basis/regexp/regexp-tests.factor

   1 USING: arrays regexp tools.test kernel sequences regexp.parser
   2 regexp.private eval strings multiline accessors ;
   3 IN: regexp.tests
   4
   5 { f } [ "b" "a*" <regexp> matches? ] unit-test
   6 { t } [ "" "a*" <regexp> matches? ] unit-test
   7 { t } [ "a" "a*" <regexp> matches? ] unit-test
   8 { t } [ "aaaaaaa" "a*"  <regexp> matches? ] unit-test
   9 { f } [ "ab" "a*" <regexp> matches? ] unit-test
  10
  11 { t } [ "abc" "abc" <regexp> matches? ] unit-test
  12 { t } [ "a" "a|b|c" <regexp> matches? ] unit-test
  13 { t } [ "b" "a|b|c" <regexp> matches? ] unit-test
  14 { t } [ "c" "a|b|c" <regexp> matches? ] unit-test
  15 { f } [ "c" "d|e|f" <regexp> matches? ] unit-test
  16
  17 { t } [ "b" "|b" <regexp> matches? ] unit-test
  18 { t } [ "b" "b|" <regexp> matches? ] unit-test
  19 { t } [ "" "b|" <regexp> matches? ] unit-test
  20 { t } [ "" "b|" <regexp> matches? ] unit-test
  21 { t } [ "" "|" <regexp> matches? ] unit-test
  22 { t } [ "" "|||||||" <regexp> matches? ] unit-test
  23
  24 { f } [ "aa" "a|b|c" <regexp> matches? ] unit-test
  25 { f } [ "bb" "a|b|c" <regexp> matches? ] unit-test
  26 { f } [ "cc" "a|b|c" <regexp> matches? ] unit-test
  27 { f } [ "cc" "d|e|f" <regexp> matches? ] unit-test
  28
  29 { f } [ "" "a+" <regexp> matches? ] unit-test
  30 { t } [ "a" "a+" <regexp> matches? ] unit-test
  31 { t } [ "aa" "a+" <regexp> matches? ] unit-test
  32
  33 { t } [ "" "a?" <regexp> matches? ] unit-test
  34 { t } [ "a" "a?" <regexp> matches? ] unit-test
  35 { f } [ "aa" "a?" <regexp> matches? ] unit-test
  36
  37 { f } [ "" "." <regexp> matches? ] unit-test
  38 { t } [ "a" "." <regexp> matches? ] unit-test
  39 { t } [ "." "." <regexp> matches? ] unit-test
  40
  41 ! Dotall mode -- when on, . matches newlines.
  42 ! Off by default.
  43 { f } [ "\n" "." <regexp> matches? ] unit-test
  44 { t } [ "\n" "(?s:.)" <regexp> matches? ] unit-test
  45 { t } [ "\n" R/ ./s matches? ] unit-test
  46 { f } [ "\n\n" "(?s:.)." <regexp> matches? ] unit-test
  47
  48 { f } [ "" ".+" <regexp> matches? ] unit-test
  49 { t } [ "a" ".+" <regexp> matches? ] unit-test
  50 { t } [ "ab" ".+" <regexp> matches? ] unit-test
  51
  52 { t } [ "\0" "[\\0]" <regexp> matches? ] unit-test
  53 { f } [ "0" "[\\0]" <regexp> matches? ] unit-test
  54
  55 { t } [ " " "[\\s]" <regexp> matches? ] unit-test
  56 { f } [ "a" "[\\s]" <regexp> matches? ] unit-test
  57 { f } [ " " "[\\S]" <regexp> matches? ] unit-test
  58 { t } [ "a" "[\\S]" <regexp> matches? ] unit-test
  59 { f } [ " " "[\\w]" <regexp> matches? ] unit-test
  60 { t } [ "a" "[\\w]" <regexp> matches? ] unit-test
  61 { t } [ " " "[\\W]" <regexp> matches? ] unit-test
  62 { f } [ "a" "[\\W]" <regexp> matches? ] unit-test
  63
  64 { t } [ "/" "\\/" <regexp> matches? ] unit-test
  65
  66 { t } [ "a" R/ a/i matches? ] unit-test
  67
  68 { t } [ "" "a|b*|c+|d?" <regexp> matches? ] unit-test
  69 { t } [ "a" "a|b*|c+|d?" <regexp> matches? ] unit-test
  70 { t } [ "c" "a|b*|c+|d?" <regexp> matches? ] unit-test
  71 { t } [ "cc" "a|b*|c+|d?" <regexp> matches? ] unit-test
  72 { f } [ "ccd" "a|b*|c+|d?" <regexp> matches? ] unit-test
  73 { t } [ "d" "a|b*|c+|d?" <regexp> matches? ] unit-test
  74
  75 { t } [ "foo" "foo|bar" <regexp> matches? ] unit-test
