def test_arg_type(self):
self.assertRaises(TypeError, pcre2nfa, None)
a = pcre2nfa('')
self.assertEqual(NFA, type(a))
test = pcre2nfa('^abc')
self.assertTrue(test.match("abc"))
self.assertFalse(test.match("dabc"))
def test_arg_type(self):
self.assertRaises(TypeError, pcre2nfa, None)
a = pcre2nfa('')
self.assertEqual(NFA, type(a))
test = pcre2nfa('^abc')
self.assertTrue(test.match("abc"))
self.assertFalse(test.match("dabc"))
def test_dotall(self):
a = pcre2nfa('/abc.def/s')
self.assertTrue(a.match('abcxdef'))
self.assertTrue(a.match('abc\ndef'))
self.assertFalse(a.match('abcdef'))
a = pcre2nfa('/abc.def/')
self.assertTrue(a.match('abcxdef'))
self.assertFalse(a.match('abc\ndef'))
def test_dotall(self):
a = pcre2nfa('/abc.def/s')
self.assertTrue(a.match('abcxdef'))
self.assertTrue(a.match('abc\ndef'))
self.assertFalse(a.match('abcdef'))
a = pcre2nfa('/abc.def/')
self.assertTrue(a.match('abcxdef'))
self.assertFalse(a.match('abc\ndef'))
def test_upto(self):
a = pcre2nfa('/a{1,2}b/i')
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('aab'))
self.assertFalse(a.match('aaa'))
a = pcre2nfa('/a{4,5}b/i')
self.assertTrue(a.match('aaaab'))
self.assertTrue(a.match('aaaaab'))
self.assertFalse(a.match('aaaaaa'))
def test_upto(self):
a = pcre2nfa('/a{1,2}b/i')
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('aab'))
self.assertFalse(a.match('aaa'))
a = pcre2nfa('/a{4,5}b/i')
self.assertTrue(a.match('aaaab'))
self.assertTrue(a.match('aaaaab'))
self.assertFalse(a.match('aaaaaa'))
def test_circ(self):
a = pcre2nfa('^a')
self.assertTrue(a.match('a'))
self.assertTrue(a.match('ab'))
self.assertFalse(a.match('ba'))
a = pcre2nfa('^\s*abc')
self.assertTrue(a.match('abc'))
self.assertTrue(a.match(' abc'))
self.assertTrue(a.match(' abcxyz'))
self.assertFalse(a.match('ababc'))
def test_circ(self):
a = pcre2nfa('^a')
self.assertTrue(a.match('a'))
self.assertTrue(a.match('ab'))
self.assertFalse(a.match('ba'))
a = pcre2nfa('^\s*abc')
self.assertTrue(a.match('abc'))
self.assertTrue(a.match(' abc'))
self.assertTrue(a.match(' abcxyz'))
self.assertFalse(a.match('ababc'))
def test_case_insensitive(self):
a = pcre2nfa('/abcde/i')
self.assertTrue(a.match('ABCDEFGHI'))
self.assertTrue(a.match('abcdefghi'))
self.assertTrue(a.match('aBcDeFgHi'))
self.assertFalse(a.match('afBgChDiE'))
a = pcre2nfa('/abcde/')
self.assertFalse(a.match('ABCDEFGHI'))
self.assertTrue(a.match('abcdefghi'))
self.assertFalse(a.match('aBcDeFgHi'))
self.assertFalse(a.match('afBgChDiE'))
def test_case_insensitive(self):
a = pcre2nfa('/abcde/i')
self.assertTrue(a.match('ABCDEFGHI'))
self.assertTrue(a.match('abcdefghi'))
self.assertTrue(a.match('aBcDeFgHi'))
self.assertFalse(a.match('afBgChDiE'))
a = pcre2nfa('/abcde/')
self.assertFalse(a.match('ABCDEFGHI'))
self.assertTrue(a.match('abcdefghi'))
self.assertFalse(a.match('aBcDeFgHi'))
self.assertFalse(a.match('afBgChDiE'))
def test_notwhitespace(self):
a = pcre2nfa('\S')
self.assertFalse(a.match('\x0c'))
self.assertFalse(a.match(' '))
self.assertFalse(a.match('\n'))
self.assertFalse(a.match('\t'))
self.assertTrue(a.match('a'))
a = pcre2nfa('a\Sb')
self.assertTrue(a.match('aab'))
