def pre_process(path):
corpus = orangecontrib.text.Corpus.from_file(path)
p = preprocess.Preprocessor(
transformers=[
preprocess.LowercaseTransformer(),
preprocess.UrlRemover(),
preprocess.HtmlTransformer()
],
tokenizer=preprocess.RegexpTokenizer('\w+'),
normalizer=preprocess.PorterStemmer(),
filters=[
preprocess.StopwordsFilter('english'),
preprocess.RegexpFilter(
'\.|,|:|;|!|\?|\(|\)|\||\+|\'|\"|‘|’|“|”|\'|\’|…|\-|–|—|\$|&|\*|>|<'
)
])
return p(corpus)
def test_preprocess_words(self):
corpus = Corpus.from_file("book-excerpts")
words = [
"House",
"dóctor",
"boy",
"way",
"Rum https://google.com",
"https://google.com",
"<p>abra<b>cadabra</b><p>",
]
pp_list = [
preprocess.LowercaseTransformer(),
preprocess.StripAccentsTransformer(),
preprocess.UrlRemover(),
preprocess.HtmlTransformer(),
]
for p in pp_list:
corpus = p(corpus)
self.assertListEqual(
["house", "doctor", "boy", "way", "rum", "abracadabra"],
_preprocess_words(corpus, words, dummy_callback),
)
words = ["House", "dóctor", "boys", "way", "Rum"]
pp_list = [preprocess.SnowballStemmer()]
for p in pp_list:
corpus = p(corpus)
self.assertListEqual(
["hous", "doctor", "boy", "way", "rum"],
_preprocess_words(corpus, words, dummy_callback),
)
def test_html(self):
transformer = preprocess.HtmlTransformer()
self.assertEqual(transformer._preprocess('<p>abra<b>cadabra</b><p>'),
'abracadabra')