def test_lexicalStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_word_count.txt")
tokensCount, vocabularyCount, lexicalRichness = conv.stats.getLexicalStats()
self.assertEqual(tokensCount, 18)
self.assertEqual(vocabularyCount, 17)
self.assertEqual(lexicalRichness, vocabularyCount/tokensCount)
def test_delayStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_delay_conv.txt")
df = conv.stats.generateStats(IConvStats.STATS_NAME_DELAY)
self.assertEqual(df.loc['S1'].values[0],
(np.timedelta64(10, 'h') + np.timedelta64(8, 's')))
self.assertEqual(df.loc['S2'].values[0], np.timedelta64(10, 's'))
def test_wordUsageByHourAndYear(self):
words = ['the', 'hello', 'a', 'when']
conv = self.getConversation(mio.getResourcesPath() + PlottingTestCase.TEST_FILE)
stats = WordsCountStats(conv)
stats.loadWordsCount(['month', 'year'])
mplot.plotWordsCount(stats, words)
mplot.plotWordsCount(stats, words, 's1')
mplot.plotWordsCount(stats, words, 's2')
def savePlotAsImage(plot, filename):
folderPath = os.path.join(mio.getResourcesPath(), 'imgs')
if not os.path.exists(folderPath):
os.makedirs(folderPath)
if SAVE_PLOT:
filepath = os.path.join(folderPath, filename)
plot.savefig(filepath)
else:
pass
def test_intervalStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_interval_conv.txt")
start, end, interval = conv.stats.getIntervalStats()
days = conv.stats.getDaysWithoutMessages()
self.assertEqual(start, Timestamp('2014-01-01 00:00:00'))
self.assertEqual(end, Timestamp('2014.01.07 00:10:00'))
self.assertEqual(interval, datetime.timedelta(6, 600))
self.assertTrue(set(days) == set(['2014.01.02', '2014.01.06']))
def test_lexicalStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_word_count.txt")
tokensCount, vocabularyCount, lexicalRichness = conv.stats.getLexicalStats(
)
self.assertEqual(tokensCount, 18)
self.assertEqual(vocabularyCount, 17)
self.assertEqual(lexicalRichness, vocabularyCount / tokensCount)
def test_singleWordCount(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_word_count.txt")
totalCount = conv.stats.getWordCountStats(word='hello')
s1Count = conv.stats.getWordCountStats(word='hello', sender='S1')
s2Count = conv.stats.getWordCountStats(word='hello', sender='S2')
self.assertEqual(totalCount, 3)
self.assertEqual(s1Count, 2)
self.assertEqual(s2Count, 1)
def test_emoticonStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_emoticons_conv.txt")
numEmoticons = conv.stats.getEmoticonsStats()[0]
numEmoticonsS1 = conv.stats.getEmoticonsStats(sender='S1')[0]
numEmoticonsS2 = conv.stats.getEmoticonsStats(sender='S2')[0]
self.assertEqual(numEmoticons, 6)
self.assertEqual(numEmoticonsS1, 3)
self.assertEqual(numEmoticonsS2, 3)
def savePlotAsImage(plot, filename):
folderPath = os.path.join(mio.getResourcesPath(), 'imgs')
if not os.path.exists(folderPath):
os.makedirs(folderPath)
if SAVE_PLOT:
filepath = os.path.join(folderPath, filename)
plot.savefig(filepath)
else:
pass
def test_emoticonStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_emoticons_conv.txt")
numEmoticons = conv.stats.getEmoticonsStats()[0]
numEmoticonsS1 = conv.stats.getEmoticonsStats(sender='S1')[0]
numEmoticonsS2 = conv.stats.getEmoticonsStats(sender='S2')[0]
self.assertEqual(numEmoticons, 6)
self.assertEqual(numEmoticonsS1, 3)
self.assertEqual(numEmoticonsS2, 3)
def test_singleWordCount(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_word_count.txt")