self.assertFalse(a.match('\t\n'))
self.assertFalse(a.match('a b'))
def test_notwhitespace(self):
a = pcre2nfa('\S')
self.assertFalse(a.match('\x0c'))
self.assertFalse(a.match(' '))
self.assertFalse(a.match('\n'))
self.assertFalse(a.match('\t'))
self.assertTrue(a.match('a'))
a = pcre2nfa('a\Sb')
self.assertTrue(a.match('aab'))
self.assertFalse(a.match('\t\n'))
self.assertFalse(a.match('a b'))
def test_not_class_range(self):
a = pcre2nfa('/a[^a-z]b/')
for c in range(ord('a'), ord('z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(32, 60):
self.assertTrue(a.match('a' + chr(c) + 'b'))
a = pcre2nfa('/a[^a-z]b/i')
for c in range(ord('A'), ord('Z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(ord('a'), ord('z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(32, 57):
self.assertTrue(a.match('a' + chr(c) + 'b'))
def test_notwordchar(self):
a = pcre2nfa('\W')
self.assertTrue(a.match('\x0c'))
self.assertFalse(a.match('A'))
self.assertFalse(a.match('Z'))
self.assertFalse(a.match('_'))
self.assertFalse(a.match('a'))
self.assertFalse(a.match('z'))
self.assertTrue(a.match('\n'))
a = pcre2nfa('a\Wb')
self.assertFalse(a.match('aab'))
self.assertFalse(a.match(' '))
self.assertTrue(a.match('a b'))
def test_notwhitespace(self):
a = pcre2nfa('\S{1,3}')
self.assertFalse(a.match('\x0c'))
self.assertFalse(a.match(' '))
self.assertFalse(a.match('\n'))
self.assertFalse(a.match('\t'))
self.assertTrue(a.match('a'))
a = pcre2nfa('a\S{1,3}b')
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a\tb'))
self.assertTrue(a.match('axxb'))
self.assertTrue(a.match('axyzb'))
self.assertFalse(a.match('attttb'))
def test_notwhitespace(self):
a = pcre2nfa('\S{1,3}')
self.assertFalse(a.match('\x0c'))
self.assertFalse(a.match(' '))
self.assertFalse(a.match('\n'))
self.assertFalse(a.match('\t'))
self.assertTrue(a.match('a'))
a = pcre2nfa('a\S{1,3}b')
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a\tb'))
self.assertTrue(a.match('axxb'))
self.assertTrue(a.match('axyzb'))
self.assertFalse(a.match('attttb'))
def test_notwordchar(self):
a = pcre2nfa('\W')
self.assertTrue(a.match('\x0c'))
self.assertFalse(a.match('A'))
self.assertFalse(a.match('Z'))
self.assertFalse(a.match('_'))
self.assertFalse(a.match('a'))
self.assertFalse(a.match('z'))
self.assertTrue(a.match('\n'))
a = pcre2nfa('a\Wb')
self.assertFalse(a.match('aab'))
self.assertFalse(a.match(' '))
self.assertTrue(a.match('a b'))
def test_not_class_range(self):
a = pcre2nfa('/a[^a-z]b/')
for c in range(ord('a'), ord('z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(32, 60):
self.assertTrue(a.match('a' + chr(c) + 'b'))
a = pcre2nfa('/a[^a-z]b/i')
for c in range(ord('A'), ord('Z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(ord('a'), ord('z')):
self.assertFalse(a.match('a' + chr(c) + 'b'))
for c in range(32, 57):
self.assertTrue(a.match('a' + chr(c) + 'b'))
def test_char_nocase(self):
a = pcre2nfa('/a/i')
self.assertTrue(a.match('a'))
self.assertTrue(a.match('A'))
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('bA'))
self.assertFalse(a.match('b'))
def test_char(self):
a = pcre2nfa('a')
self.assertTrue(a.match('a'))
