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