  76 { t } [ "bar" "foo|bar" <regexp> matches? ] unit-test
  77 { f } [ "foobar" "foo|bar" <regexp> matches? ] unit-test
  78
  79 { f } [ "" "(a)" <regexp> matches? ] unit-test
  80 { t } [ "a" "(a)" <regexp> matches? ] unit-test
  81 { f } [ "aa" "(a)" <regexp> matches? ] unit-test
  82 { t } [ "aa" "(a*)" <regexp> matches? ] unit-test
  83
  84 { f } [ "aababaaabbac" "(a|b)+" <regexp> matches? ] unit-test
  85 { t } [ "ababaaabba" "(a|b)+" <regexp> matches? ] unit-test
  86
  87 { f } [ "" "a{1}" <regexp> matches? ] unit-test
  88 { t } [ "a" "a{1}" <regexp> matches? ] unit-test
  89 { f } [ "aa" "a{1}" <regexp> matches? ] unit-test
  90
  91 { f } [ "a" "a{2,}" <regexp> matches? ] unit-test
  92 { t } [ "aaa" "a{2,}" <regexp> matches? ] unit-test
  93 { t } [ "aaaa" "a{2,}" <regexp> matches? ] unit-test
  94 { t } [ "aaaaa" "a{2,}" <regexp> matches? ] unit-test
  95
  96 { t } [ "" "a{,2}" <regexp> matches? ] unit-test
  97 { t } [ "a" "a{,2}" <regexp> matches? ] unit-test
  98 { t } [ "aa" "a{,2}" <regexp> matches? ] unit-test
  99 { f } [ "aaa" "a{,2}" <regexp> matches? ] unit-test
 100 { f } [ "aaaa" "a{,2}" <regexp> matches? ] unit-test
 101 { f } [ "aaaaa" "a{,2}" <regexp> matches? ] unit-test
 102
 103 { f } [ "" "a{1,3}" <regexp> matches? ] unit-test
 104 { t } [ "a" "a{1,3}" <regexp> matches? ] unit-test
 105 { t } [ "aa" "a{1,3}" <regexp> matches? ] unit-test
 106 { t } [ "aaa" "a{1,3}" <regexp> matches? ] unit-test
 107 { f } [ "aaaa" "a{1,3}" <regexp> matches? ] unit-test
 108
 109 { f } [ "" "[a]" <regexp> matches? ] unit-test
 110 { t } [ "a" "[a]" <regexp> matches? ] unit-test
 111 { t } [ "a" "[abc]" <regexp> matches? ] unit-test
 112 { f } [ "b" "[a]" <regexp> matches? ] unit-test
 113 { f } [ "d" "[abc]" <regexp> matches? ] unit-test
 114 { t } [ "ab" "[abc]{1,2}" <regexp> matches? ] unit-test
 115 { f } [ "abc" "[abc]{1,2}" <regexp> matches? ] unit-test
 116
 117 { f } [ "" "[^a]" <regexp> matches? ] unit-test
 118 { f } [ "a" "[^a]" <regexp> matches? ] unit-test
 119 { f } [ "a" "[^abc]" <regexp> matches? ] unit-test
 120 { t } [ "b" "[^a]" <regexp> matches? ] unit-test
 121 { t } [ "d" "[^abc]" <regexp> matches? ] unit-test
 122 { f } [ "ab" "[^abc]{1,2}" <regexp> matches? ] unit-test
 123 { f } [ "abc" "[^abc]{1,2}" <regexp> matches? ] unit-test
 124
 125 { t } [ "]" "[]]" <regexp> matches? ] unit-test
 126 { f } [ "]" "[^]]" <regexp> matches? ] unit-test
 127 { t } [ "a" "[^]]" <regexp> matches? ] unit-test
 128
 129 [ "^" "[^]" <regexp> matches? ] must-fail
 130 { t } [ "^" "[]^]" <regexp> matches? ] unit-test
 131 { t } [ "]" "[]^]" <regexp> matches? ] unit-test
 132
 133 { t } [ "[" "[[]" <regexp> matches? ] unit-test
 134 { f } [ "^" "[^^]" <regexp> matches? ] unit-test
 135 { t } [ "a" "[^^]" <regexp> matches? ] unit-test
 136
 137 { t } [ "-" "[-]" <regexp> matches? ] unit-test
 138 { f } [ "a" "[-]" <regexp> matches? ] unit-test
 139 { f } [ "-" "[^-]" <regexp> matches? ] unit-test
 140 { t } [ "a" "[^-]" <regexp> matches? ] unit-test
 141
 142 { t } [ "-" "[-a]" <regexp> matches? ] unit-test
 143 { t } [ "a" "[-a]" <regexp> matches? ] unit-test
 144 { t } [ "-" "[a-]" <regexp> matches? ] unit-test
 145 { t } [ "a" "[a-]" <regexp> matches? ] unit-test
 146 { f } [ "b" "[a-]" <regexp> matches? ] unit-test
 147 { f } [ "-" "[^-]" <regexp> matches? ] unit-test
