Lib/test/regex_tests.py

   1 # Regex test suite and benchmark suite v1.5a2
   2 # Due to the use of r"aw" strings, this file will
   3 # only work with Python 1.5 or higher.
   4
   5 # The 3 possible outcomes for each pattern
   6 [SUCCEED, FAIL, SYNTAX_ERROR] = range(3)
   7
   8 # Benchmark suite (needs expansion)
   9 #
  10 # The benchmark suite does not test correctness, just speed.  The
  11 # first element of each tuple is the regex pattern; the second is a
  12 # string to match it against.  The benchmarking code will embed the
  13 # second string inside several sizes of padding, to test how regex
  14 # matching performs on large strings.
  15
  16 benchmarks = [
  17         ('Python', 'Python'),                     # Simple text literal
  18         ('.*Python', 'Python'),                   # Bad text literal
  19         ('.*Python.*', 'Python'),                 # Worse text literal
  20         ('.*\\(Python\\)', 'Python'),             # Bad text literal with grouping
  21
  22         ('(Python\\|Perl\\|Tcl', 'Perl'),          # Alternation
  23         ('\\(Python\\|Perl\\|Tcl\\)', 'Perl'),     # Grouped alternation
  24         ('\\(Python\\)\\1', 'PythonPython'),       # Backreference
  25 #       ('\\([0a-z][a-z]*,\\)+', 'a5,b7,c9,'),     # Disable the fastmap optimization
  26         ('\\([a-z][a-z0-9]*,\\)+', 'a5,b7,c9,')    # A few sets
  27 ]
  28
  29 # Test suite (for verifying correctness)
  30 #
  31 # The test suite is a list of 5- or 3-tuples.  The 5 parts of a
  32 # complete tuple are:
  33 # element 0: a string containing the pattern
  34 #         1: the string to match against the pattern
  35 #         2: the expected result (SUCCEED, FAIL, SYNTAX_ERROR)
  36 #         3: a string that will be eval()'ed to produce a test string.
  37 #            This is an arbitrary Python expression; the available
  38 #            variables are "found" (the whole match), and "g1", "g2", ...
  39 #            up to "g10" contain the contents of each group, or the
  40 #            string 'None' if the group wasn't given a value.
  41 #         4: The expected result of evaluating the expression.
  42 #            If the two don't match, an error is reported.
  43 #
  44 # If the regex isn't expected to work, the latter two elements can be omitted.
  45
  46 tests = [
  47 ('abc', 'abc', SUCCEED,
  48  'found', 'abc'),
  49 ('abc', 'xbc', FAIL),
  50 ('abc', 'axc', FAIL),
  51 ('abc', 'abx', FAIL),
  52 ('abc', 'xabcy', SUCCEED,
  53  'found', 'abc'),
  54 ('abc', 'ababc', SUCCEED,
  55  'found', 'abc'),
  56 ('ab*c', 'abc', SUCCEED,
  57  'found', 'abc'),
