def test_add_empty_next_sentence(self):
letter = XMLData("b", 10.0, 10.0, "Font", 23.0, True, False)
next_sentence = ""
p = PerthExtractor()
expected_result = ["b"]
actual_result = p.test_add_sentence(next_sentence, letter)
self.assertTrue(expected_result == actual_result)
def test_letter_none(self):
next_sentence = "aaa"
letter = None
p = PerthExtractor()
expected_result = ["aaa"]
actual_result = p.test_add_sentence(next_sentence, letter)
self.assertTrue(expected_result == actual_result)
def test_return_empty_String(self):
letter = self.xml_data_space
next_sentence = "Hello"
p = PerthExtractor()
expected_result = ""
actual_result = p.add_sentence(next_sentence, letter)
self.assertTrue(expected_result == actual_result)
def test_dont_enter_while(self):
xml_data = []
xml_data.append(XMLData("", 10.0, 15.0, "Font", 25.0, False, True))
xml_data.append(XMLData("", 10.0, 15.0, "Font", 25.0, False, True))
expected_result = []
p = PerthExtractor()
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_remove_footer(self):
xml_data = self.add_three_gap()
xml_data.append(XMLData("", 10.0, 49.0, "Font", 23.0, False, True))
xml_data.insert(0, XMLData("", 10.0, 49.0, "Font", 23.0, False, True))
expected_result = []
p = PerthExtractor()
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_skip_on_two_gap(self):
xml_data = self.add_two_gap()
xml_data.insert(0, XMLData("a", 10.0, 100.0, "Font", 23.0, False,
True))
xml_data.append(XMLData("b", 16.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("b", 770.0, 100.0, "Font", 23.0, False, True))
expected_result = []
p = PerthExtractor()
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_create_sentences_none(self):
captured_output = io.StringIO()
sys.stdout = captured_output
xml_data = None
p = PerthExtractor()
expected_result = []
actual_result = p.create_sentences(xml_data)
sys.stdout = sys.__stdout__
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_skip_all_gaps(self):
xml_data = []
xml_data.append(XMLData(".", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("a", 270.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("b", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("c", 26.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("d", 10.0, 10.0, "Font", 23.0, False, True))
expected_result = []
p = PerthExtractor()
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_none_next_sentence(self):
captured_output = io.StringIO()
sys.stdout = captured_output
letter = XMLData("b", 10.0, 10.0, "Font", 23.0, True, False)
next_sentence = None
p = PerthExtractor()
expected_result = []
actual_result = p.test_add_sentence(next_sentence, letter)
sys.stdout = sys.__stdout__
self.assertTrue(expected_result == actual_result)
def test_two_gap_add(self):
#xml_data = self.add_two_gap()
xml_data = []
xml_data.append(XMLData("A", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 26.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("d", 26.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("d", 10.0, 10.0, "Font", 23.0, False, True))
expected_result = []
p = PerthExtractor()
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_two_words_no_removal(self):
sentences = []
sentences.append("aaaa")
sentences.append("bbbb")
expected_result = sentences
actual_result = PerthExtractor.call_remove(self, sentences)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_create_remove_footer(self):
xml_data = []
xml_data.append(XMLData("a", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("b", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("c", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("d", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData(".", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("z", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("z", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("z", 10.0, 100.0, "Font", 23.0, False, True))
p = PerthExtractor()
expected_result = ["bcd."]
