src/external/pcre2-10.32/testdata/testinput11

   1 # This set of tests is for the 16-bit and 32-bit libraries' basic (non-UTF)
   2 # features that are not compatible with the 8-bit library, or which give
   3 # different output in 16-bit or 32-bit mode. The output for the two widths is
   4 # different, so they have separate output files.
   5
   6 #forbid_utf
   7 #newline_default LF ANY ANYCRLF
   8
   9 /[^\x{c4}]/IB
  10
  11 /\x{100}/I
  12
  13 /  (?: [\040\t] |  \(
  14 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  15 \)  )*                          # optional leading comment
  16 (?:    (?:
  17 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  18 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  19 |
  20 " (?:                      # opening quote...
  21 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
  22 |                     #    or
  23 \\ [^\x80-\xff]           #   Escaped something (something != CR)
  24 )* "  # closing quote
  25 )                    # initial word
  26 (?:  (?: [\040\t] |  \(
  27 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  28 \)  )*  \.  (?: [\040\t] |  \(
  29 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  30 \)  )*   (?:
  31 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  32 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  33 |
  34 " (?:                      # opening quote...
  35 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
  36 |                     #    or
  37 \\ [^\x80-\xff]           #   Escaped something (something != CR)
  38 )* "  # closing quote
  39 )  )* # further okay, if led by a period
  40 (?: [\040\t] |  \(
  41 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  42 \)  )*  @  (?: [\040\t] |  \(
  43 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  44 \)  )*    (?:
  45 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  46 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  47 |   \[                         # [
  48 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
  49 \]                        #           ]
  50 )                           # initial subdomain
  51 (?:                                  #
  52 (?: [\040\t] |  \(
  53 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  54 \)  )*  \.                        # if led by a period...
  55 (?: [\040\t] |  \(
  56 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  57 \)  )*   (?:
  58 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  59 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  60 |   \[                         # [
  61 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
  62 \]                        #           ]
  63 )                     #   ...further okay
  64 )*
  65 # address
  66 |                     #  or
  67 (?:
  68 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  69 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  70 |
  71 " (?:                      # opening quote...
  72 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
  73 |                     #    or
  74 \\ [^\x80-\xff]           #   Escaped something (something != CR)
  75 )* "  # closing quote
  76 )             # one word, optionally followed by....
  77 (?:
  78 [^()<>@,;:".\\\[\]\x80-\xff\000-\010\012-\037]  |  # atom and space parts, or...
  79 \(
  80 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  81 \)       |  # comments, or...
  82
  83 " (?:                      # opening quote...
  84 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
  85 |                     #    or
  86 \\ [^\x80-\xff]           #   Escaped something (something != CR)
  87 )* "  # closing quote
  88 # quoted strings
  89 )*
  90 <  (?: [\040\t] |  \(
  91 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  92 \)  )*                     # leading <
  93 (?:  @  (?: [\040\t] |  \(
  94 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
  95 \)  )*    (?:
  96 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
  97 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
  98 |   \[                         # [
  99 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 100 \]                        #           ]
 101 )                           # initial subdomain
 102 (?:                                  #
 103 (?: [\040\t] |  \(
 104 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 105 \)  )*  \.                        # if led by a period...
 106 (?: [\040\t] |  \(
 107 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 108 \)  )*   (?:
 109 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 110 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 111 |   \[                         # [
 112 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 113 \]                        #           ]
 114 )                     #   ...further okay
 115 )*
 116
 117 (?:  (?: [\040\t] |  \(
 118 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 119 \)  )*  ,  (?: [\040\t] |  \(
 120 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 121 \)  )*  @  (?: [\040\t] |  \(
 122 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 123 \)  )*    (?:
 124 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 125 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 126 |   \[                         # [
 127 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 128 \]                        #           ]
 129 )                           # initial subdomain
 130 (?:                                  #
 131 (?: [\040\t] |  \(
 132 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 133 \)  )*  \.                        # if led by a period...
 134 (?: [\040\t] |  \(
 135 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 136 \)  )*   (?:
 137 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 138 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 139 |   \[                         # [
 140 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 141 \]                        #           ]
 142 )                     #   ...further okay
 143 )*
 144 )* # further okay, if led by comma
 145 :                                # closing colon
 146 (?: [\040\t] |  \(
 147 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 148 \)  )*  )? #       optional route
 149 (?:
 150 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 151 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 152 |
 153 " (?:                      # opening quote...
 154 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
 155 |                     #    or
 156 \\ [^\x80-\xff]           #   Escaped something (something != CR)
 157 )* "  # closing quote
 158 )                    # initial word
 159 (?:  (?: [\040\t] |  \(
 160 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 161 \)  )*  \.  (?: [\040\t] |  \(
 162 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 163 \)  )*   (?:
 164 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 165 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 166 |
 167 " (?:                      # opening quote...
 168 [^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
 169 |                     #    or
 170 \\ [^\x80-\xff]           #   Escaped something (something != CR)
 171 )* "  # closing quote
 172 )  )* # further okay, if led by a period
 173 (?: [\040\t] |  \(
 174 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 175 \)  )*  @  (?: [\040\t] |  \(
 176 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 177 \)  )*    (?:
 178 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 179 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 180 |   \[                         # [
 181 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 182 \]                        #           ]
 183 )                           # initial subdomain
 184 (?:                                  #
 185 (?: [\040\t] |  \(
 186 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 187 \)  )*  \.                        # if led by a period...
