def test_position2(self): stream = HTMLInputStream("abc\nd") self.assertEquals(stream.position(), (1, 0)) self.assertEquals(stream.char(), u"a") self.assertEquals(stream.position(), (1, 1)) self.assertEquals(stream.char(), u"b") self.assertEquals(stream.position(), (1, 2)) self.assertEquals(stream.char(), u"c") self.assertEquals(stream.position(), (1, 3)) self.assertEquals(stream.char(), u"\n") self.assertEquals(stream.position(), (2, 0)) self.assertEquals(stream.char(), u"d") self.assertEquals(stream.position(), (2, 1))
def test_position(self): stream = HTMLInputStream(codecs.BOM_UTF8 + "a\nbb\nccc\nddd") self.assertEquals(stream.position(), (1, 0)) self.assertEquals(stream.charsUntil('c'),u"a\nbb\n") self.assertEquals(stream.position(), (3, 0)) stream.unget("a\nbb\n") self.assertEquals(stream.position(), (1, 0)) self.assertEquals(stream.charsUntil('c'),u"a\nbb\n") self.assertEquals(stream.position(), (3, 0)) stream.unget("\n") self.assertEquals(stream.char(), u"\n") self.assertEquals(stream.position(), (3, 0)) self.assertEquals(stream.charsUntil('e'),u"ccc\nddd") self.assertEquals(stream.position(), (4, 3))
def test_position(self): stream = HTMLInputStream(codecs.BOM_UTF8 + "a\nbb\nccc\nddde\nf\ngh") self.assertEquals(stream.position(), (1, 0)) self.assertEquals(stream.charsUntil('c'), u"a\nbb\n") self.assertEquals(stream.position(), (3, 0)) stream.unget(u"\n") self.assertEquals(stream.position(), (2, 2)) self.assertEquals(stream.charsUntil('c'), u"\n") self.assertEquals(stream.position(), (3, 0)) stream.unget(u"\n") self.assertEquals(stream.position(), (2, 2)) self.assertEquals(stream.char(), u"\n") self.assertEquals(stream.position(), (3, 0)) self.assertEquals(stream.charsUntil('e'), u"ccc\nddd") self.assertEquals(stream.position(), (4, 3)) self.assertEquals(stream.charsUntil('h'), u"e\nf\ng") self.assertEquals(stream.position(), (6, 1))
def test_bom(self): stream = HTMLInputStream(codecs.BOM_UTF8 + "'") self.assertEquals(stream.charEncoding[0], u'utf-8') self.assertEquals(stream.char(), u"'")
def test_char_win1252(self): stream = HTMLInputStream(u"\xa9\xf1\u2019".encode(u'windows-1252')) self.assertEquals(stream.charEncoding[0], u'windows-1252') self.assertEquals(stream.char(), u"\xa9") self.assertEquals(stream.char(), u"\xf1") self.assertEquals(stream.char(), u"\u2019")
def test_char_utf8(self): stream = HTMLInputStream(u'\u2018'.encode(u'utf-8'), encoding=u'utf-8') self.assertEquals(stream.charEncoding[0], u'utf-8') self.assertEquals(stream.char(), u'\u2018')
def test_char_ascii(self): stream = HTMLInputStream("'", encoding=u'ascii') self.assertEquals(stream.charEncoding[0], u'ascii') self.assertEquals(stream.char(), u"'")
def test_char_utf8(): stream = HTMLInputStream('\u2018'.encode('utf-8'), encoding='utf-8') assert stream.charEncoding[0].name == 'utf-8' assert stream.char() == '\u2018'
def test_char_win1252(self): stream = HTMLInputStream('\xa9\xf1\u2019'.encode('windows-1252')) self.assertEqual(stream.charEncoding[0], 'windows-1252') self.assertEqual(stream.char(), '\xa9') self.assertEqual(stream.char(), '\xf1') self.assertEqual(stream.char(), '\u2019')
def test_char_utf8(self): stream = HTMLInputStream('\u2018'.encode('utf-8'), encoding='utf-8') self.assertEqual(stream.charEncoding[0].name, 'utf-8') self.assertEqual(stream.char(), '\u2018')
def test_char_null(self): stream = HTMLInputStream("\x00") self.assertEquals(stream.char(), u'\ufffd')
def test_char_win1252(): stream = HTMLInputStream("\xa9\xf1\u2019".encode('windows-1252')) assert stream.charEncoding[0].name == 'windows-1252' assert stream.char() == "\xa9" assert stream.char() == "\xf1" assert stream.char() == "\u2019"
def test_char_ascii(self): stream = HTMLInputStream(b"'", encoding='ascii') self.assertEqual(stream.charEncoding[0], 'ascii') self.assertEqual(stream.char(), "'")
def test_bom(self): stream = HTMLInputStream(codecs.BOM_UTF8 + b"'") self.assertEqual(stream.charEncoding[0].name, 'utf-8') self.assertEqual(stream.char(), "'")
def test_bom(self): stream = HTMLInputStream(codecs.BOM_UTF8 + b"'") self.assertEqual(stream.charEncoding[0], 'utf-8') self.assertEqual(stream.char(), "'")
def test_char_ascii(self): stream = HTMLInputStream(b"'", encoding='ascii') self.assertEqual(stream.charEncoding[0].name, 'windows-1252') self.assertEqual(stream.char(), "'")
def test_char_utf8(self): stream = HTMLInputStream('\u2018'.encode('utf-8'), encoding='utf-8') self.assertEqual(stream.charEncoding[0], 'utf-8') self.assertEqual(stream.char(), '\u2018')
def test_char_win1252(self): stream = HTMLInputStream("\xa9\xf1\u2019".encode('windows-1252')) self.assertEqual(stream.charEncoding[0].name, 'windows-1252') self.assertEqual(stream.char(), "\xa9") self.assertEqual(stream.char(), "\xf1") self.assertEqual(stream.char(), "\u2019")
def test_bom(): stream = HTMLInputStream(codecs.BOM_UTF8 + b"'") assert stream.charEncoding[0].name == 'utf-8' assert stream.char() == "'"
def test_char_win1252(self): stream = HTMLInputStream("\xa9\xf1\u2019".encode('windows-1252')) self.assertEqual(stream.charEncoding[0], 'windows-1252') self.assertEqual(stream.char(), "\xa9") self.assertEqual(stream.char(), "\xf1") self.assertEqual(stream.char(), "\u2019")
def test_char_ascii(): stream = HTMLInputStream(b"'", encoding='ascii') assert stream.charEncoding[0].name == 'windows-1252' assert stream.char() == "'"