def test_countWords_countsCorrectlyForTwoFiles(self):
# arrange
inputReader = InputReader(FILE)
inputReaderMock = Mock()
inputReaderMock.return_value = TEXT
inputReader.readInput = inputReaderMock
inputReader2 = InputReader(FILE2)
inputReaderMock2 = Mock()
inputReaderMock2.return_value = TEXT2
inputReader2.readInput = inputReaderMock2
sourceLoader = SourceLoader()
sourceLoaderMock = Mock()
sourceLoaderMock.return_value = [inputReader, inputReader2]
sourceLoader.loadSources = sourceLoaderMock
wordsToExcludeInputReader = InputReader(EMPTY)
wordsToExcludeInputReaderMock = Mock()
wordsToExcludeInputReaderMock.return_value = WORDS_TO_EXCLUDE
wordsToExcludeInputReader.readInput = wordsToExcludeInputReaderMock
inputCleaner = InputCleaner(wordsToExcludeInputReader,
InputReader(EMPTY))
wordCounter = WordCounter(inputCleaner, sourceLoader)
# act
output = wordCounter.countWords()
# assert
self.assertEqual(output, OUTPUT2)
def test_loadSources_returnsEmptyArrayWhenNothingIsInput(self):
# arrange
readInputMock = Mock()
readInputMock.return_value = EMPTY
inputReader = InputReader(EMPTY)
inputReader.readInput = readInputMock
sourceLoader = SourceFromCommandLineLoader(inputReaderType = InputReader, inputReader = inputReader)
# act
sources = sourceLoader.loadSources()
# assert
self.assertEqual(sources, [])
def test_loadSources_returnsArrayOfLengthOneForOneInput(self):
# arrange
readInputMock = Mock()
readInputMock.side_effect = [INPUT, EMPTY]
inputReader = InputReader(EMPTY)
inputReader.readInput = readInputMock
sourceLoader = SourceFromCommandLineLoader(inputReaderType = InputReader, inputReader = inputReader)
# act
sources = sourceLoader.loadSources()
# assert
self.assertEqual(len(sources), 1)
self.assertEqual(type(sources[0]), InputReader)
def test_countWords_countsCorrectlyForNoFile(self):
# assert
sourceLoader = SourceLoader()
sourceLoaderMock = Mock()
sourceLoaderMock.return_value = []
sourceLoader.loadSources = sourceLoaderMock
inputCleaner = InputCleaner(InputReader(EMPTY), InputReader(EMPTY))
wordCounter = WordCounter(inputCleaner, sourceLoader)
# act
output = wordCounter.countWords()
# assert
self.assertEqual(output, {})
def test_cleanInputCleansInputCorrectly(self):
# arrange
symbolsReader = InputReader(EMPTY)
wordsReader = InputReader(EMPTY)
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
inputCleaner.symbolsToStrip = [',', '.', '!']
inputCleaner.wordsToExclude = ["der", "die"]
# act
text = inputCleaner.cleanInput(TEXT)
# assert
self.assertEqual(text.replace(WHITESPACE, EMPTY),
CLEANED_TEXT.replace(WHITESPACE, EMPTY))
def test_loadSources_returnsEmptyListForFileNotFoundAssertion(self):
# arrange
readInputMock = Mock()
readInputMock.side_effect = [INPUT, EMPTY]
inputReader = InputReader(EMPTY)
inputReader.readInput = readInputMock
assertionErrorInputReaderMock = Mock()
assertionErrorInputReaderMock.side_effect = AssertionError
sourceLoader = SourceFromCommandLineLoader(inputReaderType = assertionErrorInputReaderMock, inputReader = inputReader)
# act
sources = sourceLoader.loadSources()
# assert
self.assertEqual(sources, [])
def test_countWords_countsCorrectlyForEmptyFile(self):
# arrange
inputReader = InputReader(FILE)
inputReaderMock = Mock()
inputReaderMock.return_value = EMPTY
inputReader.readInput = inputReaderMock
sourceLoader = SourceLoader()
sourceLoaderMock = Mock()
sourceLoaderMock.side_effect = [[inputReader]]
sourceLoader.loadSources = sourceLoaderMock
inputCleaner = InputCleaner(InputReader(EMPTY), InputReader(EMPTY))
wordCounter = WordCounter(inputCleaner, sourceLoader)
# act
output = wordCounter.countWords()
# assert
expected = {}
expected[FILE] = defaultdict(int)
self.assertEqual(output, expected)
def loadSymbolsToStrip(self, symbolsToStripInputReader: InputReader):
"""
Retruns the symbols that are to be stripped (provided InputReader is expected to read only symbols)
"""
symbolsToStrip: List[str] = []
# try to load symbols that are to be stripped from the input
try:
# read the symbols
symbols = symbolsToStripInputReader.readInput()
# add every symbol to the list of symbols to strip if it is not the empty symbol
for symbol in symbols:
if symbol != constants.EMPTY:
symbolsToStrip.append(symbol)
finally:
return symbolsToStrip
def test_loadWordsToExclude_returnsEmptyWhenNothingToRead(self):
# arrange
symbolsReader = InputReader(EMPTY)
symbolsReaderMock = Mock()
symbolsReader.return_value = EMPTY
symbolsReader.readInput = symbolsReaderMock
wordsReader = InputReader(EMPTY)
wordsReaderMock = Mock()
wordsReaderMock.return_value = EMPTY
wordsReader.readInput = wordsReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