  58 ('ab*bc', 'abc', SUCCEED,
  59  'found', 'abc'),
  60 ('ab*bc', 'abbc', SUCCEED,
  61  'found', 'abbc'),
  62 ('ab*bc', 'abbbbc', SUCCEED,
  63  'found', 'abbbbc'),
  64 ('ab+bc', 'abbc', SUCCEED,
  65  'found', 'abbc'),
  66 ('ab+bc', 'abc', FAIL),
  67 ('ab+bc', 'abq', FAIL),
  68 ('ab+bc', 'abbbbc', SUCCEED,
  69  'found', 'abbbbc'),
  70 ('ab?bc', 'abbc', SUCCEED,
  71  'found', 'abbc'),
  72 ('ab?bc', 'abc', SUCCEED,
  73  'found', 'abc'),
  74 ('ab?bc', 'abbbbc', FAIL),
  75 ('ab?c', 'abc', SUCCEED,
  76  'found', 'abc'),
  77 ('^abc$', 'abc', SUCCEED,
  78  'found', 'abc'),
  79 ('^abc$', 'abcc', FAIL),
  80 ('^abc', 'abcc', SUCCEED,
  81  'found', 'abc'),
  82 ('^abc$', 'aabc', FAIL),
  83 ('abc$', 'aabc', SUCCEED,
  84  'found', 'abc'),
  85 ('^', 'abc', SUCCEED,
  86  'found+"-"', '-'),
  87 ('$', 'abc', SUCCEED,
  88  'found+"-"', '-'),
  89 ('a.c', 'abc', SUCCEED,
  90  'found', 'abc'),
  91 ('a.c', 'axc', SUCCEED,
  92  'found', 'axc'),
  93 ('a.*c', 'axyzc', SUCCEED,
  94  'found', 'axyzc'),
  95 ('a.*c', 'axyzd', FAIL),
  96 ('a[bc]d', 'abc', FAIL),
  97 ('a[bc]d', 'abd', SUCCEED,
  98  'found', 'abd'),
  99 ('a[b-d]e', 'abd', FAIL),
 100 ('a[b-d]e', 'ace', SUCCEED,
 101  'found', 'ace'),
 102 ('a[b-d]', 'aac', SUCCEED,
 103  'found', 'ac'),
 104 ('a[-b]', 'a-', SUCCEED,
 105  'found', 'a-'),
 106 ('a[b-]', 'a-', SUCCEED,
 107  'found', 'a-'),
 108 ('a[]b', '-', SYNTAX_ERROR),
 109 ('a[', '-', SYNTAX_ERROR),
 110 ('a\\', '-', SYNTAX_ERROR),
 111 ('abc\\)', '-', SYNTAX_ERROR),
 112 ('\\(abc', '-', SYNTAX_ERROR),
 113 ('a]', 'a]', SUCCEED,
 114  'found', 'a]'),
 115 ('a[]]b', 'a]b', SUCCEED,
 116  'found', 'a]b'),
 117 ('a[^bc]d', 'aed', SUCCEED,
 118  'found', 'aed'),
 119 ('a[^bc]d', 'abd', FAIL),
 120 ('a[^-b]c', 'adc', SUCCEED,
 121  'found', 'adc'),
 122 ('a[^-b]c', 'a-c', FAIL),
 123 ('a[^]b]c', 'a]c', FAIL),
 124 ('a[^]b]c', 'adc', SUCCEED,
 125  'found', 'adc'),
 126 ('\\ba\\b', 'a-', SUCCEED,
 127  '"-"', '-'),
 128 ('\\ba\\b', '-a', SUCCEED,
 129  '"-"', '-'),
 130 ('\\ba\\b', '-a-', SUCCEED,
 131  '"-"', '-'),
 132 ('\\by\\b', 'xy', FAIL),
 133 ('\\by\\b', 'yz', FAIL),
 134 ('\\by\\b', 'xyz', FAIL),
 135 ('ab\\|cd', 'abc', SUCCEED,
 136  'found', 'ab'),
 137 ('ab\\|cd', 'abcd', SUCCEED,
 138  'found', 'ab'),
 139 ('\\(\\)ef', 'def', SUCCEED,
 140  'found+"-"+g1', 'ef-'),
 141 ('$b', 'b', FAIL),
 142 ('a(b', 'a(b', SUCCEED,
 143  'found+"-"+g1', 'a(b-None'),
 144 ('a(*b', 'ab', SUCCEED,
 145  'found', 'ab'),
 146 ('a(*b', 'a((b', SUCCEED,
 147  'found', 'a((b'),
 148 ('a\\\\b', 'a\\b', SUCCEED,
 149  'found', 'a\\b'),
 150 ('\\(\\(a\\)\\)', 'abc', SUCCEED,
