Martin v. Löwis | b96e0e5 | 2000-09-19 16:35:39 +0000 | [diff] [blame] | 1 | # Tests StringIO and cStringIO |
| 2 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 3 | import unittest |
| 4 | import StringIO |
| 5 | import cStringIO |
| 6 | import types |
Antoine Pitrou | a5016cd | 2011-10-21 22:22:04 +0200 | [diff] [blame] | 7 | import array |
Serhiy Storchaka | 276f1d5 | 2013-02-09 13:47:43 +0200 | [diff] [blame] | 8 | import sys |
Barry Warsaw | 04f357c | 2002-07-23 19:04:11 +0000 | [diff] [blame] | 9 | from test import test_support |
Martin v. Löwis | b96e0e5 | 2000-09-19 16:35:39 +0000 | [diff] [blame] | 10 | |
Jeremy Hylton | cafd495 | 2001-02-09 23:44:22 +0000 | [diff] [blame] | 11 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 12 | class TestGenericStringIO(unittest.TestCase): |
| 13 | # use a class variable MODULE to define which module is being tested |
Jeremy Hylton | cafd495 | 2001-02-09 23:44:22 +0000 | [diff] [blame] | 14 | |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 15 | # Line of data to test as string |
| 16 | _line = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ!' |
| 17 | |
| 18 | # Constructor to use for the test data (._line is passed to this |
| 19 | # constructor) |
| 20 | constructor = str |
| 21 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 22 | def setUp(self): |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 23 | self._line = self.constructor(self._line) |
| 24 | self._lines = self.constructor((self._line + '\n') * 5) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 25 | self._fp = self.MODULE.StringIO(self._lines) |
Fred Drake | 9543833 | 2000-09-28 04:25:33 +0000 | [diff] [blame] | 26 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 27 | def test_reads(self): |
| 28 | eq = self.assertEqual |
Walter Dörwald | bb9c739 | 2004-11-01 17:10:19 +0000 | [diff] [blame] | 29 | self.assertRaises(TypeError, self._fp.seek) |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 30 | eq(self._fp.read(10), self._line[:10]) |
Serhiy Storchaka | 8d7d6bc | 2013-02-13 12:26:58 +0200 | [diff] [blame] | 31 | eq(self._fp.read(0), '') |
| 32 | eq(self._fp.readline(0), '') |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 33 | eq(self._fp.readline(), self._line[10:] + '\n') |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 34 | eq(len(self._fp.readlines(60)), 2) |
Benjamin Peterson | 4147216 | 2009-12-13 17:29:16 +0000 | [diff] [blame] | 35 | self._fp.seek(0) |
| 36 | eq(self._fp.readline(-1), self._line + '\n') |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 37 | |
| 38 | def test_writes(self): |
| 39 | f = self.MODULE.StringIO() |
Walter Dörwald | bb9c739 | 2004-11-01 17:10:19 +0000 | [diff] [blame] | 40 | self.assertRaises(TypeError, f.seek) |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 41 | f.write(self._line[:6]) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 42 | f.seek(3) |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 43 | f.write(self._line[20:26]) |
| 44 | f.write(self._line[52]) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 45 | self.assertEqual(f.getvalue(), 'abcuvwxyz!') |
| 46 | |
| 47 | def test_writelines(self): |
| 48 | f = self.MODULE.StringIO() |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 49 | f.writelines([self._line[0], self._line[1], self._line[2]]) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 50 | f.seek(0) |
| 51 | self.assertEqual(f.getvalue(), 'abc') |
| 52 | |
Michael W. Hudson | 10402a3 | 2005-09-22 09:19:01 +0000 | [diff] [blame] | 53 | def test_writelines_error(self): |
| 54 | def errorGen(): |
| 55 | yield 'a' |
| 56 | raise KeyboardInterrupt() |
| 57 | f = self.MODULE.StringIO() |
| 58 | self.assertRaises(KeyboardInterrupt, f.writelines, errorGen()) |
| 59 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 60 | def test_truncate(self): |
| 61 | eq = self.assertEqual |
| 62 | f = self.MODULE.StringIO() |