 148 { t } [ "a" "[^-]" <regexp> matches? ] unit-test
 149
 150 { f } [ "-" "[a-c]" <regexp> matches? ] unit-test
 151 { t } [ "-" "[^a-c]" <regexp> matches? ] unit-test
 152 { t } [ "b" "[a-c]" <regexp> matches? ] unit-test
 153 { f } [ "b" "[^a-c]" <regexp> matches? ] unit-test
 154
 155 { t } [ "-" "[a-c-]" <regexp> matches? ] unit-test
 156 { f } [ "-" "[^a-c-]" <regexp> matches? ] unit-test
 157
 158 { t } [ "\\" "[\\\\]" <regexp> matches? ] unit-test
 159 { f } [ "a" "[\\\\]" <regexp> matches? ] unit-test
 160 { f } [ "\\" "[^\\\\]" <regexp> matches? ] unit-test
 161 { t } [ "a" "[^\\\\]" <regexp> matches? ] unit-test
 162
 163 { t } [ "0" "[\\d]" <regexp> matches? ] unit-test
 164 { f } [ "a" "[\\d]" <regexp> matches? ] unit-test
 165 { f } [ "0" "[^\\d]" <regexp> matches? ] unit-test
 166 { t } [ "a" "[^\\d]" <regexp> matches? ] unit-test
 167
 168 { t } [ "a" "[a-z]{1,}|[A-Z]{2,4}|b*|c|(f|g)*" <regexp> matches? ] unit-test
 169 { t } [ "a" "[a-z]{1,2}|[A-Z]{3,3}|b*|c|(f|g)*" <regexp> matches? ] unit-test
 170 { t } [ "a" "[a-z]{1,2}|[A-Z]{3,3}" <regexp> matches? ] unit-test
 171
 172 { t } [ "1000" "\\d{4,6}" <regexp> matches? ] unit-test
 173 { t } [ "1000" "[0-9]{4,6}" <regexp> matches? ] unit-test
 174
 175 { t } [ "abc" "\\p{Lower}{3}" <regexp> matches? ] unit-test
 176 { f } [ "ABC" "\\p{Lower}{3}" <regexp> matches? ] unit-test
 177 { t } [ "ABC" "\\p{Upper}{3}" <regexp> matches? ] unit-test
 178 { f } [ "abc" "\\p{Upper}{3}" <regexp> matches? ] unit-test
 179 { f } [ "abc" "[\\p{Upper}]{3}" <regexp> matches? ] unit-test
 180 { t } [ "ABC" "[\\p{Upper}]{3}" <regexp> matches? ] unit-test
 181
 182 { t } [ "" "\\Q\\E" <regexp> matches? ] unit-test
 183 { f } [ "a" "\\Q\\E" <regexp> matches? ] unit-test
 184 { t } [ "|*+" "\\Q|*+\\E" <regexp> matches? ] unit-test
 185 { f } [ "abc" "\\Q|*+\\E" <regexp> matches? ] unit-test
 186 { t } [ "s" "\\Qs\\E" <regexp> matches? ] unit-test
 187
 188 { t } [ "S" "\\0123" <regexp> matches? ] unit-test
 189 { t } [ "SXY" "\\0123XY" <regexp> matches? ] unit-test
 190 { t } [ "x" "\\x78" <regexp> matches? ] unit-test
 191 { f } [ "y" "\\x78" <regexp> matches? ] unit-test
 192 { t } [ "x" "\\u0078" <regexp> matches? ] unit-test
 193 { f } [ "y" "\\u0078" <regexp> matches? ] unit-test
 194
 195 { t } [ "ab" "a+b" <regexp> matches? ] unit-test
 196 { f } [ "b" "a+b" <regexp> matches? ] unit-test
 197 { t } [ "aab" "a+b" <regexp> matches? ] unit-test
 198 { f } [ "abb" "a+b" <regexp> matches? ] unit-test
 199
 200 { t } [ "abbbb" "ab*" <regexp> matches? ] unit-test
 201 { t } [ "a" "ab*" <regexp> matches? ] unit-test
 202 { f } [ "abab" "ab*" <regexp> matches? ] unit-test
 203
 204 { f } [ "x" "\\." <regexp> matches? ] unit-test
 205 { t } [ "." "\\." <regexp> matches? ] unit-test
 206
 207 { t } [ "aaaab" "a+ab" <regexp> matches? ] unit-test
 208 { f } [ "aaaxb" "a+ab" <regexp> matches? ] unit-test
 209 { t } [ "aaacb" "a+cb" <regexp> matches? ] unit-test
 210
 211 { "aaa" } [ "aaacb" "a*" <regexp> first-match >string ] unit-test
 212 { "aa" } [ "aaacb" "aa?" <regexp> first-match >string ] unit-test
 213
 214 { t } [ "aaa" R/ AAA/i matches? ] unit-test
 215 { f } [ "aax" R/ AAA/i matches? ] unit-test
 216 { t } [ "aaa" R/ A*/i matches? ] unit-test
 217 { f } [ "aaba" R/ A*/i matches? ] unit-test
 218 { t } [ "b" R/ [AB]/i matches? ] unit-test
