def test_DocxParser(self):
docx = DocxParser()
docx.parse('files/Test.docx')
text = [
'i', 'test', 'poop', 'test', 'anim', 'test', 'anim', 'googl',
'link'
]
assert docx.get_processed_stems() == text
def test_get_processed_stems(self):
abstract = DocxParser()
abstract.parse('files/Test.docx')
text = [
'i', 'test', 'poop', 'test', 'anim', 'test', 'anim', 'googl',
'link'
]
assert abstract.get_processed_stems() == text
def test_get_stems(self):
abstract = DocxParser()
abstract.parse('files/Test.docx')
text = [
'i', 'am', 'test', 'here', 'poops,', 'test', 'animals,', 'test',
'anim', 'something.', 'googl', 'link'
]
assert abstract.get_stems() == text
def docx_test():
docx_parser = DocxParser()
docx_parser.parse(r'D:\Test22.docx')
print(docx_parser.get_processed_stems())
def docx():
docx_parser = DocxParser()
docx_parser.parse(r'files/Test2.docx')
print('docx parser', docx_parser.get_processed_stems(),
len(docx_parser.get_processed_stems()))
def test_get_processed_text(self):
abstract = DocxParser()
abstract.parse('files/Test.docx')
text = "I test poops Test animals tests animal google link"
assert abstract.get_processed_text() == text
def test_get_text(self):
abstract = DocxParser()
abstract.parse('files/Test.docx')
text = "I am test here poops, Test animals, tests animal something. google link"
assert abstract.get_text() == text