words = inputCleaner.wordsToExclude
# assert
self.assertEqual(words, [])
def test_loadWordsToExclude_returnsEmptyOnError(self):
# arrange
symbolsReader = InputReader(EMPTY)
symbolsReaderMock = Mock()
symbolsReader.return_value = EMPTY
symbolsReader.readInput = symbolsReaderMock
wordsReader = InputReader(EMPTY)
wordsReaderMock = Mock()
wordsReaderMock.side_effect = [AssertionError]
wordsReader.readInput = wordsReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
words = inputCleaner.wordsToExclude
# assert
self.assertEqual(words, [])
def test_loadWordsToExclude_returnsWords(self):
# arrange
symbolsReader = InputReader(EMPTY)
symbolsReaderMock = Mock()
symbolsReader.return_value = EMPTY
symbolsReader.readInput = symbolsReaderMock
wordsReader = InputReader(EMPTY)
wordsReaderMock = Mock()
wordsReaderMock.return_value = WORDS
wordsReader.readInput = wordsReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
words = inputCleaner.wordsToExclude
# assert
self.assertEqual(len(words), 3)
self.assertEqual(words, ["der", "die", "das"])
def test_loadSymbolsToStrip_returnsEmptyWhenErrorOccurs(self):
# arrange
readSymbolsMock = Mock()
readSymbolsMock.side_effect = [AssertionError]
symbolsReader = InputReader(EMPTY)
symbolsReader.readInput = readSymbolsMock
wordReaderMock = Mock()
wordReaderMock.return_value = EMPTY
wordsReader = InputReader(EMPTY)
wordsReader.readInput = wordReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
symbols = inputCleaner.symbolsToStrip
# assert
self.assertEqual(symbols, [])
def test_loadSymbolsToStrip_returnsEmptyWhenNothingToRead(self):
# arrange
readSymbolsMock = Mock()
readSymbolsMock.return_value = EMPTY
symbolsReader = InputReader(EMPTY)
symbolsReader.readInput = readSymbolsMock
wordReaderMock = Mock()
wordReaderMock.return_value = EMPTY
wordsReader = InputReader(EMPTY)
wordsReader.readInput = wordReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
symbols = inputCleaner.symbolsToStrip
# assert
self.assertEqual(symbols, [])
def loadWordsToExclude(self, wordsToExcludeInputReader: InputReader):
"""
Returns the words that are to be excluded (provided InputReader is expected to read comma separated values - csv)
"""
wordsToExclude: List[str] = []
# try to load words that are to be excluded
try:
# read the words
words = wordsToExcludeInputReader.readInput()
# cast to lower case
words = words.lower()
# remove whitespaces
words = words.replace(constants.WHITESPACE, constants.EMPTY)
# split the words by comma
splittedWords = words.split(constants.COMMA)
# remove the empty symbol
if constants.EMPTY in splittedWords:
splittedWords.remove(constants.EMPTY)
# add the words to the words to be excluded
wordsToExclude.extend(splittedWords)
finally:
return wordsToExclude
def test_loadSymbolsToStrip_returnsSymbols(self):
# arrange
readSymbolsMock = Mock()
readSymbolsMock.return_value = SYMBOLS
symbolsReader = InputReader(EMPTY)
symbolsReader.readInput = readSymbolsMock
wordReaderMock = Mock()
wordReaderMock.return_value = EMPTY
wordsReader = InputReader(EMPTY)
wordsReader.readInput = wordReaderMock
# act
inputCleaner: InputCleaner = InputCleaner(wordsReader, symbolsReader)
symbols = inputCleaner.symbolsToStrip
# assert
self.assertEqual(len(symbols), 5)
self.assertEqual(symbols, [".", ",", ";", ":", "!"])
def test(self):
reader = InputReader(EMPTY)
reader.readInput()
from VV_E1_WordCount.InputReader import InputReader, FileReader, CommandLineReader
from VV_E1_WordCount.InputCleaner import InputCleaner
from VV_E1_WordCount.SourceLoader import SourceFromCommandLineLoader, SourceLoader
from VV_E1_WordCount.OutputWriter import OutputToConsoleWriter
from VV_E1_WordCount.WordCounter import WordCounter
import VV_E1_WordCount.constants as constants
# try to get the location of the file configuring the words to exclude
wordsToExcludePath = ""
wordsToExcludeInputReader = None
if len(sys.argv) == 3:
wordsToExcludePath = sys.argv[2]
try:
wordsToExcludeInputReader = FileReader(wordsToExcludePath)
except (AssertionError):
wordsToExcludeInputReader = InputReader(constants.EMPTY)
# try to get the location of the file configuring the symbols to strip
symbolsToStripPath = ""
symbolsToStripInputReader = None
if len(sys.argv) >= 2:
symbolsToStripPath = sys.argv[1]
try:
symbolsToStripInputReader = FileReader(symbolsToStripPath)
except (AssertionError):
symbolsToStripInputReader = InputReader(constants.EMPTY)
# composition root
inputCleaner: InputCleaner = InputCleaner(wordsToExcludeInputReader,
symbolsToStripInputReader)
sourceLoader: SourceFromCommandLineLoader = SourceFromCommandLineLoader(