def setUp(self):
self.xml_data_space = XMLData(" ", 10.0, 10.0, "Font", 23.0, False,
True)
self.xml_data_B = XMLData("B", 15.0, 70.0, "Font", 23.0, False, True)
self.xml_data_y_1 = XMLData(" ", 10.0, 10.0, "Font", 25.0, False, True)
self.xml_data_y_2 = XMLData(" ", 10.0, 15.0, "Font", 25.0, False, True)
self.xml_data_y_3 = XMLData(" ", 10.0, 20.0, "Font", 25.0, False, True)
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 = VincentExtractor.check_for_repeated_phrases(self, xml_data, 0)
self.assertTrue(expected_result == actual_result)
def test_xml_no_space(self):
xml_data = []
xml_data.append(XMLData("b", 270.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("c", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData("d", 10.0, 10.0, "Font", 23.0, False, True))
expected_result = 13.0
actual_result = VincentExtractor.get_space_size(self, xml_data)
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_true_on_end(self):
xml_data = []
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 23.0, False, True))
expected_result = 2
actual_result = PerthExtractor.check_for_repeated_phrases(
self, xml_data, 0)
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_false_on_second_space(self):
xml_data = []
xml_data.append(XMLData(" ", 10.0, 10.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 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 = VincentExtractor.get_space_size(self, xml_data)
self.assertTrue(expected_result == actual_result)
tured_output = io.StringIO()
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_two_bracket_cs(self):
xml_data = self.xml_data_list
xml_data[0].character = "("
xml_data[1].character = " "
xml_data[2].character = " "
xml_data[3].character = ")"
xml_data.append(XMLData(" ", 170.0, 60.0, "Font", 23.0, False, True))
xml_data.append(XMLData(" ", 180.0, 60.0, "Font", 23.0, False, True))
expected_result = ["( ) "]
actual_result = self.v.create_sentences(xml_data)
self.assertEqual(expected_result, 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_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_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_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 characters(self, content):
""" For each character, add it to the XMLData list.
Parse out the data from each line in the XML document if it has a character.
If any invalid data is passed to XMLData class, skip that character.
Args:
content: The character that is being parsed.
Returns:
Void
"""
if (self.tag == "text" and self.b_box != ""):
self.current_char = content
b_box_values = self.process_b_box((self.b_box))
' Commented this out, just for now - Pushed back 19 lines'
#if not "\n" in content:
if "\n" not in content:
#print("Keep the \n")
next_letter = XMLData(content, b_box_values[0],
b_box_values[1], self.font, self.size,
self.bold, self.italics)
if (next_letter.valid == True):
self.xml_data.append(next_letter)
else: # It is a \n
if (len(self.xml_data) > 1):
# If the previous letter was a \n keep it, else, ignore // Used to be a "." that we check to keep it
if (self.xml_data[-1]).character == "\n":
next_letter = XMLData(content, b_box_values[0],
b_box_values[1], self.font,
self.size, self.bold,
self.italics)
if (next_letter.valid == True):
self.xml_data.append(next_letter)
self.next_sentence = self.next_sentence + content
# If we reach a full stop then finish the sentence and add it to sentences
if content == ".":
self.sentences.append(self.next_sentence)
self.next_sentence = ""
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_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_three_gap_add(self):
xml_data = self.xml_data_list
xml_data[1].character = " "
xml_data[2].character = " "
xml_data[3].character = " "
xml_data[4].character = "b"
xml_data[4].b_y = 60
xml_data[5].b_y = 70.0
xml_data[7].b_x = 150.0
xml_data.append(XMLData(" ", 160.0, 60.0, "Font", 23.0, False, True))
v = VincentExtractor()
expected_result = ["a b"]
actual_result = v.create_sentences(xml_data)