| 63 | f.write(self._lines) |
| 64 | f.seek(10) |
| 65 | f.truncate() |
| 66 | eq(f.getvalue(), 'abcdefghij') |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 67 | f.truncate(5) |
| 68 | eq(f.getvalue(), 'abcde') |
Tim Peters | 037b3ee | 2004-08-21 06:55:43 +0000 | [diff] [blame] | 69 | f.write('xyz') |
| 70 | eq(f.getvalue(), 'abcdexyz') |
Martin v. Löwis | cffcc8b | 2006-11-19 10:41:41 +0000 | [diff] [blame] | 71 | self.assertRaises(IOError, f.truncate, -1) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 72 | f.close() |
| 73 | self.assertRaises(ValueError, f.write, 'frobnitz') |
| 74 | |
Raymond Hettinger | 5475f23 | 2003-08-08 12:20:03 +0000 | [diff] [blame] | 75 | def test_closed_flag(self): |
| 76 | f = self.MODULE.StringIO() |
| 77 | self.assertEqual(f.closed, False) |
| 78 | f.close() |
| 79 | self.assertEqual(f.closed, True) |
| 80 | f = self.MODULE.StringIO("abc") |
| 81 | self.assertEqual(f.closed, False) |
| 82 | f.close() |
| 83 | self.assertEqual(f.closed, True) |
| 84 | |
Walter Dörwald | 197e832 | 2006-03-15 22:13:13 +0000 | [diff] [blame] | 85 | def test_isatty(self): |
| 86 | f = self.MODULE.StringIO() |
| 87 | self.assertRaises(TypeError, f.isatty, None) |
| 88 | self.assertEqual(f.isatty(), False) |
| 89 | f.close() |
| 90 | self.assertRaises(ValueError, f.isatty) |
| 91 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 92 | def test_iterator(self): |
| 93 | eq = self.assertEqual |
Benjamin Peterson | 5c8da86 | 2009-06-30 22:57:08 +0000 | [diff] [blame] | 94 | unless = self.assertTrue |
Raymond Hettinger | 352f947 | 2003-04-24 15:50:11 +0000 | [diff] [blame] | 95 | eq(iter(self._fp), self._fp) |
Barry Warsaw | 4565350 | 2001-09-25 21:40:04 +0000 | [diff] [blame] | 96 | # Does this object support the iteration protocol? |
Raymond Hettinger | 352f947 | 2003-04-24 15:50:11 +0000 | [diff] [blame] | 97 | unless(hasattr(self._fp, '__iter__')) |
| 98 | unless(hasattr(self._fp, 'next')) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 99 | i = 0 |
| 100 | for line in self._fp: |
| 101 | eq(line, self._line + '\n') |
| 102 | i += 1 |
| 103 | eq(i, 5) |
Walter Dörwald | 0af5d93 | 2006-03-15 08:23:53 +0000 | [diff] [blame] | 104 | self._fp.close() |
| 105 | self.assertRaises(ValueError, self._fp.next) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 106 | |
Benjamin Peterson | 4fa68ab | 2011-05-26 09:56:41 -0500 | [diff] [blame] | 107 | def test_getvalue(self): |
| 108 | self._fp.close() |
| 109 | self.assertRaises(ValueError, self._fp.getvalue) |
| 110 | |
Serhiy Storchaka | 276f1d5 | 2013-02-09 13:47:43 +0200 | [diff] [blame] | 111 | @test_support.bigmemtest(test_support._2G + 2**26, memuse=2.001) |
| 112 | def test_reads_from_large_stream(self, size): |
| 113 | linesize = 2**26 # 64 MiB |
| 114 | lines = ['x' * (linesize - 1) + '\n'] * (size // linesize) + \ |
| 115 | ['y' * (size % linesize)] |
| 116 | f = self.MODULE.StringIO(''.join(lines)) |
| 117 | for i, expected in enumerate(lines): |
| 118 | line = f.read(len(expected)) |
| 119 | self.assertEqual(len(line), len(expected)) |
| 120 | self.assertEqual(line, expected) |
| 121 | self.assertEqual(f.read(), '') |
| 122 | f.seek(0) |
| 123 | for i, expected in enumerate(lines): |
| 124 | line = f.readline() |
| 125 | self.assertEqual(len(line), len(expected)) |
| 126 | self.assertEqual(line, expected) |
| 127 | self.assertEqual(f.readline(), '') |
| 128 | f.seek(0) |
| 129 | self.assertEqual(f.readlines(), lines) |
| 130 | self.assertEqual(f.readlines(), []) |
| 131 | f.seek(0) |
| 132 | self.assertEqual(f.readlines(size), lines) |
| 133 | self.assertEqual(f.readlines(), []) |
| 134 | |
| 135 | # In worst case cStringIO requires 2 + 1 + 1/2 + 1/2**2 + ... = 4 |
| 136 | # bytes per input character. |
| 137 | @test_support.bigmemtest(test_support._2G, memuse=4) |
| 138 | def test_writes_to_large_stream(self, size): |
| 139 | s = 'x' * 2**26 # 64 MiB |