 219 { f } [ "c" R/ [AB]/i matches? ] unit-test
 220 { t } [ "c" R/ [A-Z]/i matches? ] unit-test
 221 { f } [ "3" R/ [A-Z]/i matches? ] unit-test
 222
 223 { t } [ "a" "(?i:a)" <regexp> matches? ] unit-test
 224 { t } [ "a" "(?i:a)" <regexp> matches? ] unit-test
 225 { t } [ "A" "(?i:a)" <regexp> matches? ] unit-test
 226 { t } [ "A" "(?i:a)" <regexp> matches? ] unit-test
 227
 228 { t } [ "a" R/ (?-i:a)/i matches? ] unit-test
 229 { t } [ "a" R/ (?-i:a)/i matches? ] unit-test
 230 { f } [ "A" R/ (?-i:a)/i matches? ] unit-test
 231 { f } [ "A" R/ (?-i:a)/i matches? ] unit-test
 232
 233 { f } [ "A" "[a-z]" <regexp> matches? ] unit-test
 234 { t } [ "A" R/ [a-z]/i matches? ] unit-test
 235
 236 { f } [ "A" "\\p{Lower}" <regexp> matches? ] unit-test
 237 { t } [ "A" R/ \p{Lower}/i matches? ] unit-test
 238
 239 { t } [ "abc" R/ abc/r matches? ] unit-test
 240 { t } [ "abc" R/ a[bB][cC]/r matches? ] unit-test
 241
 242 { t } [ 3 "xabc" R/ abc/r match-index-from >boolean ] unit-test
 243 { t } [ 3 "xabc" R/ a[bB][cC]/r match-index-from >boolean ] unit-test
 244
 245 { 2 } [ 0 "llamallol" R/ ll/ match-index-from ] unit-test
 246 { 5 } [ 8 "lolmallol" R/ lol/r match-index-from ] unit-test
 247
 248 { t } [ "s@f" "[a-z.-]@[a-z]" <regexp> matches? ] unit-test
 249 { f } [ "a" "[a-z.-]@[a-z]" <regexp> matches? ] unit-test
 250 { t } [ ".o" "\\.[a-z]" <regexp> matches? ] unit-test
 251
 252 { t } [ "abc*" "[^\\*]*\\*" <regexp> matches? ] unit-test
 253 { t } [ "bca" "[^a]*a" <regexp> matches? ] unit-test
 254
 255 { } [
 256     "(0[lL]?|[1-9]\\d{0,9}(\\d{0,9}[lL])?|0[xX]\\p{XDigit}{1,8}(\\p{XDigit}{0,8}[lL])?|0[0-7]{1,11}([0-7]{0,11}[lL])?|([0-9]+\\.[0-9]*|\\.[0-9]+)([eE][+-]?[0-9]+)?[fFdD]?|[0-9]+([eE][+-]?[0-9]+[fFdD]?|([eE][+-]?[0-9]+)?[fFdD]))"
 257     <regexp> drop
 258 ] unit-test
 259
 260 { } [ "(\\$[\\p{XDigit}]|[\\p{Digit}])" <regexp> drop ] unit-test
 261
 262 ! Comment inside a regular expression
 263 { t } [ "ac" "a(?#boo)c" <regexp> matches? ] unit-test
 264
 265 { } [ "USING: regexp kernel ; R/ -{3}[+]{1,6}(?:!!)?\\s/ drop" eval( -- ) ] unit-test
 266
 267 { } [ "USING: regexp kernel ; R/ (ftp|http|https):\\/\\/(\\w+:?\\w*@)?(\\S+)(:[0-9]+)?(\\/\\|\\/([\\w#!:.?+=&%@!\\-\\/]))?/ drop" eval( -- ) ] unit-test
 268
 269 { } [ "USING: regexp kernel ; R/ \\*[^\s*][^*]*\\*/ drop" eval( -- ) ] unit-test
 270
 271 { "ab" } [ "ab" "(a|ab)(bc)?" <regexp> first-match >string ] unit-test
 272 { "abc" } [ "abc" "(a|ab)(bc)?" <regexp> first-match >string ] unit-test
 273
 274 { "ab" } [ "ab" "(ab|a)(bc)?" <regexp> first-match >string ] unit-test
 275 { "abc" } [ "abc" "(ab|a)(bc)?" <regexp> first-match >string ] unit-test
 276
 277 { "b" } [ "aaaaaaaaaaaaaaaaaaaaaaab" "((a*)*b)*b" <regexp> first-match >string ] unit-test
 278
 279 { T{ slice { from 5 } { to 10 } { seq "hellohello" } } }
 280 [ "hellohello" R/ hello/r first-match ]
 281 unit-test
 282
 283 { { "1" "2" "3" "4" } }
 284 [ "1ABC2DEF3GHI4" R/ [A-Z]+/ re-split [ >string ] map ] unit-test
 285
 286 { { "1" "2" "3" "4" "" } }
 287 [ "1ABC2DEF3GHI4JK" R/ [A-Z]+/ re-split [ >string ] map ] unit-test
 288
 289 { { "" } } [ "" R/ =/ re-split [ >string ] map ] unit-test
 290
 291 { { "a" "" } } [ "a=" R/ =/ re-split [ >string ] map ] unit-test
 292
 293 { { "he" "o" } } [ "hello" R/ l+/ re-split [ >string ] map ] unit-test