 188 (?: [\040\t] |  \(
 189 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 190 \)  )*   (?:
 191 [^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
 192 (?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
 193 |   \[                         # [
 194 (?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
 195 \]                        #           ]
 196 )                     #   ...further okay
 197 )*
 198 #       address spec
 199 (?: [\040\t] |  \(
 200 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 201 \)  )*  > #                  trailing >
 202 # name and address
 203 )  (?: [\040\t] |  \(
 204 (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
 205 \)  )*                       # optional trailing comment
 206 /Ix
 207
 208 /[\h]/B
 209     >\x09<
 210
 211 /[\h]+/B
 212     >\x09\x20\xa0<
 213
 214 /[\v]/B
 215
 216 /[^\h]/B
 217
 218 /\h+/I
 219     \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
 220     \x{3001}\x{2fff}\x{200a}\xa0\x{2000}
 221
 222 /[\h\x{dc00}]+/IB
 223     \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
 224     \x{3001}\x{2fff}\x{200a}\xa0\x{2000}
 225
 226 /\H+/I
 227     \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
 228     \x{2000}\x{200a}\x{1fff}\x{200b}
 229     \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
 230     \xa0\x{3000}\x9f\xa1\x{2fff}\x{3001}
 231
 232 /[\H\x{d800}]+/
 233     \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
 234     \x{2000}\x{200a}\x{1fff}\x{200b}
 235     \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
 236     \xa0\x{3000}\x9f\xa1\x{2fff}\x{3001}
 237
 238 /\v+/I
 239     \x{2027}\x{2030}\x{2028}\x{2029}
 240     \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
 241
 242 /[\v\x{dc00}]+/IB
 243     \x{2027}\x{2030}\x{2028}\x{2029}
 244     \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
 245
 246 /\V+/I
 247     \x{2028}\x{2029}\x{2027}\x{2030}
 248     \x85\x0a\x0b\x0c\x0d\x09\x0e\x84\x86
 249
 250 /[\V\x{d800}]+/
 251     \x{2028}\x{2029}\x{2027}\x{2030}
 252     \x85\x0a\x0b\x0c\x0d\x09\x0e\x84\x86
 253
 254 /\R+/I,bsr=unicode
 255     \x{2027}\x{2030}\x{2028}\x{2029}
 256     \x09\x0e\x84\x86\x85\x0a\x0b\x0c\x0d
 257
 258 /\x{d800}\x{d7ff}\x{dc00}\x{dc00}\x{dcff}\x{dd00}/I
 259     \x{d800}\x{d7ff}\x{dc00}\x{dc00}\x{dcff}\x{dd00}
 260
 261 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/B
 262
 263 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/Bi
 264
 265 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/B
 266
 267 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/Bi
 268
 269 /(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF)XX/mark
 270     XX
 271
 272 /(*:0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDE)XX/mark
 273     XX
 274
 275 /\u0100/B,alt_bsux,allow_empty_class,match_unset_backref
 276
 277 /[\u0100-\u0200]/B,alt_bsux,allow_empty_class,match_unset_backref
 278
 279 /\ud800/B,alt_bsux,allow_empty_class,match_unset_backref
 280
 281 /^\x{ffff}+/i
 282     \x{ffff}
 283
 284 /^\x{ffff}?/i
 285     \x{ffff}
 286
 287 /^\x{ffff}*/i
 288     \x{ffff}
 289
 290 /^\x{ffff}{3}/i
 291     \x{ffff}\x{ffff}\x{ffff}
 292
 293 /^\x{ffff}{0,3}/i
 294     \x{ffff}
 295
 296 /[^\x00-a]{12,}[^b-\xff]*/B
 297
 298 /[^\s]*\s* [^\W]+\W+ [^\d]*?\d0 [^\d\w]{4,6}?\w*A/B
 299
 300 /a*[b-\x{200}]?a#a*[b-\x{200}]?b#[a-f]*[g-\x{200}]*#[g-\x{200}]*[a-c]*#[g-\x{200}]*[a-h]*/B
 301
 302 /^[\x{1234}\x{4321}]{2,4}?/
 303     \x{1234}\x{1234}\x{1234}
 304
 305 # Check maximum non-UTF character size for the 16-bit library.
 306
 307 /\x{ffff}/
 308     A\x{ffff}B
 309
 310 /\x{10000}/
 311
 312 /\o{20000}/
 313
 314 # Check maximum character size for the 32-bit library. These will all give
 315 # errors in the 16-bit library.
 316
 317 /\x{110000}/
 318
 319 /\x{7fffffff}/
 320
 321 /\x{80000000}/
 322
 323 /\x{ffffffff}/
 324
 325 /\x{100000000}/
 326
 327 /\o{17777777777}/
 328
 329 /\o{20000000000}/
 330
 331 /\o{37777777777}/
 332
 333 /\o{40000000000}/
 334
 335 /\x{7fffffff}\x{7fffffff}/I
 336
 337 /\x{80000000}\x{80000000}/I
 338
 339 /\x{ffffffff}\x{ffffffff}/I
 340
 341 # Non-UTF characters
 342
 343 /.{2,3}/
 344     \x{400000}\x{400001}\x{400002}\x{400003}
 345
 346 /\x{400000}\x{800000}/IBi
 347
 348 # Check character ranges
 349
 350 /[\H]/IB
 351
 352 /[\V]/IB
 353
 354 /(*THEN:\[A]{65501})/expand
 355
 356 # We can use pcre2test's utf8_input modifier to create wide pattern characters,
 357 # even though this test is run when UTF is not supported.
 358
 359 /ab������z/utf8_input
 360     ab������z
 361     ab\x{7fffffff}z
 362
 363 /abÿý¿¿¿¿¿z/utf8_input
 364     abÿý¿¿¿¿¿z
 365     ab\x{ffffffff}z
 366
 367 /abÿAz/utf8_input
 368     abÿAz
 369     ab\x{80000041}z
 370
 371 # End of testinput11