Python TEIDocument Examples

Example #1

File: test_document.py Project: kcl-ddh/bacalhau

 def setUp(self):
     self.filepath = "tests/corpus/a.xml"
     self.doc = TEIDocument(
         self.filepath,
         nltk.tokenize.regexp.WordPunctTokenizer(),
         nltk.corpus.stopwords.words("english"),
         '//tei:body/tei:div[@type = "dummy"]',
     )

Example #2

File: test_document.py Project: kcl-ddh/bacalhau

class TestDocument(unittest.TestCase):
    def setUp(self):
        self.filepath = "tests/corpus/a.xml"
        self.doc = TEIDocument(
            self.filepath,
            nltk.tokenize.regexp.WordPunctTokenizer(),
            nltk.corpus.stopwords.words("english"),
            '//tei:body/tei:div[@type = "dummy"]',
        )

    def test_get_text_count(self):
        self.assertEqual(2, self.doc.get_text_count())

    def test_get_texts(self):
        texts = self.doc.get_texts()
        self.assertEqual(2, len(texts))

    def test_get_term_data(self):
        term_data = self.doc.get_term_data()
        self.assertIsNotNone(term_data)