def test_sanitize_sortbysize(self):
wordlist = ["word", "is", "his", "short"]
expect = ["short", "word", "his", "is"]
test = WStext(wordlist, maxlength=11)
answer = test.sanitize_sortbysize(wordlist)
assert answer == expect
wordlist = ["longing", "notshorter"]
expect = ["notshorter", "longing"]
test = WStext(wordlist, 11)
answer = test.sanitize_sortbysize(wordlist)
assert answer == expect
def test_sanitize_nosubwords(self):
wordlist = ["short", "word", "is", "his"]
expect = ["short", "word", "his"]
reject = []
test = WStext(wordlist, maxlength=11)
good, bad = test.sanitize_nosubwords(wordlist, reject)
assert good == expect
assert bad == ["is"]
wordlist = ["short", "notshorter", "longing", "long", "verylongword"]
expect = ["notshorter", "longing", "verylongword"]
reject = []
test = WStext(wordlist)
good, bad = test.sanitize_nosubwords(wordlist, reject)
assert good == expect
assert bad == ["short", "long"]
def test_sanitize_length(self):
wordlist = ["short", "notshorter", "longing", "verylongword"]
expect = ["notshorter", "longing"]
expectreject = ["verylongword", "short"]
test = WStext(wordlist, maxlength=11)
assert test.wordlist == expect
assert test.wordlist_rejected == expectreject
def setUpAll(self):
self.wsstreamio = WSstreamIO()
self.wstext = WStext()
self.wstext.setLanguage('to')
fh = self.wsstreamio.open('unicode.txt')
self.wsstreamio.load(fh)
self.lines = self.wsstreamio.lines
self.originalEncoding = self.wsstreamio.originalEncoding
def test_sanitize_words(self):
wordlist = ["word", "is", "his", "short"]
expect = ["short", "word", "his"]
reject = []
test = WStext(wordlist, maxlength=11)
good, bad = test.sanitize_words(wordlist, maxlength=11)
assert good == expect
assert bad == ["is"]
def test_sanitize_content(self):
wordlist = ["short", "not6shorter", "longing", "verylongword"]
expect = ["longing", "shorter", "not"]
rejected = ["short", "verylongword"]
test = WStext(wordlist, maxlength=11, lingua="en")
for word in expect:
assert word in test.wordlist
for word in rejected:
assert word in test.wordlist_rejected
def setUpAll(self):
self.words = [
'this', 'is', 'my', 'matrix', 'test', 'collection', 'not', 'very',
'large', 'but', 'hoping'
]
self.wstext = WStext(lingua="en")