self.assertEqual(expected_result, actual_result)
def test_b_y_none(self):
xml_bad = XMLData("a", 344.22, 'y', "Some Font", 24, True, True)
self.assertTrue(xml_bad.b_y == None)
def test_italics_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, True,
"Italics")
self.assertTrue(xml_bad.italics == None)
def test_italics_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, True,
"Italics")
self.assertFalse(xml_bad.valid)
def test_bold_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, "Bold", True)
self.assertTrue(xml_bad.bold == None)
def test_bold_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, "Bold", True)
self.assertFalse(xml_bad.valid)
def test_size_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", "Size", True, True)
self.assertTrue(xml_bad.size == None)
def test_size_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", "Size", True, True)
self.assertFalse(xml_bad.valid)
def test_font_none(self):
xml_bad = XMLData("a", 344.22, 544.22, 1, 24, True, True)
self.assertTrue(xml_bad.font == None)
def setUp(self):
self.xmldata1 = XMLData("a", 2.6, 5.2, "sans", 12.0, True, False)
def test_font_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, 1, 24, True, True)
self.assertFalse(xml_bad.valid)
def test_b_y_invalid(self):
xml_bad = XMLData("a", 344.22, 'y', "Some Font", 24, True, True)
self.assertFalse(xml_bad.valid)
class TestXMLData(TestCase):
def setUp(self):
self.xmldata1 = XMLData("a", 2.6, 5.2, "sans", 12.0, True, False)
def test_print_all_data(self):
xml_expected_output = """Character: a || b_x: 2.6 || b_y: 5.2 || Font: sans || Size: 12.0 || Bold: True || Italics: False"""
self.assertTrue(self.xmldata1.print_all_data() == xml_expected_output)
def test_character_expected(self):
self.assertTrue(self.xmldata1.character == "a")
def test_x_b_box_expected(self):
self.assertTrue(self.xmldata1.b_x == 2.6)
def test_x_y_box_expected(self):
self.assertTrue(self.xmldata1.b_y == 5.2)
def test_font_expected(self):
self.assertTrue(self.xmldata1.font == "sans")
def test_size_expected(self):
self.assertTrue(self.xmldata1.size == 12.0)
def test_bold_expected(self):
self.assertTrue(self.xmldata1.bold)
def test_italics_expected(self):
self.assertFalse(self.xmldata1.italics)
def test_character_invalid(self):
xml_bad = XMLData(3, 344.22, 554.22, "Some Font", 24, True, True)
self.assertFalse(xml_bad.valid)
def test_character_none(self):
xml_bad = XMLData(3, 344.22, 554.22, "Some Font", 24, True, True)
self.assertTrue(xml_bad.character == None)
def test_b_x_invalid(self):
xml_bad = XMLData("a", 's', 554.22, "Some Font", 24, True, True)
self.assertFalse(xml_bad.valid)
def test_b_x_none(self):
xml_bad = XMLData("a", 's', 554.22, "Some Font", 24, True, True)
self.assertTrue(xml_bad.b_x == None)
def test_b_y_invalid(self):
xml_bad = XMLData("a", 344.22, 'y', "Some Font", 24, True, True)
self.assertFalse(xml_bad.valid)
def test_b_y_none(self):
xml_bad = XMLData("a", 344.22, 'y', "Some Font", 24, True, True)
self.assertTrue(xml_bad.b_y == None)
def test_font_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, 1, 24, True, True)
self.assertFalse(xml_bad.valid)
def test_font_none(self):
xml_bad = XMLData("a", 344.22, 544.22, 1, 24, True, True)
self.assertTrue(xml_bad.font == None)
def test_size_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", "Size", True, True)
self.assertFalse(xml_bad.valid)
def test_size_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", "Size", True, True)
self.assertTrue(xml_bad.size == None)
def test_bold_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, "Bold", True)
self.assertFalse(xml_bad.valid)
def test_bold_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, "Bold", True)
self.assertTrue(xml_bad.bold == None)
def test_italics_invalid(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, True,
"Italics")
self.assertFalse(xml_bad.valid)
def test_italics_none(self):
xml_bad = XMLData("a", 344.22, 544.22, "Some Font", 24, True,
"Italics")
self.assertTrue(xml_bad.italics == None)
