def test_from_data_frame(): not_data_frame = 'test_str' not_int = 'test_str' params = [[not_data_frame, -1], [DataFrame(), not_int]] for df, index in params: with pytest.raises(AssertionError): Document.from_data_frame(df, index) df = read_excel(example_excel_file) index = 0 doc = Document.from_data_frame(df, index) assert doc.index == index assert sorted(doc.fields.keys()) == sorted(df.columns) assert str(doc.fields) == str(df.to_dict('records')[index]) assert doc.analyzed_sentences == dict()
def data_frame_to_document_list(data_frame): documents = [] for i in range(len(data_frame)): d = Document.from_data_frame(data_frame=data_frame, index=i) documents.append(d) return documents