self.assertFalse(a.match('A'))
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('ba'))
self.assertFalse(a.match('b'))
def test_dotall_n_caseless(self):
a = pcre2nfa('/abc.ef/is')
self.assertTrue(a.match('abcdef'))
self.assertTrue(a.match('ABCDEF'))
self.assertTrue(a.match('aBcDeF'))
self.assertTrue(a.match('abC%ef'))
self.assertTrue(a.match('abc\nef'))
def test_brazero_any(self):
a = pcre2nfa('a(b|c)*d')
self.assertTrue(a.match('abd'))
self.assertTrue(a.match('ad'))
self.assertTrue(a.match('acd'))
self.assertTrue(a.match('abbbbd'))
self.assertFalse(a.match('aed'))
a = pcre2nfa('a(b*|cd)*e')
self.assertTrue(a.match('abe'))
self.assertTrue(a.match('ae'))
self.assertTrue(a.match('acde'))
self.assertTrue(a.match('abcde'))
self.assertTrue(a.match('abbbbe'))
self.assertTrue(a.match('acdbcde'))
self.assertFalse(a.match('abcbde'))
def test_plus(self):
a = pcre2nfa('a0+0')
self.assertFalse(a.match('a0'))
self.assertTrue(a.match('a00'))
self.assertTrue(a.match('a000'))
self.assertTrue(a.match('a0000'))
self.assertFalse(a.match('a01'))
def test_crquery(self):
a = pcre2nfa('a[0-1]?a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a1a'))
self.assertFalse(a.match('a10a'))
self.assertFalse(a.match('a9a'))
def test_any(self):
a = pcre2nfa('a.{2,4}a')
self.assertFalse(a.match('aaa'))
self.assertTrue(a.match('abba'))
self.assertTrue(a.match('ababaa'))
self.assertTrue(a.match('axxxxa'))
self.assertFalse(a.match('axxxxxa'))
def test_star(self):
a = pcre2nfa('a0{2,}(a|b)')
self.assertFalse(a.match('aa'))
self.assertFalse(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00x'))
def test_any(self):
a = pcre2nfa('a.*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('aaa'))
self.assertTrue(a.match('aba'))
self.assertTrue(a.match('acdefa'))
self.assertFalse(a.match('axxxxxx'))
def test_digit(self):
a = pcre2nfa('a\d{2,3}b')
self.assertTrue(a.match('a12b'))
self.assertTrue(a.match('a123b'))
self.assertFalse(a.match('a1b'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a1234b'))
def test_notwhitespace(self):
a = pcre2nfa('a\S+b')
self.assertTrue(a.match('axxb'))
self.assertTrue(a.match('axb'))
self.assertFalse(a.match('a b'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a b'))
def test_crquery(self):
a = pcre2nfa('a[0-1]?a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a1a'))
self.assertFalse(a.match('a10a'))
self.assertFalse(a.match('a9a'))
def test_posstar(self):
a = pcre2nfa('a0*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00b'))
def test_notminplus(self):
a = pcre2nfa('a[^x]+?b')
self.assertTrue(a.match('ayb'))
self.assertTrue(a.match('ayzb'))
self.assertFalse(a.match('axb'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('axxxxxb'))
def test_crrange_range(self):
a = pcre2nfa('^[abc]{2,4}x')
self.assertFalse(a.match('bx'))
self.assertTrue(a.match('acx'))
self.assertTrue(a.match('abcx'))
self.assertTrue(a.match('bbbbx'))
self.assertFalse(a.match('bbbbbx'))
self.assertFalse(a.match('abca'))
a = pcre2nfa('^[abc]{0,2}x')
self.assertTrue(a.match('x'))
self.assertTrue(a.match('bx'))
self.assertTrue(a.match('acx'))
self.assertFalse(a.match('bbbx'))