 294
 295 { { "h" "llo" } } [ "hello" R/ e+/ re-split [ >string ] map ] unit-test
 296
 297 { { "" "h" "" "l" "l" "o" "" } } [ "hello" R/ e*/ re-split [ >string ] map ] unit-test
 298
 299 { { { 0 5 "hellohello" } { 5 10 "hellohello" } } }
 300 [ "hellohello" R/ hello/ [ 3array ] map-matches ]
 301 unit-test
 302
 303 { { { 5 10 "hellohello" } { 0 5 "hellohello" } } }
 304 [ "hellohello" R/ hello/r [ 3array ] map-matches ]
 305 unit-test
 306
 307 { { "ABC" "DEF" "GHI" } }
 308 [ "1ABC2DEF3GHI4" R/ [A-Z]+/ all-matching-subseqs ] unit-test
 309
 310 { { "ee" "e" } } [ "heellohello" R/ e+/ all-matching-subseqs ] unit-test
 311 { { "e" "ee" } } [ "heellohello" R/ e+/r all-matching-subseqs ] unit-test
 312
 313 { 3 } [ "1ABC2DEF3GHI4" R/ [A-Z]+/ count-matches ] unit-test
 314
 315 { 3 } [ "1ABC2DEF3GHI4" R/ [A-Z]+/r count-matches ] unit-test
 316
 317 { 1 } [ "" R/ / count-matches ] unit-test
 318
 319 { 1 } [ "" R/ /r count-matches ] unit-test
 320
 321 { 0 } [ "123" R/ [A-Z]+/ count-matches ] unit-test
 322
 323 { 0 } [ "123" R/ [A-Z]+/r count-matches ] unit-test
 324
 325 { 6 } [ "hello" R/ e*/ count-matches ] unit-test
 326
 327 { 6 } [ "hello" R/ e*/r count-matches ] unit-test
 328
 329 { 11 } [ "hello world" R/ l*/ count-matches ] unit-test
 330
 331 { 11 } [ "hello world" R/ l*/r count-matches ] unit-test
 332
 333 { 1 } [ "hello" R/ e+/ count-matches ] unit-test
 334
 335 { 2 } [ "hello world" R/ l+/r count-matches ] unit-test
 336
 337 { "1.2.3.4." } [ "1ABC2DEF3GHI4JK" R/ [A-Z]+/ "." re-replace ] unit-test
 338 { "XhXXlXlXoX XwXoXrXlXdX" } [ "hello world" R/ e*/ "X" re-replace ] unit-test
 339 { "-- title --" } [ "== title ==" R/ =/ "-" re-replace ] unit-test
 340
 341 { "abc" } [ "a/   \\bc" "/.*\\" <regexp> "" re-replace ] unit-test
 342 { "ac" } [ "a/   \\bc" R/ \/.*\\./ "" re-replace ] unit-test
 343 { "abc" } [ "a/   \\bc" R/ \/.*\\/ "" re-replace ] unit-test
 344
 345 { "" } [ "ab" "a(?!b)" <regexp> first-match >string ] unit-test
 346 { "a" } [ "ac" "a(?!b)" <regexp> first-match >string ] unit-test
 347 { t } [ "fxxbar" ".{3}(?!foo)bar" <regexp> matches? ] unit-test
 348 { t } [ "foobar" ".{3}(?!foo)bar" <regexp> matches? ] unit-test
 349 { t } [ "fxxbar" "(?!foo).{3}bar" <regexp> matches? ] unit-test
 350 { f } [ "foobar" "(?!foo).{3}bar" <regexp> matches? ] unit-test
 351 { "a" } [ "ab" "a(?=b)(?=b)" <regexp> first-match >string ] unit-test
 352 { "a" } [ "ba" "(?<=b)(?<=b)a" <regexp> first-match >string ] unit-test
 353 { "a" } [ "cab" "(?<=c)a(?=b)" <regexp> first-match >string ] unit-test
 354
 355 { 3 } [ "foobar" "foo(?=bar)" <regexp> first-match length ] unit-test
 356 { f } [ "foobxr" "foo(?=bar)" <regexp> first-match ] unit-test
 357
 358 ! Bug in parsing word
 359 { t } [ "a" R/ a/ matches? ] unit-test
 360
 361 ! Testing negation
 362 { f } [ "a" R/ (?~a)/ matches? ] unit-test
 363 { t } [ "aa" R/ (?~a)/ matches? ] unit-test
 364 { t } [ "bb" R/ (?~a)/ matches? ] unit-test
 365 { t } [ "" R/ (?~a)/ matches? ] unit-test
 366
 367 { f } [ "a" R/ (?~a+|b)/ matches? ] unit-test
 368 { f } [ "aa" R/ (?~a+|b)/ matches? ] unit-test
 369 { t } [ "bb" R/ (?~a+|b)/ matches? ] unit-test
 370 { f } [ "b" R/ (?~a+|b)/ matches? ] unit-test
 371 { t } [ "" R/ (?~a+|b)/ matches? ] unit-test
 372
 373 ! Intersecting classes