| 140 | f = self.MODULE.StringIO() |
| 141 | n = size |
| 142 | while n > len(s): |
| 143 | f.write(s) |
| 144 | n -= len(s) |
| 145 | s = None |
| 146 | f.write('x' * n) |
| 147 | self.assertEqual(len(f.getvalue()), size) |
| 148 | |
| 149 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 150 | class TestStringIO(TestGenericStringIO): |
| 151 | MODULE = StringIO |
| 152 | |
Marc-André Lemburg | f853be9 | 2002-01-06 17:15:05 +0000 | [diff] [blame] | 153 | def test_unicode(self): |
| 154 | |
Michael W. Hudson | e1c67d1 | 2002-05-13 09:42:16 +0000 | [diff] [blame] | 155 | if not test_support.have_unicode: return |
| 156 | |
Marc-André Lemburg | f853be9 | 2002-01-06 17:15:05 +0000 | [diff] [blame] | 157 | # The StringIO module also supports concatenating Unicode |
| 158 | # snippets to larger Unicode strings. This is tested by this |
| 159 | # method. Note that cStringIO does not support this extension. |
Tim Peters | e441860 | 2002-02-16 07:34:19 +0000 | [diff] [blame] | 160 | |
Marc-André Lemburg | f853be9 | 2002-01-06 17:15:05 +0000 | [diff] [blame] | 161 | f = self.MODULE.StringIO() |
| 162 | f.write(self._line[:6]) |
| 163 | f.seek(3) |
| 164 | f.write(unicode(self._line[20:26])) |
| 165 | f.write(unicode(self._line[52])) |
| 166 | s = f.getvalue() |
| 167 | self.assertEqual(s, unicode('abcuvwxyz!')) |
| 168 | self.assertEqual(type(s), types.UnicodeType) |
| 169 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 170 | class TestcStringIO(TestGenericStringIO): |
| 171 | MODULE = cStringIO |
| 172 | |
Antoine Pitrou | a5016cd | 2011-10-21 22:22:04 +0200 | [diff] [blame] | 173 | def test_array_support(self): |
Antoine Pitrou | f678e82 | 2011-10-21 22:22:43 +0200 | [diff] [blame] | 174 | # Issue #1730114: cStringIO should accept array objects |
Antoine Pitrou | a5016cd | 2011-10-21 22:22:04 +0200 | [diff] [blame] | 175 | a = array.array('B', [0,1,2]) |
| 176 | f = self.MODULE.StringIO(a) |
| 177 | self.assertEqual(f.getvalue(), '\x00\x01\x02') |
| 178 | |
Antoine Pitrou | 5a77fe9 | 2011-10-22 21:26:01 +0200 | [diff] [blame] | 179 | def test_unicode(self): |
| 180 | |
| 181 | if not test_support.have_unicode: return |
| 182 | |
| 183 | # The cStringIO module converts Unicode strings to character |
| 184 | # strings when writing them to cStringIO objects. |
| 185 | # Check that this works. |
| 186 | |
| 187 | f = self.MODULE.StringIO() |
| 188 | f.write(u'abcde') |
| 189 | s = f.getvalue() |
| 190 | self.assertEqual(s, 'abcde') |
| 191 | self.assertEqual(type(s), str) |
| 192 | |
| 193 | f = self.MODULE.StringIO(u'abcde') |
| 194 | s = f.getvalue() |
| 195 | self.assertEqual(s, 'abcde') |
| 196 | self.assertEqual(type(s), str) |
| 197 | |
| 198 | self.assertRaises(UnicodeEncodeError, self.MODULE.StringIO, u'\xf4') |
| 199 | |
Antoine Pitrou | a5016cd | 2011-10-21 22:22:04 +0200 | [diff] [blame] | 200 | |
Finn Bock | 793ead5 | 2001-12-09 20:06:32 +0000 | [diff] [blame] | 201 | import sys |
| 202 | if sys.platform.startswith('java'): |
| 203 | # Jython doesn't have a buffer object, so we just do a useless |
| 204 | # fake of the buffer tests. |
| 205 | buffer = str |
| 206 | |
Marc-André Lemburg | e47df7a | 2001-09-24 17:34:52 +0000 | [diff] [blame] | 207 | class TestBufferStringIO(TestStringIO): |
| 208 | constructor = buffer |
| 209 | |
| 210 | class TestBuffercStringIO(TestcStringIO): |
| 211 | constructor = buffer |
| 212 | |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 213 | |
| 214 | def test_main(): |
Florent Xicluna | 0762788 | 2010-03-21 01:14:24 +0000 | [diff] [blame] | 215 | test_support.run_unittest(TestStringIO, TestcStringIO) |
| 216 | with test_support.check_py3k_warnings(("buffer.. not supported", |
| 217 | DeprecationWarning)): |
| 218 | test_support.run_unittest(TestBufferStringIO, TestBuffercStringIO) |
Barry Warsaw | 7f8ff47 | 2001-09-22 04:33:47 +0000 | [diff] [blame] | 219 | |
| 220 | if __name__ == '__main__': |
| 221 | test_main() |