def test_similar_values_different_facets_then_return_no_question(self):
question_generator = QuestionGenerator()
facet_a1 = Facet('NameA', 'ValueA')
facet_a2 = Facet('NameA', 'ValueA')
facets_by_document = {'1': [facet_a1], '2': [facet_a2], '3': [facet_a2]}
questions = question_generator.generate_questions(facets_by_document)
self.assertEqual(len(questions), 0)
def test_one_discriminating_facet_then_return_one_element_question(self):
question_generator = QuestionGenerator()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facets_by_document = {'1': [facet_a1], '2': [facet_a2], '3': [facet_a2]}
questions = question_generator.generate_questions(facets_by_document)
self.assertEqual(len(questions), 1)
self.assertEqual(questions[0].question.facet_name, "NameA")
self.assertEqual(sorted(questions[0].question.facet_values), ['ValueA1', 'ValueA2'])
def test_similar_values_different_facets_then_return_one_question(self):
question_generator = QuestionGenerator()
facet_a = Facet('NameA', 'ValueA')
facet_a2 = Facet('NameA', 'ValueA')
facet_b = Facet('NameB', 'ValueB')
facets_by_document = {'1': [facet_a2, facet_b], '2': [facet_a]}
questions = question_generator.generate_questions(facets_by_document)
self.assertEqual(questions[0].facet_name, "NameB")
self.assertEqual(sorted(questions[0].facet_values), ['ValueB'])
def test_when_no_discriminating_facet_then_return_no_facet(self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facet_b = Facet('NameB', 'ValueB')
facets_by_document = {'1': [facet_a1], '2': [facet_a2], '3': [facet_b]}
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 0)
def test_create_facet_dictionary(self):
facet_dictionary = FacetDictionary()
tuples = [('uri1', [Facet('@y', '21'), Facet('@m', '9')]), ('uri2', [Facet('@y', '19')]), ('uri3', [Facet('@m', '9')])]
facets_by_document = dict(tuples)
facets = facet_dictionary.create_facet_dict(facets_by_document)
self.assertEqual(2, len(facets))
self.assertEqual(1, list(facets.keys()).index('@y'))
self.assertEqual(0, list(facets.get('@y')).index('19'))
def generate_data(self):
facetA = Facet("FacetA", "FacetValueA")
facetA2 = Facet("FacetA", "FacetValueA2")
facetB = Facet("FacetB", "FacetValueB")
facetC = Facet("FacetC", "FacetValueC")
return {'Document1': [facetA, facetB],
'Document2': [facetB, facetC],
'Document3': [facetA, facetC],
'Document4': [facetA2, facetA],
'Document5': [facetA2, facetC]}
def test_similar_values_different_facets_then_return_one_question(self):
question_generator = QuestionGenerator()
facet_a1 = Facet('NameA', 'ValueA')
facet_a2 = Facet('NameA', 'ValueA')
facet_a3 = Facet('NameA', 'ValueA')
facet_b1 = Facet('NameB', 'ValueB1')
facet_b2 = Facet('NameB', 'ValueB2')
facets_by_document = {'1': [facet_a1, facet_b1], '2': [facet_a2, facet_b1], '3': [facet_a3, facet_b2]}
questions = question_generator.generate_questions(facets_by_document)
self.assertEqual(len(questions), 1)
self.assertEqual(questions[0].question.facet_name, "NameB")
self.assertEqual(sorted(questions[0].question.facet_values), ['ValueB1', 'ValueB2'])
def test_when_facet_dont_have_2_values_then_return_0_facet(self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a = Facet('NameA', 'ValueA')
facet_b = Facet('NameB', 'ValueB')
facets_by_document = {
'1': [facet_a],
'2': [facet_a],
'3': [facet_b],
'4': [facet_a]
}
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 0)