 374 { t } [ "ab" R/ ac|\p{Lower}b/ matches? ] unit-test
 375 { t } [ "ab" R/ ac|[a-z]b/ matches? ] unit-test
 376 { t } [ "ac" R/ ac|\p{Lower}b/ matches? ] unit-test
 377 { t } [ "ac" R/ ac|[a-z]b/ matches? ] unit-test
 378 { t } [ "ac" R/ [a-zA-Z]c|\p{Lower}b/ matches? ] unit-test
 379 { t } [ "ab" R/ [a-zA-Z]c|\p{Lower}b/ matches? ] unit-test
 380 { t } [ "πb" R/ [a-zA-Z]c|\p{Lower}b/ matches? ] unit-test
 381 { f } [ "πc" R/ [a-zA-Z]c|\p{Lower}b/ matches? ] unit-test
 382 { f } [ "Ab" R/ [a-zA-Z]c|\p{Lower}b/ matches? ] unit-test
 383
 384 { t } [ "aaaa" R/ .*a./ matches? ] unit-test
 385
 386 { f } [ "ab" R/ (?~ac|\p{Lower}b)/ matches? ] unit-test
 387 { f } [ "ab" R/ (?~ac|[a-z]b)/ matches? ] unit-test
 388 { f } [ "ac" R/ (?~ac|\p{Lower}b)/ matches? ] unit-test
 389 { f } [ "ac" R/ (?~ac|[a-z]b)/ matches? ] unit-test
 390 { f } [ "ac" R/ (?~[a-zA-Z]c|\p{Lower}b)/ matches? ] unit-test
 391 { f } [ "ab" R/ (?~[a-zA-Z]c|\p{Lower}b)/ matches? ] unit-test
 392 { f } [ "πb" R/ (?~[a-zA-Z]c|\p{Lower}b)/ matches? ] unit-test
 393 { t } [ "πc" R/ (?~[a-zA-Z]c|\p{Lower}b)/ matches? ] unit-test
 394 { t } [ "Ab" R/ (?~[a-zA-Z]c|\p{Lower}b)/ matches? ] unit-test
 395
 396 ! DFA is compiled when needed, or when literal
 397 { regexp-initial-word } [ "foo" <regexp> dfa>> ] unit-test
 398 { f } [ R/ foo/ dfa>> \ regexp-initial-word = ] unit-test
 399
 400 { t } [ "a" R/ ^a/ matches? ] unit-test
 401 { f } [ "\na" R/ ^a/ matches? ] unit-test
 402 { f } [ "\r\na" R/ ^a/ matches? ] unit-test
 403 { f } [ "\ra" R/ ^a/ matches? ] unit-test
 404
 405 { 1 } [ "a" R/ ^a/ count-matches ] unit-test
 406 { 0 } [ "\na" R/ ^a/ count-matches ] unit-test
 407 { 0 } [ "\r\na" R/ ^a/ count-matches ] unit-test
 408 { 0 } [ "\ra" R/ ^a/ count-matches ] unit-test
 409
 410 { t } [ "a" R/ a$/ matches? ] unit-test
 411 { f } [ "a\n" R/ a$/ matches? ] unit-test
 412 { f } [ "a\r" R/ a$/ matches? ] unit-test
 413 { f } [ "a\r\n" R/ a$/ matches? ] unit-test
 414
 415 { 1 } [ "a" R/ a$/ count-matches ] unit-test
 416 { 0 } [ "a\n" R/ a$/ count-matches ] unit-test
 417 { 0 } [ "a\r" R/ a$/ count-matches ] unit-test
 418 { 0 } [ "a\r\n" R/ a$/ count-matches ] unit-test
 419
 420 { t } [ "a" R/ a$|b$/ matches? ] unit-test
 421 { t } [ "b" R/ a$|b$/ matches? ] unit-test
 422 { f } [ "ab" R/ a$|b$/ matches? ] unit-test
 423 { t } [ "ba" R/ ba$|b$/ matches? ] unit-test
 424
 425 { t } [ "a" R/ \Aa/ matches? ] unit-test
 426 { f } [ "\na" R/ \Aaa/ matches? ] unit-test
 427 { f } [ "\r\na" R/ \Aa/ matches? ] unit-test
 428 { f } [ "\ra" R/ \Aa/ matches? ] unit-test
 429
 430 { t } [ "a" R/ \Aa/m matches? ] unit-test
 431 { f } [ "\na" R/ \Aaa/m matches? ] unit-test
 432 { f } [ "\r\na" R/ \Aa/m matches? ] unit-test
 433 { f } [ "\ra" R/ \Aa/m matches? ] unit-test
 434 { 0 } [ "\ra" R/ \Aa/m count-matches ] unit-test
 435
 436 { f } [ "\r\n\n\n\nam" R/ ^am/m matches? ] unit-test
 437 { 1 } [ "\r\n\n\n\nam" R/ ^am/m count-matches ] unit-test
 438
 439 { t } [ "a" R/ \Aa\z/m matches? ] unit-test
 440 { f } [ "a\n" R/ \Aa\z/m matches? ] unit-test
 441
 442 { f } [ "a\r\n" R/ \Aa\Z/m matches? ] unit-test
 443 { f } [ "a\n" R/ \Aa\Z/m matches? ] unit-test
 444 { 1 } [ "a\r\n" R/ \Aa\Z/m count-matches ] unit-test
 445 { 1 } [ "a\n" R/ \Aa\Z/m count-matches ] unit-test
 446
 447 { t } [ "a" R/ \Aa\Z/m matches? ] unit-test
