def test_to_data_frame():
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test')
test.add_file(file_name="file2", label='file2', content='other file')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.add_dictionary(file_name="dict2.txt", label="dict2", content="test")
test.count()
test.test_option = TestOptions(formula="")
test.save_formula()
test.generate_scores()
test.generate_averages()
assert isinstance(test.to_data_frame(), type(pd.DataFrame()))
def test_generate_files_raw_counts_tables():
test = ContentAnalysisModel()
test.test_option = TestOptions(formula="[]")
test.save_formula()
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.add_dictionary(file_name="dict2.txt", label="dict2", content="test2")
html_tables = test.generate_document_results(test.count())
for html_table in html_tables:
assert html_table["name"] == "file1"
def test_generate_corpus_counts_table():
test = ContentAnalysisModel()
test.test_option = TestOptions(formula="[]")
test.save_formula()
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.add_dictionary(file_name="dict2.txt", label="dict2", content="test2")
html_table = test.generate_corpus_results(test.count())
assert html_table[0][0] == "dict1"
assert html_table[-1][0] == "dict2"
def test_generate_corpus_counts_table():
test = ContentAnalysisModel()
test.test_option = TestOptions(formula="[]")
test.save_formula()
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.add_dictionary(file_name="dict2.txt", label="dict2", content="test2")
html_table = test.generate_corpus_counts_table(test.count(),
test.dictionary_colors)
assert html_table.startswith("<table")
assert html_table.endswith("</table>")
def test_generate_averages():
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.test_option = TestOptions(formula="0")
test.save_formula()
test.generate_scores()
test.generate_averages()
assert test.averages == [1.0, 0.0, 1.0, 0.0]
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test')
test.add_file(file_name="file2", label='file2', content='other file')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.test_option = TestOptions(formula="0")
test.save_formula()
test.generate_scores()
test.generate_averages()
assert test.averages == [0.5, 0.0, 1.5, 0.0]
test.count()
test.test_option = TestOptions(formula="4*[dict1]**2")
test.save_formula()
test.generate_scores()
test.generate_averages()
assert test.averages == [0.5, 2.0, 1.5, 2.0]
def test_generate_scores():
test = ContentAnalysisModel(TestOptions(formula=""))
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.generate_scores()
assert test.scores[0] == 0.0
test.test_option = TestOptions(formula="[dict1]")
test.save_formula()
test.generate_scores()
assert test.scores[0] == 1
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test a')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.test_option = TestOptions(formula="[dict1]")
test.save_formula()
test.generate_scores()
assert test.scores[0] == 0.5
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.test_option = TestOptions(formula="[dict1]*2")
test.save_formula()
test.generate_scores()
assert test.scores[0] == 1
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test a')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
test.test_option = TestOptions(formula="[dict1]")
test.save_formula()
test.generate_scores()
assert test.scores[0] == round(1 / 3, 3)
def test_count_words():
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
assert test.counters[0][0] == 1
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='test test test')
test.add_dictionary(file_name="dict1.txt", label="dict1", content="test")
test.count()
assert test.counters[0][0] == 3
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test')
test.add_dictionary(file_name="dict1.txt", label="dict1",
content="test, a")
test.count()
assert test.counters[0][0] == 2
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test')
test.add_dictionary(file_name="dict1.txt", label="dict1",
content="test, a, a test")
test.count()
assert test.counters[0][0] == 1
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test test')
test.add_dictionary(file_name="dict1.txt", label="dict1",
content="test, a, a test")
test.count()
assert test.counters[0][0] == 2
test = ContentAnalysisModel()
test.add_file(file_name="file1", label='file1', content='a test test a')
test.add_dictionary(file_name="dict1.txt", label="dict1",
content="test, a, a test")
test.count()
assert test.counters[0][0] == 3