self.assertFalse(a.match('bbbbx'))
self.assertFalse(a.match('abca'))
def test_posplus(self):
a = pcre2nfa('a0+a')
self.assertFalse(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00b'))
def test_notposplus(self):
a = pcre2nfa('a[^x]+x')
self.assertTrue(a.match('aax'))
self.assertTrue(a.match('abx'))
self.assertTrue(a.match('acdefx'))
self.assertFalse(a.match('axx'))
self.assertFalse(a.match('ax'))
def test_wordchar(self):
a = pcre2nfa(':\w{2}:')
self.assertFalse(a.match(':%%:'))
self.assertTrue(a.match(':aA:'))
self.assertTrue(a.match(':zZ:'))
self.assertTrue(a.match(':_a:'))
self.assertFalse(a.match(':aaa:'))
def test_notposplus(self):
a = pcre2nfa('a[^x]+x')
self.assertTrue(a.match('aax'))
self.assertTrue(a.match('abx'))
self.assertTrue(a.match('acdefx'))
self.assertFalse(a.match('axx'))
self.assertFalse(a.match('ax'))
def test_whitespace(self):
a = pcre2nfa('\s')
self.assertTrue(a.match('\x0c'))
self.assertTrue(a.match(' '))
self.assertTrue(a.match('\n'))
self.assertTrue(a.match('\t'))
self.assertFalse(a.match('a'))
def test_char_nocase(self):
a = pcre2nfa('/a/i')
self.assertTrue(a.match('a'))
self.assertTrue(a.match('A'))
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('bA'))
self.assertFalse(a.match('b'))
def test_digit(self):
a = pcre2nfa('a\d{2,3}b')
self.assertTrue(a.match('a12b'))
self.assertTrue(a.match('a123b'))
self.assertFalse(a.match('a1b'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a1234b'))
def test_brazero_any(self):
a = pcre2nfa('a(b|c)*d')
self.assertTrue(a.match('abd'))
self.assertTrue(a.match('ad'))
self.assertTrue(a.match('acd'))
self.assertTrue(a.match('abbbbd'))
self.assertFalse(a.match('aed'))
a = pcre2nfa('a(b*|cd)*e')
self.assertTrue(a.match('abe'))
self.assertTrue(a.match('ae'))
self.assertTrue(a.match('acde'))
self.assertTrue(a.match('abcde'))
self.assertTrue(a.match('abbbbe'))
self.assertTrue(a.match('acdbcde'))
self.assertFalse(a.match('abcbde'))
def test_notwhitespace(self):
a = pcre2nfa('a\S+b')
self.assertTrue(a.match('axxb'))
self.assertTrue(a.match('axb'))
self.assertFalse(a.match('a b'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('a b'))
def test_star(self):
a = pcre2nfa('a0{2,}(a|b)')
self.assertFalse(a.match('aa'))
self.assertFalse(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00x'))
def test_wordchar(self):
a = pcre2nfa(':\w{2}:')
self.assertFalse(a.match(':%%:'))
self.assertTrue(a.match(':aA:'))
self.assertTrue(a.match(':zZ:'))
self.assertTrue(a.match(':_a:'))
self.assertFalse(a.match(':aaa:'))
def test_posstar(self):
a = pcre2nfa('a0*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00b'))
def test_notminplus(self):
a = pcre2nfa('a[^x]+?b')
self.assertTrue(a.match('ayb'))
self.assertTrue(a.match('ayzb'))
self.assertFalse(a.match('axb'))
self.assertFalse(a.match('ab'))
self.assertFalse(a.match('axxxxxb'))
def test_posplus(self):
a = pcre2nfa('a0+a')
self.assertFalse(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a00a'))
self.assertTrue(a.match('a000a'))
self.assertFalse(a.match('a00b'))
def test_plus(self):
a = pcre2nfa('a0+0')
self.assertFalse(a.match('a0'))
self.assertTrue(a.match('a00'))
