def test_sorting(self):
" Test the various sorting APIs "
german = """Sonntag Montag Dienstag Januar Februar März Fuße Fluße Flusse flusse fluße flüße flüsse""".split()
german_good = (
"""Dienstag Februar flusse Flusse fluße Fluße flüsse flüße Fuße Januar März Montag Sonntag""".split()
)
french = """dimanche lundi mardi janvier février mars déjà Meme deja même dejà bpef bœg Boef Mémé bœf boef bnef pêche pèché pêché pêche pêché""".split()
french_good = """bnef boef Boef bœf bœg bpef deja dejà déjà dimanche février janvier lundi mardi mars Meme Mémé même pèché pêche pêche pêché pêché""".split() # noqa
# Test corner cases
sort_key = icu.sort_key
s = "\U0001f431"
self.ae(
sort_key(s),
sort_key(s.encode(sys.getdefaultencoding())),
"UTF-8 encoded object not correctly decoded to generate sort key",
)
self.ae(s.encode("utf-16"), s.encode("utf-16"), "Undecodable bytestring not returned as itself")
self.ae(b"", sort_key(None))
self.ae(0, icu.strcmp(None, b""))
self.ae(0, icu.strcmp(s, s.encode(sys.getdefaultencoding())))
# Test locales
with make_collation_func("dsk", "de", func="sort_key") as dsk:
self.ae(german_good, sorted(german, key=dsk))
with make_collation_func("dcmp", "de", template="_strcmp_template") as dcmp:
for x in german:
for y in german:
self.ae(cmp(dsk(x), dsk(y)), dcmp(x, y))
with make_collation_func("fsk", "fr", func="sort_key") as fsk:
self.ae(french_good, sorted(french, key=fsk))
with make_collation_func("fcmp", "fr", template="_strcmp_template") as fcmp:
for x in french:
for y in french:
self.ae(cmp(fsk(x), fsk(y)), fcmp(x, y))
with make_collation_func("ssk", "es", func="sort_key") as ssk:
self.assertNotEqual(ssk("peña"), ssk("pena"))
with make_collation_func("scmp", "es", template="_strcmp_template") as scmp:
self.assertNotEqual(0, scmp("pena", "peña"))
for k, v in {"pèché": "peche", "flüße": "Flusse", "Štepánek": "ŠtepaneK"}.iteritems():
self.ae(0, icu.primary_strcmp(k, v))
# Test different types of collation
self.ae(icu.primary_sort_key("Aä"), icu.primary_sort_key("aa"))
self.assertLess(icu.numeric_sort_key("something 2"), icu.numeric_sort_key("something 11"))
self.assertLess(icu.case_sensitive_sort_key("A"), icu.case_sensitive_sort_key("a"))
self.ae(0, icu.strcmp("a", "A"))
self.ae(cmp("a", "A"), icu.case_sensitive_strcmp("a", "A"))
self.ae(0, icu.primary_strcmp("ä", "A"))
def test_sorting(self):
' Test the various sorting APIs '
german = '''Sonntag Montag Dienstag Januar Februar März Fuße Fluße Flusse flusse fluße flüße flüsse'''.split()
german_good = '''Dienstag Februar flusse Flusse fluße Fluße flüsse flüße Fuße Januar März Montag Sonntag'''.split()
french = '''dimanche lundi mardi janvier février mars déjà Meme deja même dejà bpef bœg Boef Mémé bœf boef bnef pêche pèché pêché pêche pêché'''.split()
french_good = '''bnef boef Boef bœf bœg bpef deja dejà déjà dimanche février janvier lundi mardi mars Meme Mémé même pèché pêche pêche pêché pêché'''.split() # noqa
# Test corner cases
sort_key = icu.sort_key
s = '\U0001f431'
self.ae(sort_key(s), sort_key(s.encode(sys.getdefaultencoding())), 'UTF-8 encoded object not correctly decoded to generate sort key')
self.ae(s.encode('utf-16'), s.encode('utf-16'), 'Undecodable bytestring not returned as itself')
self.ae(b'', sort_key(None))
self.ae(0, icu.strcmp(None, b''))
self.ae(0, icu.strcmp(s, s.encode(sys.getdefaultencoding())))
# Test locales
with make_collation_func('dsk', 'de', func='sort_key') as dsk:
self.ae(german_good, sorted(german, key=dsk))
