|
| 1 | +import unittest, codecs |
| 2 | + |
| 3 | +#RELEASE remove |
| 4 | +if __name__ == '__main__': |
| 5 | + import os, sys |
| 6 | + os.chdir(os.path.split(os.path.abspath(__file__))[0]) |
| 7 | + sys.path.insert(0, os.path.abspath(os.path.join(os.pardir, "src"))) |
| 8 | + |
| 9 | +from inputstream import HTMLInputStream |
| 10 | +#END RELEASE |
| 11 | + |
| 12 | +#RELEASE add |
| 13 | +#import html5lib |
| 14 | +#from html5lib.inputstream import HTMLInputStream |
| 15 | +#END RELEASE |
| 16 | + |
| 17 | +class HTMLInputStreamTest(unittest.TestCase): |
| 18 | + |
| 19 | + def test_char_ascii(self): |
| 20 | + stream = HTMLInputStream("'") |
| 21 | + self.assertEquals(stream.charEncoding, 'ascii') |
| 22 | + self.assertEquals(stream.char(), "'") |
| 23 | + |
| 24 | + def test_char_null(self): |
| 25 | + stream = HTMLInputStream("\x00") |
| 26 | + self.assertEquals(stream.char(), u'\ufffd') |
| 27 | + |
| 28 | + def test_char_utf8(self): |
| 29 | + stream = HTMLInputStream(u'\u2018'.encode('utf-8')) |
| 30 | + self.assertEquals(stream.charEncoding, 'utf-8') |
| 31 | + self.assertEquals(stream.char(), u'\u2018') |
| 32 | + |
| 33 | + def test_char_win1252(self): |
| 34 | + stream = HTMLInputStream(u'\u2018'.encode('windows-1252')) |
| 35 | + self.assertEquals(stream.charEncoding, 'windows-1252') |
| 36 | + self.assertEquals(stream.char(), u'\u2018') |
| 37 | + |
| 38 | + def test_bom(self): |
| 39 | + stream = HTMLInputStream(codecs.BOM_UTF8 + "'") |
| 40 | + self.assertEquals(stream.charEncoding, 'utf-8') |
| 41 | + self.assertEquals(stream.char(), "'") |
| 42 | + |
| 43 | + def test_utf_16(self): |
| 44 | + stream = HTMLInputStream((' '*1025).encode('utf-16')) |
| 45 | + self.assert_(stream.charEncoding in ['utf-16-le','utf-16-be']) |
| 46 | + self.assertEquals(len(stream.charsUntil(' ',True)),1025) |
| 47 | + |
| 48 | + def test_newlines(self): |
| 49 | + stream = HTMLInputStream(codecs.BOM_UTF8 + "a\nbb\r\nccc\rdddd") |
| 50 | + self.assertEquals(stream.tell, 0) |
| 51 | + self.assertEquals(stream.charsUntil('x'),u"a\nbb\nccc\ndddd") |
| 52 | + self.assertEquals(stream.tell, 14) |
| 53 | + self.assertEquals(stream.position(), (4,5)) |
| 54 | + self.assertEquals(stream.newLines, [0,1,4,8]) |
| 55 | + |
| 56 | +def buildTestSuite(): |
| 57 | + return unittest.defaultTestLoader.loadTestsFromName(__name__) |
| 58 | + |
| 59 | +def main(): |
| 60 | + buildTestSuite() |
| 61 | + unittest.main() |
| 62 | + |
| 63 | +if __name__ == '__main__': |
| 64 | + main() |
0 commit comments