-[:digit:]]ネ/' => 'False [] range "ネ-[:digit:]" {#} m/ネ[ネ-[:digit:]{#}]ネ/', 396 '/ネ[\d-\s]ネ/' => 'False [] range "\d-" {#} m/ネ[\d-{#}\s]ネ/', 397 '/ネ[a\zb]ネ/' => 'Unrecognized escape \z in character class passed through {#} m/ネ[a\z{#}b]ネ/', 398 '/ネ(?c)ネ/' => 'Useless (?c) - use /gc modifier {#} m/ネ(?c{#})ネ/', 399 '/utf8 ネ (?ogc) ネ/' => [ 400 'Useless (?o) - use /o modifier {#} m/utf8 ネ (?o{#}gc) ネ/', 401 'Useless (?g) - use /g modifier {#} m/utf8 ネ (?og{#}c) ネ/', 402 'Useless (?c) - use /gc modifier {#} m/utf8 ネ (?ogc{#}) ネ/', 403 ], 404 405); 406 407push @warning, @warnings_utf8; 408 409my @experimental_regex_sets = ( 410 '/(?[ \t ])/' => 'The regex_sets feature is experimental {#} m/(?[{#} \t ])/', 411 'use utf8; /utf8 ネ (?[ [\tネ] ])/' => do { use utf8; 'The regex_sets feature is experimental {#} m/utf8 ネ (?[{#} [\tネ] ])/' }, 412 '/noutf8 ネ (?[ [\tネ] ])/' => 'The regex_sets feature is experimental {#} m/noutf8 ネ (?[{#} [\tネ] ])/', 413); 414 415my @deprecated = ( 416 "/(?x)latin1\\\x{85}\x{85}\\\x{85}/" => 'Escape literal pattern white space under /x {#} ' . "m/(?x)latin1\\\x{85}\x{85}{#}\\\x{85}/", 417 'use utf8; /(?x)utf8\……\…/' => 'Escape literal pattern white space under /x {#} ' . "m/(?x)utf8\\\N{NEXT LINE}\N{NEXT LINE}{#}\\\N{NEXT LINE}/", 418 '/((?# This is a comment in the middle of a token)?:foo)/' => 'In \'(?...)\', splitting the initial \'(?\' is deprecated {#} m/((?# This is a comment in the middle of a token)?{#}:foo)/', 419 '/((?# This is a comment in the middle of a token)*FAIL)/' => 'In \'(*VERB...)\', splitting the initial \'(*\' is deprecated {#} m/((?# This is a comment in the middle of a token)*{#}FAIL)/', 420); 421 422while (my ($regex, $expect) = splice @death, 0, 2) { 423 my $expect = fixup_expect($expect); 424 no warnings 'experimental::regex_sets'; 425 # skip the utf8 test on EBCDIC since they do not die 426 next if $::IS_EBCDIC && $regex =~ /utf8/; 427 428 warning_is(sub { 429 $_ = "x"; 430 eval $regex; 431 like($@, qr/\Q$expect/, $regex); 432 }, undef, "... and died without any other warnings"); 433} 434 435foreach my $ref (\@warning, \@experimental_regex_sets, \@deprecated) { 436 my $warning_type = ($ref == \@warning) 437 ? 'regexp' 438 : ($ref == \@deprecated) 439 ? 'regexp, deprecated' 440 : 'experimental::regex_sets'; 441 while (my ($regex, $expect) = splice @$ref, 0, 2) { 442 my @expect = fixup_expect($expect); 443 { 444 $_ = "x"; 445 no warnings; 446 eval $regex; 447 } 448 if (is($@, "", "$regex did not die")) { 449 my @got = capture_warnings(sub { 450 $_ = "x"; 451 eval $regex }); 452 my $count = @expect; 453 if (! is(scalar @got, scalar @expect, "... and gave expected number ($count) of warnings")) { 454 if (@got < @expect) { 455 $count = @got; 456 note "Expected warnings not gotten:\n\t" . join "\n\t", @expect[$count .. $#expect]; 457 } 458 else { 459 note "Unexpected warnings gotten:\n\t" . join("\n\t", @got[$count .. $#got]); 460 } 461 } 462 foreach my $i (0 .. $count - 1) { 463 if (! like($got[$i], qr/\Q$expect[$i]/, "... and gave expected warning")) { 464 chomp($got[$i]); 465 chomp($expect[$i]); 466 diag("GOT\n'$got[$i]'\nEXPECT\n'$expect[$i]'"); 467 } 468 else { 469 ok (0 == capture_warnings(sub { 470 $_ = "x"; 471 eval "no warnings '$warning_type'; $regex;" } 472 ), 473 "... and turning off '$warning_type' warnings suppressed it"); 474 # Test that whether the warning is on by default is 475 # correct. Experimental and deprecated warnings are; 476 # others are not. This test relies on the fact that we 477 # are outside the scope of any ‘use warnings’. 478 local $^W; 479 my $on = 'on' x ($warning_type ne 'regexp'); 480 ok !!$on == 481 capture_warnings(sub { $_ = "x"; eval $regex }), 482 "... and the warning is " . ($on||'off') 483 . " by default"; 484 } 485 } 486 } 487 } 488} 489 490done_testing(); 491