 151  'found+"-"+g1+"-"+g2', 'a-a-a'),
 152 ('\\(a\\)b\\(c\\)', 'abc', SUCCEED,
 153  'found+"-"+g1+"-"+g2', 'abc-a-c'),
 154 ('a+b+c', 'aabbabc', SUCCEED,
 155  'found', 'abc'),
 156 ('\\(a+\\|b\\)*', 'ab', SUCCEED,
 157  'found+"-"+g1', 'ab-b'),
 158 ('\\(a+\\|b\\)+', 'ab', SUCCEED,
 159  'found+"-"+g1', 'ab-b'),
 160 ('\\(a+\\|b\\)?', 'ab', SUCCEED,
 161  'found+"-"+g1', 'a-a'),
 162 ('\\)\\(', '-', SYNTAX_ERROR),
 163 ('[^ab]*', 'cde', SUCCEED,
 164  'found', 'cde'),
 165 ('abc', '', FAIL),
 166 ('a*', '', SUCCEED,
 167  'found', ''),
 168 ('a\\|b\\|c\\|d\\|e', 'e', SUCCEED,
 169  'found', 'e'),
 170 ('\\(a\\|b\\|c\\|d\\|e\\)f', 'ef', SUCCEED,
 171  'found+"-"+g1', 'ef-e'),
 172 ('abcd*efg', 'abcdefg', SUCCEED,
 173  'found', 'abcdefg'),
 174 ('ab*', 'xabyabbbz', SUCCEED,
 175  'found', 'ab'),
 176 ('ab*', 'xayabbbz', SUCCEED,
 177  'found', 'a'),
 178 ('\\(ab\\|cd\\)e', 'abcde', SUCCEED,
 179  'found+"-"+g1', 'cde-cd'),
 180 ('[abhgefdc]ij', 'hij', SUCCEED,
 181  'found', 'hij'),
 182 ('^\\(ab\\|cd\\)e', 'abcde', FAIL,
 183  'xg1y', 'xy'),
 184 ('\\(abc\\|\\)ef', 'abcdef', SUCCEED,
 185  'found+"-"+g1', 'ef-'),
 186 ('\\(a\\|b\\)c*d', 'abcd', SUCCEED,
 187  'found+"-"+g1', 'bcd-b'),
 188 ('\\(ab\\|ab*\\)bc', 'abc', SUCCEED,
 189  'found+"-"+g1', 'abc-a'),
 190 ('a\\([bc]*\\)c*', 'abc', SUCCEED,
 191  'found+"-"+g1', 'abc-bc'),
 192 ('a\\([bc]*\\)\\(c*d\\)', 'abcd', SUCCEED,
 193  'found+"-"+g1+"-"+g2', 'abcd-bc-d'),
 194 ('a\\([bc]+\\)\\(c*d\\)', 'abcd', SUCCEED,
 195  'found+"-"+g1+"-"+g2', 'abcd-bc-d'),
 196 ('a\\([bc]*\\)\\(c+d\\)', 'abcd', SUCCEED,
 197  'found+"-"+g1+"-"+g2', 'abcd-b-cd'),
 198 ('a[bcd]*dcdcde', 'adcdcde', SUCCEED,
 199  'found', 'adcdcde'),
 200 ('a[bcd]+dcdcde', 'adcdcde', FAIL),
 201 ('\\(ab\\|a\\)b*c', 'abc', SUCCEED,
 202  'found+"-"+g1', 'abc-ab'),
 203 ('\\(\\(a\\)\\(b\\)c\\)\\(d\\)', 'abcd', SUCCEED,
 204  'g1+"-"+g2+"-"+g3+"-"+g4', 'abc-a-b-d'),
 205 ('[a-zA-Z_][a-zA-Z0-9_]*', 'alpha', SUCCEED,
 206  'found', 'alpha'),
 207 ('^a\\(bc+\\|b[eh]\\)g\\|.h$', 'abh', SUCCEED,
 208  'found+"-"+g1', 'bh-None'),
 209 ('\\(bc+d$\\|ef*g.\\|h?i\\(j\\|k\\)\\)', 'effgz', SUCCEED,
 210  'found+"-"+g1+"-"+g2', 'effgz-effgz-None'),
 211 ('\\(bc+d$\\|ef*g.\\|h?i\\(j\\|k\\)\\)', 'ij', SUCCEED,
 212  'found+"-"+g1+"-"+g2', 'ij-ij-j'),
 213 ('\\(bc+d$\\|ef*g.\\|h?i\\(j\\|k\\)\\)', 'effg', FAIL),
 214 ('\\(bc+d$\\|ef*g.\\|h?i\\(j\\|k\\)\\)', 'bcdd', FAIL),
 215 ('\\(bc+d$\\|ef*g.\\|h?i\\(j\\|k\\)\\)', 'reffgz', SUCCEED,
 216  'found+"-"+g1+"-"+g2', 'effgz-effgz-None'),
 217 ('\\(\\(\\(\\(\\(\\(\\(\\(\\(a\\)\\)\\)\\)\\)\\)\\)\\)\\)', 'a', SUCCEED,
 218  'found', 'a'),
