| 1 | #!/usr/bin/env python2
|
| 2 | """
|
| 3 | split.test.py: Tests for split.py
|
| 4 | """
|
| 5 |
|
| 6 | import unittest
|
| 7 |
|
| 8 | from osh import split # module under test
|
| 9 |
|
| 10 |
|
| 11 | def _RunSplitCases(test, sp, cases):
|
| 12 | for expected_parts, s, allow_escape in cases:
|
| 13 | spans = sp.Split(s, allow_escape)
|
| 14 | if 0:
|
| 15 | print('%r: %s' % (s, spans))
|
| 16 | else:
|
| 17 | # Verbose for debugging
|
| 18 | print(repr(s))
|
| 19 | for span in spans:
|
| 20 | print(' %s %s' % span)
|
| 21 |
|
| 22 | parts = split._SpansToParts(s, spans)
|
| 23 | print('PARTS %s' % parts)
|
| 24 |
|
| 25 | test.assertEqual(expected_parts, parts,
|
| 26 | '%r: %s != %s' % (s, expected_parts, parts))
|
| 27 |
|
| 28 |
|
| 29 | class SplitTest(unittest.TestCase):
|
| 30 |
|
| 31 | def testSpansToParts(self):
|
| 32 | sp = split.IfsSplitter(split.DEFAULT_IFS, '')
|
| 33 |
|
| 34 | s = 'one\\ two'
|
| 35 | spans = sp.Split(s, False)
|
| 36 | print(spans)
|
| 37 |
|
| 38 | parts = split._SpansToParts(s, spans)
|
| 39 | self.assertEqual(['one\\', 'two'], parts)
|
| 40 |
|
| 41 | spans = sp.Split(s, True) # allow_escape
|
| 42 | parts = split._SpansToParts(s, spans)
|
| 43 | self.assertEqual(['one two'], parts)
|
| 44 |
|
| 45 | # NOTE: Only read builtin supports max_results
|
| 46 | return
|
| 47 |
|
| 48 | parts = split._SpansToParts(s, spans, max_results=1)
|
| 49 | self.assertEqual(['one\\ two'], parts)
|
| 50 |
|
| 51 | print(spans)
|
| 52 |
|
| 53 | parts = split._SpansToParts(s, spans, max_results=1)
|
| 54 | self.assertEqual(['one two'], parts)
|
| 55 |
|
| 56 | def testTrailingWhitespaceBug(self):
|
| 57 | # Bug: these differed
|
| 58 | CASES = [
|
| 59 | (['x y'], r' x\ y', True),
|
| 60 | (['ab '], r' ab\ ', True),
|
| 61 | (['ab '], r' ab\ ', True),
|
| 62 | ]
|
| 63 | sp = split.IfsSplitter(split.DEFAULT_IFS, '')
|
| 64 | _RunSplitCases(self, sp, CASES)
|
| 65 |
|
| 66 | def testDefaultIfs(self):
|
| 67 | CASES = [
|
| 68 | ([], '', True),
|
| 69 | (['a'], 'a', True),
|
| 70 | (['a'], ' a ', True),
|
| 71 | (['ab'], '\tab\n', True),
|
| 72 | (['a', 'b'], 'a b\n', True),
|
| 73 | (['a b'], r'a\ b', True),
|
| 74 | (['a\\', 'b'], r'a\ b', False),
|
| 75 | ([r'\*.sh'], r'\\*.sh', True),
|
| 76 | (['Aa', 'b', ' a b'], 'Aa b \\ a\\ b', True),
|
| 77 | ]
|
| 78 |
|
| 79 | sp = split.IfsSplitter(split.DEFAULT_IFS, '')
|
| 80 | _RunSplitCases(self, sp, CASES)
|
| 81 |
|
| 82 | self.assertEqual(r'a\ _b', sp.Escape('a _b'))
|
| 83 |
|
| 84 | def testMixedIfs(self):
|
| 85 | CASES = [
|
| 86 | ([], '', True),
|
| 87 | (['a', 'b'], 'a_b', True),
|
| 88 | (['a', 'b'], ' a b ', True),
|
| 89 | (['a', 'b'], 'a _ b', True),
|
| 90 | (['a', 'b'], ' a _ b ', True),
|
| 91 | (['a', '', 'b'], 'a _ _ b', True),
|
| 92 | (['a', '', 'b'], 'a __ b', True),
|
| 93 | (['a', '', '', 'b'], 'a _ _ _ b', True),
|
| 94 | (['a'], ' a _ ', True),
|
| 95 |
|
| 96 | # NOTES:
|
| 97 | # - This cases REQUIRES ignoring leading whitespace. The state machine
|
| 98 | # can't handle it. Contrast with the case above.
