def openTextProcessing(self):
"""
open the GUI CurrentDownload and start the text processing. Also it connects the logger
to the edit line. This Gui informs the user about which CourseData will be textually
processed.
"""
self.cleanMainWindow()
self.MainWindow = loadUi(self.pathCurrentDownload)
self.MainWindow.showMaximized()
# Activate logger to textbox
logTextBox = QTextEditLogger(self.MainWindow)
# You can format what is printed to text box
logTextBox.setFormatter(
logging.Formatter('%(asctime)s - %(levelname)s - %(message)s'))
logging.getLogger().addHandler(logTextBox)
# You can control the logging level
logging.getLogger().setLevel(logging.INFO)
# disable button until the text process is finished
self.MainWindow.but_done.setEnabled(False)
self.MainWindow.plainTextEdit.setReadOnly(True)
self.MainWindow.show()
# create the text mining logic
textProcessing = TextProcessing(database=self.database,
path=self.pathTempFolder,
textSummarizer=self.textSummarizer)
# get the CourseData from the database, which do not have a text abstraction and not raised an error in a past process
fileTypes = ['pdf', 'docx', 'pptx', 'html']
dataWithoutTextMining = self.database.getCourseDataWithoutTextMiningForParsing(
fileTypes)
# i is the position of the course data in the list
i = 1
countDataWithoutTextMining = len(dataWithoutTextMining)
self.MainWindow.plainTextEdit.appendPlainText(
"Es sind insgesamt %d Dateien textuell aufzuarbeiten!".replace(
"%d", str(countDataWithoutTextMining)))
# start for each data the text mining process
for data in dataWithoutTextMining:
self.MainWindow.plainTextEdit.appendPlainText(
"Beginn mit der %d1. Datei von %d2!".replace(
"%d1", str(i)).replace("%d2",
str(countDataWithoutTextMining)))
textProcessing.textProcessing(data)
self.MainWindow.plainTextEdit.appendPlainText(
"Fertig mit der %d1. Datei von %d2!".replace(
"%d1", str(i)).replace("%d2",
str(countDataWithoutTextMining)))
i += 1
QtWidgets.QApplication.processEvents()
# enable the button for finish this
self.MainWindow.but_done.setEnabled(True)
# deactivate the logger
logTextBox.deactivate()
# go back to main page
self.MainWindow.but_done.clicked.connect(self.goBackMainPage)
def test_textProcessor_ExceptionCheckLanguage(self, mock_apply_async, mock_pool):
"""
Test if the text processing process will work right, if a ExceptionTextAbstraction pop up
Procedure:
1. Set the mocking results
2. call the function
---------
Verification:
3. check if file handler was called right
4. check if multiprocessing was called right
5. check if text cleaner was called right
6. check if check language was called right
7. check if summarize text was called right
8. check if frequency_words was called right
9. check if update course data text fields was called right
10. check if the course data object is right
"""
def my_side_effect(text):
raise ExceptionCheckLanguage("ExceptionCheckLanguage")
dummyCourseData = CourseData(id=1, name="test.pdf", dataType="pdf")
mock_database = Mock(spec=DatabaseManager)
mock_fileToText = Mock(spec=FileToText)
mock_textAbstraction = Mock(spec=TextAbstraction)
mock_fileHandler = Mock(spec=FileHandler)
mock_checkLanguage = Mock(spec=CheckLanguage)
mock_textCleaner = Mock(spec=TextCleaner)
textProcesser = TextProcessing(database=mock_database, fileHandler=mock_fileHandler, path="C:/",
textSummarizer=mock_textAbstraction, fileToText=mock_fileToText,
checkLanguage=mock_checkLanguage, textCleaner=mock_textCleaner)
# mocking results
resultFullText = "Das ist ein Volltext"
resultCleanText = "Das ist ein sauberer Text"
# set mocking results
mock_get = MagicMock()
mock_get_function = MagicMock(return_value=resultFullText)
mock_get.get = mock_get_function
mock_apply_async.apply_async.return_value = mock_get
mock_pool.return_value = mock_apply_async
mock_fileHandler.saveFile.return_value = "C:/test.pdf"
mock_fileToText.fileToText.return_value = resultFullText
mock_textCleaner.cleaningText.return_value = resultCleanText
mock_checkLanguage.checkLanguage.side_effect = my_side_effect
textProcesser.textProcessing(dummyCourseData)
# check if FileHandler was called right
mock_fileHandler.saveFile.assert_called_with(dummyCourseData, "C:/")
self.assertTrue(mock_fileHandler.saveFile.called)
# check if multiprocessing was called right
called_args = mock_apply_async.apply_async.call_args_list[0][1]
self.assertTrue(called_args['args'] == ("C:/test.pdf", "pdf"))
self.assertTrue(called_args['func'] == mock_fileToText.fileToText)
self.assertTrue(mock_apply_async.apply_async.called)
# check if text cleaner was called right
mock_textCleaner.cleaningText.assert_called_with(resultFullText)
self.assertTrue(mock_textCleaner.cleaningText.called)
# check if check language was called right
mock_checkLanguage.checkLanguage.assert_called_with(resultCleanText)
self.assertTrue(mock_checkLanguage.checkLanguage.called)
# check if summarize text was not called
self.assertFalse(mock_textAbstraction.summarize_Text.called)
# check if frequency_words was not called
self.assertFalse(mock_textAbstraction.frequency_Words.called)
# check if update course data text fields was called right
mock_database.UpdateCourseDataTextFields.assert_called_with(dummyCourseData)
self.assertTrue(mock_database.UpdateCourseDataTextFields.called)
# check if the course data object is right
self.assertEqual(dummyCourseData.fullText, resultFullText)
self.assertEqual(dummyCourseData.abstract, None)
self.assertEqual(dummyCourseData.abstractWordFrequency, None)
self.assertEqual(dummyCourseData.error, "ExceptionCheckLanguage")