def test_when_facet_dont_have_3_documents_then_return_0_facet(self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facet_a3 = Facet('NameA', 'ValueA3')
facet_a4 = Facet('NameA', 'ValueA4')
facets_by_document = {
'1': [facet_a1, facet_a2],
'2': [facet_a3, facet_a4]
}
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 0)
def test_when_same_facet_then_return_no_questions(self):
question_generator = QuestionGenerator()
facet_a = Facet('NameA', 'ValueA')
facets_by_document = {'1': [facet_a], '2': [facet_a]}
questions = question_generator.generate_questions(facets_by_document)
self.assertEqual(len(questions), 0)
def extract_facet_from_file(file_path):
with open(file_path, errors='ignore') as jsonfile:
data = json.load(jsonfile)
name = data['FacetName']
value = data['FacetValue']
documents = []
for document in data['Documents']:
documents.append(document['ClickUri'])
return Facet(name, value), documents
def test_when_1_discriminating_facet_then_rerun_algorithm_and_return_1_facet(
self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facet_b = Facet('NameB', 'ValueB')
facets_by_document = {
'1': [facet_a1],
'2': [facet_a2],
'3': [facet_a2],
'4': [facet_b]
}
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 1)
self.assertEqual(discriminating_facets[0].name, 'NameA')
self.assertEqual(discriminating_facets[0].values,
['ValueA1', 'ValueA2'])
self.assertEqual(discriminating_facets[0].score, 1)
def test_must_not_have_facet_a_or_b_chained(self):
document_filter = DocumentFilter()
documents = document_filter.keep_documents_without_facets(self.generate_data(), [Facet("FacetA", "FacetValueA")])
documents = document_filter.keep_documents_without_facets(documents, [Facet("FacetB", "FacetValueB")])
self.assertEqual(1, len(documents))
self.assertTrue("Document1" not in documents)
self.assertTrue("Document2" not in documents)
self.assertTrue("Document3" not in documents)
self.assertTrue("Document4" not in documents)
self.assertTrue("Document5" in documents)
def test_multiple_values_then_return_multiple_questions(self):
question_generator = QuestionGenerator()
facet_a = Facet('NameA', 'ValueA')
facet_b = Facet('NameA', 'ValueB')
facet_c = Facet('NameB', 'Value1')
facet_d = Facet('NameB', 'Value2')
facet_e = Facet('NameB', 'Value3')
facets_by_document = {
'1': [facet_a, facet_d, facet_e],
'2': [facet_b, facet_c, facet_e]
}
questions = sorted(
question_generator.generate_questions(facets_by_document),
key=lambda x: x.facet_name)
self.assertEqual(questions[0].facet_name, 'NameA')
self.assertEqual(sorted(questions[0].facet_values),
['ValueA', 'ValueB'])
self.assertEqual(questions[1].facet_name, 'NameB')
self.assertEqual(sorted(questions[1].facet_values),
['Value1', 'Value2'])
def test_when_2_discriminating_facets_then_run_one_time_algorithm_and_return_2_facets(
self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facet_b = Facet('NameB', 'ValueB')
facet_c1 = Facet('NameC', 'ValueC1')
facet_c2 = Facet('NameC', 'ValueC2')
facet_c3 = Facet('NameC', 'ValueC3')
facets_by_document = {
'1': [facet_a1],
'2': [facet_a2, facet_c1],
'3': [facet_a1, facet_c2, facet_c3],
'4': [facet_b, facet_c3]
}
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 2)
self.assertEqual(discriminating_facets[0].name, 'NameC')
self.assertEqual(discriminating_facets[0].values,
['ValueC1', 'ValueC2', 'ValueC3'])
self.assertEqual(discriminating_facets[0].score, 1)
self.assertEqual(discriminating_facets[1].name, 'NameA')
self.assertEqual(discriminating_facets[1].values,