|
| 99 | # - We get three spans with index 1 because of the initial rule to
|
| 100 | # ignore whitespace, and then EMIT_EMPTY. Seems harmless for now?
|
| 101 | (['', 'a'], ' _ a _ ', True),
|
| 102 |
|
| 103 | # Backslash escape
|
| 104 | (['a b'], r'a\ b', True),
|
| 105 | (['a\\', 'b'], r'a\ b', False),
|
| 106 | ]
|
| 107 |
|
| 108 | # IFS='_ '
|
| 109 | sp = split.IfsSplitter(' ', '_')
|
| 110 | _RunSplitCases(self, sp, CASES)
|
| 111 |
|
| 112 | self.assertEqual('a\ \_b', sp.Escape('a _b'))
|
| 113 |
|
| 114 | def testWhitespaceOnly(self):
|
| 115 | CASES = [
|
| 116 | ([], '', True),
|
| 117 | ([], '\t', True),
|
| 118 | (['a'], 'a\t', True),
|
| 119 | (['a', 'b'], '\t\ta\tb\t', True),
|
| 120 |
|
| 121 | # Backslash escape
|
| 122 | (['a\tb'], 'a\\\tb', True),
|
| 123 | (['a\\', 'b'], 'a\\\tb', False),
|
| 124 | ]
|
| 125 |
|
| 126 | # IFS='_ '
|
| 127 | sp = split.IfsSplitter('\t', '')
|
| 128 | _RunSplitCases(self, sp, CASES)
|
| 129 |
|
| 130 | self.assertEqual('a b', sp.Escape('a b'))
|
| 131 | self.assertEqual('a\\\tb', sp.Escape('a\tb'))
|
| 132 |
|
| 133 | def testOtherOnly(self):
|
| 134 | CASES = [
|
| 135 | ([], '', True),
|
| 136 | ([''], '_', True),
|
| 137 | (['a'], 'a_', True),
|
| 138 | (['', '', 'a', 'b'], '__a_b_', True),
|
| 139 |
|
| 140 | # Backslash escape
|
| 141 | (['a_b'], r'a\_b', True),
|
| 142 | (['a\\', 'b'], r'a\_b', False),
|
| 143 | ]
|
| 144 |
|
| 145 | # IFS='_ '
|
| 146 | sp = split.IfsSplitter('', '_')
|
| 147 | _RunSplitCases(self, sp, CASES)
|
| 148 |
|
| 149 | def testTwoOther(self):
|
| 150 | CASES = [
|
| 151 | (['a', '', 'b', '', '', 'c', 'd'], 'a__b---c_d', True),
|
| 152 |
|
| 153 | # Backslash escape
|
| 154 | (['a_-b'], r'a\_\-b', True),
|
| 155 | (['a\\', '\\', 'b'], r'a\_\-b', False),
|
| 156 | ]
|
| 157 |
|
| 158 | # IFS='_ '
|
| 159 | sp = split.IfsSplitter('', '_-')
|
| 160 | _RunSplitCases(self, sp, CASES)
|
| 161 |
|
| 162 |
|
| 163 | if __name__ == '__main__':
|
| 164 | unittest.main()
|