 448 { f } [ "\na" R/ \Aaa\Z/m matches? ] unit-test
 449 { f } [ "\r\na" R/ \Aa\Z/m matches? ] unit-test
 450 { f } [ "\ra" R/ \Aa\Z/m matches? ] unit-test
 451
 452 { 1 } [ "a" R/ \Aa\Z/m count-matches ] unit-test
 453 { 0 } [ "\na" R/ \Aaa\Z/m count-matches ] unit-test
 454 { 0 } [ "\r\na" R/ \Aa\Z/m count-matches ] unit-test
 455 { 0 } [ "\ra" R/ \Aa\Z/m count-matches ] unit-test
 456
 457 { t } [ "a" R/ ^a/m matches? ] unit-test
 458 { f } [ "\na" R/ ^a/m matches? ] unit-test
 459 { 1 } [ "\na" R/ ^a/m count-matches ] unit-test
 460 { 1 } [ "\r\na" R/ ^a/m count-matches ] unit-test
 461 { 1 } [ "\ra" R/ ^a/m count-matches ] unit-test
 462
 463 { t } [ "a" R/ a$/m matches? ] unit-test
 464 { f } [ "a\n" R/ a$/m matches? ] unit-test
 465 { 1 } [ "a\n" R/ a$/m count-matches ] unit-test
 466 { 1 } [ "a\r" R/ a$/m count-matches ] unit-test
 467 { 1 } [ "a\r\n" R/ a$/m count-matches ] unit-test
 468
 469 { f } [ "foobxr" "foo\\z" <regexp> first-match ] unit-test
 470 { 3 } [ "foo" "foo\\z" <regexp> first-match length ] unit-test
 471
 472 { t } [ "a foo b" R/ foo/ re-contains? ] unit-test
 473 { f } [ "a bar b" R/ foo/ re-contains? ] unit-test
 474 { t } [ "foo" R/ foo/ re-contains? ] unit-test
 475
 476 { { "foo" "fxx" "fab" } } [ "fab fxx foo" R/ f../r all-matching-subseqs ] unit-test
 477
 478 { t } [ "foo" "\\bfoo\\b" <regexp> re-contains? ] unit-test
 479 { t } [ "afoob" "\\Bfoo\\B" <regexp> re-contains? ] unit-test
 480 { f } [ "afoob" "\\bfoo\\b" <regexp> re-contains? ] unit-test
 481 { f } [ "foo" "\\Bfoo\\B" <regexp> re-contains? ] unit-test
 482
 483 { 3 } [ "foo bar" "foo\\b" <regexp> first-match length ] unit-test
 484 { f } [ "fooxbar" "foo\\b" <regexp> re-contains? ] unit-test
 485 { t } [ "foo" "foo\\b" <regexp> re-contains? ] unit-test
 486 { t } [ "foo bar" "foo\\b bar" <regexp> matches? ] unit-test
 487 { f } [ "fooxbar" "foo\\bxbar" <regexp> matches? ] unit-test
 488 { f } [ "foo" "foo\\bbar" <regexp> matches? ] unit-test
 489
 490 { f } [ "foo bar" "foo\\B" <regexp> re-contains? ] unit-test
 491 { 3 } [ "fooxbar" "foo\\B" <regexp> first-match length ] unit-test
 492 { f } [ "foo" "foo\\B" <regexp> re-contains? ] unit-test
 493 { f } [ "foo bar" "foo\\B bar" <regexp> matches? ] unit-test
 494 { t } [ "fooxbar" "foo\\Bxbar" <regexp> matches? ] unit-test
 495 { f } [ "foo" "foo\\Bbar" <regexp> matches? ] unit-test
 496
 497 { t } [ "ab" "a(?=b*)" <regexp> re-contains? ] unit-test
 498 { t } [ "abbbbbc" "a(?=b*c)" <regexp> re-contains? ] unit-test
 499 { f } [ "abbbbb" "a(?=b*c)" <regexp> re-contains? ] unit-test
 500 { t } [ "ab" "a(?=b*)" <regexp> re-contains? ] unit-test
 501
 502 { "az" } [ "baz" "(?<=b)(az)" <regexp> first-match >string ] unit-test
 503 { f } [ "chaz" "(?<=b)(az)" <regexp> re-contains? ] unit-test
 504 { "a" } [ "cbaz" "(?<=b*)a" <regexp> first-match >string ] unit-test
 505 { f } [ "baz" "a(?<=b)" <regexp> re-contains? ] unit-test
 506
 507 { f } [ "baz" "(?<!b)a" <regexp> re-contains? ] unit-test
 508 { t } [ "caz" "(?<!b)a" <regexp> re-contains? ] unit-test
 509
 510 { "abcd" } [ "abcdefg" "a(?=bcdefg)bcd" <regexp> first-match >string ] unit-test
 511 { t } [ "abcdefg" "a(?#bcdefg)bcd" <regexp> re-contains? ] unit-test
 512 { t } [ "abcdefg" "a(?:bcdefg)" <regexp> matches? ] unit-test
 513
 514 { 3 } [ "caba" "(?<=b)a" <regexp> first-match from>> ] unit-test