self.assertTrue(a.match('a000'))
self.assertTrue(a.match('a0000'))
self.assertFalse(a.match('a01'))
def test_any(self):
a = pcre2nfa('a.*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('aaa'))
self.assertTrue(a.match('aba'))
self.assertTrue(a.match('acdefa'))
self.assertFalse(a.match('axxxxxx'))
def test_crstar(self):
a = pcre2nfa('a[0-1]*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a10a'))
self.assertTrue(a.match('a111000a'))
self.assertFalse(a.match('a0012100a'))
def test_dotall_n_caseless(self):
a = pcre2nfa('/abc.ef/is')
self.assertTrue(a.match('abcdef'))
self.assertTrue(a.match('ABCDEF'))
self.assertTrue(a.match('aBcDeF'))
self.assertTrue(a.match('abC%ef'))
self.assertTrue(a.match('abc\nef'))
def test_char(self):
a = pcre2nfa('a')
self.assertTrue(a.match('a'))
self.assertFalse(a.match('A'))
self.assertTrue(a.match('ab'))
self.assertTrue(a.match('ba'))
self.assertFalse(a.match('b'))
def test_whitespace(self):
a = pcre2nfa('\s')
self.assertTrue(a.match('\x0c'))
self.assertTrue(a.match(' '))
self.assertTrue(a.match('\n'))
self.assertTrue(a.match('\t'))
self.assertFalse(a.match('a'))
def test_crrange_range(self):
a = pcre2nfa('^[abc]{2,4}x')
self.assertFalse(a.match('bx'))
self.assertTrue(a.match('acx'))
self.assertTrue(a.match('abcx'))
self.assertTrue(a.match('bbbbx'))
self.assertFalse(a.match('bbbbbx'))
self.assertFalse(a.match('abca'))
a = pcre2nfa('^[abc]{0,2}x')
self.assertTrue(a.match('x'))
self.assertTrue(a.match('bx'))
self.assertTrue(a.match('acx'))
self.assertFalse(a.match('bbbx'))
self.assertFalse(a.match('bbbbx'))
self.assertFalse(a.match('abca'))
def test_crstar(self):
a = pcre2nfa('a[0-1]*a')
self.assertTrue(a.match('aa'))
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a10a'))
self.assertTrue(a.match('a111000a'))
self.assertFalse(a.match('a0012100a'))
def test_any(self):
a = pcre2nfa('a.{2,4}a')
self.assertFalse(a.match('aaa'))
self.assertTrue(a.match('abba'))
self.assertTrue(a.match('ababaa'))
self.assertTrue(a.match('axxxxa'))
self.assertFalse(a.match('axxxxxa'))
def test_notplus(self):
a = pcre2nfa('a[^x]+a')
self.assertTrue(a.match('aaa'))
self.assertTrue(a.match('aba'))
self.assertTrue(a.match('acdefa'))
self.assertFalse(a.match('axa'))
self.assertFalse(a.match('acdexsda'))
self.assertFalse(a.match('aa'))
def test_notplus(self):
a = pcre2nfa('a[^x]+a')
self.assertTrue(a.match('aaa'))
self.assertTrue(a.match('aba'))
self.assertTrue(a.match('acdefa'))
self.assertFalse(a.match('axa'))
self.assertFalse(a.match('acdexsda'))
self.assertFalse(a.match('aa'))
def test_actual(self):
a = pcre2nfa('/clsid\s*\x3a\s*\x7B?\s*'
'EC5D5118-9FDE-4A3E-84F3-C2B711740E70(\x22)?.*'
'DownloadCertificateExt\(/is')
self.assertTrue(a.match('clsid:EC5D5118-9FDE-4A3E-84F3-C2B711740E70'
'DownloadCertificateExt('))
self.assertTrue(a.match('CLSID : { EC5D5118-9FDE-4A3E-84F3-'
'C2B711740E70xxxxxxDownloadCertificateExt('))
def test_crplus(self):
# OP_CRPOSPLUS in pcre >= 8.34
a = pcre2nfa('a[0-1]+a')
self.assertTrue(a.match('a0a'))
self.assertTrue(a.match('a10a'))
self.assertTrue(a.match('a111000a'))
self.assertFalse(a.match('aa'))
self.assertFalse(a.match('a0012100a'))