['ValueA1', 'ValueA2'])
self.assertEqual(round(discriminating_facets[1].score, 2), 0.67)
def filter_document_by_facets():
content = request.get_json()
documents_to_filter = content['Documents']
documents = dict((k, facets_by_document[k]) for k in documents_to_filter
if k in facets_by_document)
if content['MustHaveFacets'] is not None:
must_have_facets = [
Facet(val['Name'], val['Value'])
for val in content['MustHaveFacets']
]
documents = DocumentFilter.keep_documents_with_facets(
documents, must_have_facets)
if content['MustNotHaveFacets'] is not None:
must_not_have_facets = [
Facet(val['Name'], val['Value'])
for val in content['MustNotHaveFacets']
]
documents = DocumentFilter.keep_documents_without_facets(
documents, must_not_have_facets)
return jsonify(list(documents.keys()))
def generate_data(self):
facetA = Facet("NameA", "ValueA")
facetB = Facet("NameB", "ValueB")
return {'document1': [facetA, facetB], 'document2': [facetA]}
def test_must_have_facet_a(self):
document_filter = DocumentFilter()
documents = document_filter.keep_documents_with_facets(self.generate_data(), [Facet("FacetA", "FacetValueA")])
self.assertEqual(3, len(documents))
self.assertTrue("Document1" in documents)
self.assertTrue("Document2" not in documents)
self.assertTrue("Document3" in documents)
self.assertTrue("Document4" in documents)
self.assertTrue("Document5" not in documents)
def test_must_not_have_facet_d(self):
document_filter = DocumentFilter()
documents = document_filter.keep_documents_without_facets(self.generate_data(), [Facet("FacetD", "FacetValueD")])
self.assertEqual(5, len(documents))
self.assertTrue("Document1" in documents)
self.assertTrue("Document2" in documents)
self.assertTrue("Document3" in documents)
self.assertTrue("Document4" in documents)
self.assertTrue("Document5" in documents)
def test_when_document_counts_in_facet_values_have_more_than_35_in_standard_deviation_then_return_0_facet(
self):
discriminating_algo = DiscriminatingFacetsAlgo()
facet_a1 = Facet('NameA', 'ValueA1')
facet_a2 = Facet('NameA', 'ValueA2')
facet_b = Facet('NameB', 'ValueB')
facets_by_document = {
'1': [facet_a1],
'2': [facet_a2],
'3': [facet_a2],
'4': [facet_a2],
'5': [facet_a2],
'6': [facet_a2],
'7': [facet_a2],
'8': [facet_a2],
'9': [facet_a2],
'10': [facet_a2],
'11': [facet_a2],
'12': [facet_a2],
'13': [facet_a2],
'14': [facet_a2],
'15': [facet_a2],
'16': [facet_a2],
'17': [facet_a2],
'18': [facet_a2],
'19': [facet_a2],
'20': [facet_a2],
'21': [facet_a2],
'22': [facet_a2],
'23': [facet_a2],
'24': [facet_a2],
'25': [facet_a2],
'26': [facet_a2],
'27': [facet_a2],
'28': [facet_a2],
'29': [facet_a2],
'30': [facet_a2],
'31': [facet_a2],
'32': [facet_a2],
'33': [facet_a2],
'34': [facet_a2],
'35': [facet_a2],
'36': [facet_a2],
'37': [facet_a2],
'38': [facet_a2],
'39': [facet_a2],
'40': [facet_a2],
'41': [facet_a2],
'42': [facet_a2],
'43': [facet_a2],
'45': [facet_a2],
'46': [facet_a2],
'47': [facet_a2],
'48': [facet_a2],
'49': [facet_a2],
'50': [facet_a2],
'51': [facet_a2],
'52': [facet_a2],
'53': [facet_a2],
'54': [facet_a2],
'55': [facet_a2],
'56': [facet_a2],
'57': [facet_a2],
'58': [facet_b]
}
documents_per_facet_value_counts = [1, 55]
standard_deviation = statistics.stdev(documents_per_facet_value_counts)
self.assertTrue(standard_deviation > 35)
discriminating_facets = discriminating_algo.get_discriminating_facets(
facets_by_document)
self.assertEqual(len(discriminating_facets), 0)