 515
 516 { t } [ "\ra" R/ .^a/ms matches? ] unit-test
 517 { f } [ "\ra" R/ .^a/mds matches? ] unit-test
 518 { t } [ "\na" R/ .^a/ms matches? ] unit-test
 519 { t } [ "\na" R/ .^a/mds matches? ] unit-test
 520
 521 { t } [ "a\r" R/ a$./ms matches? ] unit-test
 522 { f } [ "a\r" R/ a$./mds matches? ] unit-test
 523 { t } [ "a\n" R/ a$./ms matches? ] unit-test
 524 { t } [ "a\n" R/ a$./mds matches? ] unit-test
 525
 526 ! Unicode categories
 527 { t } [ "a" R/ \p{L}/ matches? ] unit-test
 528 { t } [ "A" R/ \p{L}/ matches? ] unit-test
 529 { f } [ " " R/ \p{L}/ matches? ] unit-test
 530 { f } [ "a" R/ \P{L}/ matches? ] unit-test
 531 { f } [ "A" R/ \P{L}/ matches? ] unit-test
 532 { t } [ " " R/ \P{L}/ matches? ] unit-test
 533
 534 { t } [ "a" R/ \p{Ll}/ matches? ] unit-test
 535 { f } [ "A" R/ \p{Ll}/ matches? ] unit-test
 536 { f } [ " " R/ \p{Ll}/ matches? ] unit-test
 537 { f } [ "a" R/ \P{Ll}/ matches? ] unit-test
 538 { t } [ "A" R/ \P{Ll}/ matches? ] unit-test
 539 { t } [ " " R/ \P{Ll}/ matches? ] unit-test
 540
 541 { t } [ "a" R/ \p{script=Latin}/ matches? ] unit-test
 542 { f } [ " " R/ \p{script=Latin}/ matches? ] unit-test
 543 { f } [ "a" R/ \P{script=Latin}/ matches? ] unit-test
 544 { t } [ " " R/ \P{script=Latin}/ matches? ] unit-test
 545
 546 ! These should be case-insensitive
 547 { f } [ " " R/ \p{l}/ matches? ] unit-test
 548 { f } [ "a" R/ \P{l}/ matches? ] unit-test
 549 { f } [ "a" R/ \P{ll}/ matches? ] unit-test
 550 { t } [ " " R/ \P{LL}/ matches? ] unit-test
 551 { f } [ "a" R/ \P{sCriPt = latin}/ matches? ] unit-test
 552 { t } [ " " R/ \P{SCRIPT = laTIn}/ matches? ] unit-test
 553
 554 ! Logical operators
 555 { t } [ "a" R/ [\p{script=latin}\p{lower}]/ matches? ] unit-test
 556 { t } [ "π" R/ [\p{script=latin}\p{lower}]/ matches? ] unit-test
 557 { t } [ "A" R/ [\p{script=latin}\p{lower}]/ matches? ] unit-test
 558 { f } [ "3" R/ [\p{script=latin}\p{lower}]/ matches? ] unit-test
 559
 560 { t } [ "a" R/ [\p{script=latin}||\p{lower}]/ matches? ] unit-test
 561 { t } [ "π" R/ [\p{script=latin}||\p{lower}]/ matches? ] unit-test
 562 { t } [ "A" R/ [\p{script=latin}||\p{lower}]/ matches? ] unit-test
 563 { f } [ "3" R/ [\p{script=latin}||\p{lower}]/ matches? ] unit-test
 564
 565 { t } [ "a" R/ [\p{script=latin}&&\p{lower}]/ matches? ] unit-test
 566 { f } [ "π" R/ [\p{script=latin}&&\p{lower}]/ matches? ] unit-test
 567 { f } [ "A" R/ [\p{script=latin}&&\p{lower}]/ matches? ] unit-test
 568 { f } [ "3" R/ [\p{script=latin}&&\p{lower}]/ matches? ] unit-test
 569
 570 { f } [ "a" R/ [\p{script=latin}~~\p{lower}]/ matches? ] unit-test
 571 { t } [ "π" R/ [\p{script=latin}~~\p{lower}]/ matches? ] unit-test
 572 { t } [ "A" R/ [\p{script=latin}~~\p{lower}]/ matches? ] unit-test
 573 { f } [ "3" R/ [\p{script=latin}~~\p{lower}]/ matches? ] unit-test
 574
 575 { f } [ "a" R/ [\p{script=latin}--\p{lower}]/ matches? ] unit-test
 576 { f } [ "π" R/ [\p{script=latin}--\p{lower}]/ matches? ] unit-test
 577 { t } [ "A" R/ [\p{script=latin}--\p{lower}]/ matches? ] unit-test
 578 { f } [ "3" R/ [\p{script=latin}--\p{lower}]/ matches? ] unit-test
 579
 580 { t } [ " " R/ \P{alpha}/ matches? ] unit-test
 581 { f } [ "" R/ \P{alpha}/ matches? ] unit-test
 582 { f } [ "a " R/ \P{alpha}/ matches? ] unit-test
 583 { f } [ "a" R/ \P{alpha}/ matches? ] unit-test