 219 ('multiple words of text', 'uh-uh', FAIL),
 220 ('multiple words', 'multiple words, yeah', SUCCEED,
 221  'found', 'multiple words'),
 222 ('\\(.*\\)c\\(.*\\)', 'abcde', SUCCEED,
 223  'found+"-"+g1+"-"+g2', 'abcde-ab-de'),
 224 ('(\\(.*\\), \\(.*\\))', '(a, b)', SUCCEED,
 225  'g2+"-"+g1', 'b-a'),
 226 ('[k]', 'ab', FAIL),
 227 ('a[-]?c', 'ac', SUCCEED,
 228  'found', 'ac'),
 229 ('\\(abc\\)\\1', 'abcabc', SUCCEED,
 230  'g1', 'abc'),
 231 ('\\([a-c]*\\)\\1', 'abcabc', SUCCEED,
 232  'g1', 'abc'),
 233 ('^\\(.+\\)?B', 'AB', SUCCEED,
 234  'g1', 'A'),
 235 ('\\(a+\\).\\1$', 'aaaaa', SUCCEED,
 236  'found+"-"+g1', 'aaaaa-aa'),
 237 ('^\\(a+\\).\\1$', 'aaaa', FAIL),
 238 ('\\(abc\\)\\1', 'abcabc', SUCCEED,
 239  'found+"-"+g1', 'abcabc-abc'),
 240 ('\\([a-c]+\\)\\1', 'abcabc', SUCCEED,
 241  'found+"-"+g1', 'abcabc-abc'),
 242 ('\\(a\\)\\1', 'aa', SUCCEED,
 243  'found+"-"+g1', 'aa-a'),
 244 ('\\(a+\\)\\1', 'aa', SUCCEED,
 245  'found+"-"+g1', 'aa-a'),
 246 ('\\(a+\\)+\\1', 'aa', SUCCEED,
 247  'found+"-"+g1', 'aa-a'),
 248 ('\\(a\\).+\\1', 'aba', SUCCEED,
 249  'found+"-"+g1', 'aba-a'),
 250 ('\\(a\\)ba*\\1', 'aba', SUCCEED,
 251  'found+"-"+g1', 'aba-a'),
 252 ('\\(aa\\|a\\)a\\1$', 'aaa', SUCCEED,
 253  'found+"-"+g1', 'aaa-a'),
 254 ('\\(a\\|aa\\)a\\1$', 'aaa', SUCCEED,
 255  'found+"-"+g1', 'aaa-a'),
 256 ('\\(a+\\)a\\1$', 'aaa', SUCCEED,
 257  'found+"-"+g1', 'aaa-a'),
 258 ('\\([abc]*\\)\\1', 'abcabc', SUCCEED,
 259  'found+"-"+g1', 'abcabc-abc'),
 260 ('\\(a\\)\\(b\\)c\\|ab', 'ab', SUCCEED,
 261  'found+"-"+g1+"-"+g2', 'ab-None-None'),
 262 ('\\(a\\)+x', 'aaax', SUCCEED,
 263  'found+"-"+g1', 'aaax-a'),
 264 ('\\([ac]\\)+x', 'aacx', SUCCEED,
 265  'found+"-"+g1', 'aacx-c'),
 266 ('\\([^/]*/\\)*sub1/', 'd:msgs/tdir/sub1/trial/away.cpp', SUCCEED,
 267  'found+"-"+g1', 'd:msgs/tdir/sub1/-tdir/'),
 268 ('\\([^.]*\\)\\.\\([^:]*\\):[T ]+\\(.*\\)', 'track1.title:TBlah blah blah', SUCCEED,
 269  'found+"-"+g1+"-"+g2+"-"+g3', 'track1.title:TBlah blah blah-track1-title-Blah blah blah'),
 270 ('\\([^N]*N\\)+', 'abNNxyzN', SUCCEED,
 271  'found+"-"+g1', 'abNNxyzN-xyzN'),
 272 ('\\([^N]*N\\)+', 'abNNxyz', SUCCEED,
 273  'found+"-"+g1', 'abNN-N'),
 274 ('\\([abc]*\\)x', 'abcx', SUCCEED,
 275  'found+"-"+g1', 'abcx-abc'),
 276 ('\\([abc]*\\)x', 'abc', FAIL),
 277 ('\\([xyz]*\\)x', 'abcx', SUCCEED,
 278  'found+"-"+g1', 'x-'),
 279 ('\\(a\\)+b\\|aac', 'aac', SUCCEED,
 280  'found+"-"+g1', 'aac-None'),
 281 ('\<a', 'a', SUCCEED, 'found', 'a'),
 282 ('\<a', '!', FAIL),
 283 ('a\<b', 'ab', FAIL),
 284 ('a\>', 'ab', FAIL),
 285 ('a\>', 'a!', SUCCEED, 'found', 'a'),
 286 ('a\>', 'a', SUCCEED, 'found', 'a'),
 287 ]