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_remove_short_sentence(self):
sentences = []
sentences.append("ab")
expected_result = []
actual_result = PerthExtractor.call_remove(self, sentences)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_immediate_exit(self):
xml_data = []
xml_data.append(self.xml_data_space)
expected_result = 13.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_xml_none(self):
captured_output = io.StringIO()
sys.stdout = captured_output
xml_data = None
expected_result = 0.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
sys.stdout = sys.__stdout__
self.assertTrue(expected_result == actual_result)
def test_false_on_space(self):
xml_data = []
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("b", 10.0, 100.0, "Font", 23.0, False, True))
xml_data.append(XMLData("c", 10.0, 100.0, "Font", 23.0, False, True))
expected_result = 0.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_xml_only_bold(self):
xml_data = []
xml_data.append(XMLData("b", 270.0, 10.0, "Bold", 23.0, False, True))
xml_data.append(XMLData("c", 10.0, 10.0, "Bold", 23.0, False, True))
xml_data.append(XMLData("d", 10.0, 10.0, "Bold", 23.0, False, True))
expected_result = 13.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_false_on_y_gap(self):
xml_data = []
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 24.0, False, True))
xml_data.append(XMLData("b", 10.0, 100.0, "Font", 24.0, False, True))
expected_result = 1
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
self.assertTrue(expected_result == actual_result)
def test_finding_space_gap(self):
xml_data = []
xml_data.append(self.xml_data_space)
xml_data.append(self.xml_data_B)
expected_result = 5.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_repeated_xml_none(self):
captured_output = io.StringIO()
sys.stdout = captured_output
xml_data = None
expected_result = 0.0
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
sys.stdout = sys.__stdout__
self.assertTrue(expected_result == actual_result)
def test_reach_list_end_immediately(self):
xml_data = []
xml_data.append(self.xml_data_space)
expected_result = 1
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 1)
self.assertTrue(expected_result == actual_result)
def test_immediate_different_line(self):
xml_data = []
xml_data.append(self.xml_data_y_1)
xml_data.append(self.xml_data_space)
expected_result = 1.0
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
self.assertTrue(expected_result == actual_result)
def test_removed_character(self):
xml_data = []
xml_data.append(self.xml_data_y_1)
xml_data.append(self.xml_data_y_2)
xml_data.append(self.xml_data_y_3)
expected_result = 2.0
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
self.assertTrue(expected_result == actual_result)
def test_execute_false_if(self):
xml_data = []
xml_data.append(self.xml_data_space)
xml_data.append(self.xml_data_space)
xml_data.append(self.xml_data_B)
expected_result = 5.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_double_loop(self):
xml_data = []
xml_data.append(self.xml_data_y_1)
xml_data.append(self.xml_data_y_1)
xml_data.append(self.xml_data_y_1)
xml_data.append(self.xml_data_space)
expected_result = 3.0
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
self.assertTrue(expected_result == actual_result)
def determine_factory(self, file_name):
"""Determine which CouncilExtractor will parse the XMLData.
Using the name of the file passed in, determine which factory class,
a CouncilExtractor, will parse the XMLData list.
Args:
file_name: The name of the file being parsed.
Returns: PerthExtractor() or some other CouncilExtractor once they are implemented.
"""
try:
file_name = file_name.lower()
if "perth" in file_name:
#print("Do the Perth one")
return PerthExtractor()
# Add a return Vincent class once we make a Vincent parser.
elif "vincent" in file_name:
return VincentExtractor()
#print("do the vincent one")
# Add a return Stirling class once we make the Stirling parser.
elif "stirling" in file_name:
print("Do the Stirling one")
# If there isn't a recognised council in the file_name string then (for now),
# raise an error, and later possibly later it will be changed to just parse
# it for Perth.
# If None is passed, throw a type error
else:
raise ValueError(file_name)
except (TypeError, AttributeError) as e:
message = "File_Name is not a String type: %s" % (e.args)
raise TypeError(e.__str__() + "\n" + message)
def test_xml_empty(self):
xml_data = []
expected_result = 13.0
actual_result = PerthExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
def test_only_spaces(self):
sentences = []
sentences.append(" ")
expected_result = sentences
actual_result = PerthExtractor.call_remove(self, sentences)
self.assertTrue(expected_result == actual_result)
def test_create_xml_empty(self):
xml_data = []
p = PerthExtractor()
expected_result = []
actual_result = p.create_sentences(xml_data)
self.assertTrue(all(elem in expected_result for elem in actual_result))
def test_one_length(self):
sentences = []
sentences.append("a")
expected_result = []
actual_result = PerthExtractor.call_remove(self, sentences)
self.assertTrue(actual_result == expected_result)