with make_collation_func('dcmp', 'de', template='_strcmp_template') as dcmp:
for x in german:
for y in german:
self.ae(cmp(dsk(x), dsk(y)), dcmp(x, y))
with make_collation_func('fsk', 'fr', func='sort_key') as fsk:
self.ae(french_good, sorted(french, key=fsk))
with make_collation_func('fcmp', 'fr', template='_strcmp_template') as fcmp:
for x in french:
for y in french:
self.ae(cmp(fsk(x), fsk(y)), fcmp(x, y))
with make_collation_func('ssk', 'es', func='sort_key') as ssk:
self.assertNotEqual(ssk('peña'), ssk('pena'))
with make_collation_func('scmp', 'es', template='_strcmp_template') as scmp:
self.assertNotEqual(0, scmp('pena', 'peña'))
for k, v in {u'pèché': u'peche', u'flüße':u'Flusse', u'Štepánek':u'ŠtepaneK'}.iteritems():
self.ae(0, icu.primary_strcmp(k, v))
# Test different types of collation
self.ae(icu.primary_sort_key('Aä'), icu.primary_sort_key('aa'))
self.assertLess(icu.numeric_sort_key('something 2'), icu.numeric_sort_key('something 11'))
self.assertLess(icu.case_sensitive_sort_key('A'), icu.case_sensitive_sort_key('a'))
self.ae(0, icu.strcmp('a', 'A'))
self.ae(cmp('a', 'A'), icu.case_sensitive_strcmp('a', 'A'))
self.ae(0, icu.primary_strcmp('ä', 'A'))
def test_sorting(self):
' Test the various sorting APIs '
german = '''Sonntag Montag Dienstag Januar Februar März Fuße Fluße Flusse flusse fluße flüße flüsse'''.split()
german_good = '''Dienstag Februar flusse Flusse fluße Fluße flüsse flüße Fuße Januar März Montag Sonntag'''.split()
french = '''dimanche lundi mardi janvier février mars déjà Meme deja même dejà bpef bœg Boef Mémé bœf boef bnef pêche pèché pêché pêche pêché'''.split()
french_good = '''bnef boef Boef bœf bœg bpef deja dejà déjà dimanche février janvier lundi mardi mars Meme Mémé même pèché pêche pêche pêché pêché'''.split() # noqa
# Test corner cases
sort_key = icu.sort_key
s = '\U0001f431'
self.ae(sort_key(s), sort_key(s.encode(sys.getdefaultencoding())), 'UTF-8 encoded object not correctly decoded to generate sort key')
self.ae(s.encode('utf-16'), s.encode('utf-16'), 'Undecodable bytestring not returned as itself')
self.ae(b'', sort_key(None))
self.ae(0, icu.strcmp(None, b''))
self.ae(0, icu.strcmp(s, s.encode(sys.getdefaultencoding())))
# Test locales
with make_collation_func('dsk', 'de', func='sort_key') as dsk:
self.ae(german_good, sorted(german, key=dsk))
with make_collation_func('dcmp', 'de', template='_strcmp_template') as dcmp:
for x in german:
for y in german:
self.ae(cmp(dsk(x), dsk(y)), dcmp(x, y))
with make_collation_func('fsk', 'fr', func='sort_key') as fsk:
self.ae(french_good, sorted(french, key=fsk))
with make_collation_func('fcmp', 'fr', template='_strcmp_template') as fcmp:
for x in french:
for y in french:
self.ae(cmp(fsk(x), fsk(y)), fcmp(x, y))
with make_collation_func('ssk', 'es', func='sort_key') as ssk:
self.assertNotEqual(ssk('peña'), ssk('pena'))
with make_collation_func('scmp', 'es', template='_strcmp_template') as scmp:
self.assertNotEqual(0, scmp('pena', 'peña'))
for k, v in iteritems({u'pèché': u'peche', u'flüße':u'Flusse', u'Štepánek':u'ŠtepaneK'}):
self.ae(0, icu.primary_strcmp(k, v))
# Test different types of collation
self.ae(icu.primary_sort_key('Aä'), icu.primary_sort_key('aa'))
self.assertLess(icu.numeric_sort_key('something 2'), icu.numeric_sort_key('something 11'))
self.assertLess(icu.case_sensitive_sort_key('A'), icu.case_sensitive_sort_key('a'))
self.ae(0, icu.strcmp('a', 'A'))
self.ae(cmp('a', 'A'), icu.case_sensitive_strcmp('a', 'A'))
self.ae(0, icu.primary_strcmp('ä', 'A'))