totalCount = conv.stats.getWordCountStats(word='hello')
s1Count = conv.stats.getWordCountStats(word='hello', sender='S1')
s2Count = conv.stats.getWordCountStats(word='hello', sender='S2')
self.assertEqual(totalCount, 3)
self.assertEqual(s1Count, 2)
self.assertEqual(s2Count, 1)
def test_wordsUsedJustByStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_word_count.txt")
wordsUsedBy = conv.stats.getWordsBySender(usedJustBy=True)
wordsSaidJustByS1 = wordsUsedBy['S1']
wordsSaidJustByS2 = wordsUsedBy['S2']
sender = {'S2': {'fine', 'sender2', 'xd', ':d'},
'S1': {'are', 'how', 'you', 'sender1', 'bye'}}
self.assertEqual(wordsSaidJustByS1, sender['S1'])
self.assertEqual(wordsSaidJustByS2, sender['S2'])
def test_intervalStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_interval_conv.txt")
start, end, interval = conv.stats.getIntervalStats()
days = conv.stats.getDaysWithoutMessages()
self.assertEqual(start, Timestamp('2014-01-01 00:00:00'))
self.assertEqual(end, Timestamp('2014.01.07 00:10:00'))
self.assertEqual(interval, datetime.timedelta(6, 600))
self.assertTrue(set(days) == set(['2014.01.02', '2014.01.06']))
def test_singleWordCount(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_word_count.txt")
stats = conv.stats.generateStats(IConvStats.STATS_NAME_WORDCOUNT)
totalCount = stats.getWordsCount(['hello']).values[0]
s1Count = stats.getWordsCount(['hello'], 'S1').values[0]
s2Count = stats.getWordsCount(['hello'], 'S2').values[0]
self.assertEqual(totalCount, 3)
self.assertEqual(s1Count, 2)
self.assertEqual(s2Count, 1)
def test_wordsUsedJustByStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_word_count.txt")
wordsUsedBy = conv.stats.getWordsBySender(usedJustBy=True)
wordsSaidJustByS1 = wordsUsedBy['S1']
wordsSaidJustByS2 = wordsUsedBy['S2']
sender = {
'S2': {'fine', 'sender2', 'xd', ':d'},
'S1': {'are', 'how', 'you', 'sender1', 'bye'}
}
self.assertEqual(wordsSaidJustByS1, sender['S1'])
self.assertEqual(wordsSaidJustByS2, sender['S2'])
def test_firstAndLastOccurences(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_words_trends.txt")
stats = WordsCountStats(conv)
stats.loadWordsCount(['date'])
# Present word, total
res = stats.getWordFirstAndLastOccurences('dog')
self.assertEqual(res, ('2015.10.04', '2015.10.04'))
# Present word, sender specified
res = stats.getWordFirstAndLastOccurences('dog', 'S1')
self.assertEqual(res, ('2015.10.04', '2015.10.04'))
# Word not present
res = stats.getWordFirstAndLastOccurences('log', 'S1')
self.assertEqual(res, (-1, -1))
def test_basicLengthStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_basic_conv.txt")
testSet = set(['S1', 'S2'])
self.assertEqual(testSet, conv.senders)
totalNumS1, totalLengthS1, avgLegthS1 = conv.stats.getBasicLengthStats('S1')
self.assertEqual(totalNumS1, 3)
self.assertEqual(totalLengthS1, 110)
self.assertEqual(avgLegthS1, 110/3)
totalNumS2, totalLengthS2, avgLegthS2 = conv.stats.getBasicLengthStats('S2')
self.assertEqual(totalNumS2, 3)
self.assertEqual(totalLengthS2, 56)
self.assertEqual(avgLegthS2, 56/3)
totalNum, totalLength, avgLegth = conv.stats.getBasicLengthStats()
self.assertEqual(totalNum, 6)
self.assertEqual(totalLength, 166)
self.assertEqual(avgLegth, 166/6)
def test_basicLengthStats(self):
conv = self.getConversation(mio.getResourcesPath() +
"\\unittest\\test_basic_conv.txt")
testSet = set(['S1', 'S2'])
self.assertEqual(testSet, conv.senders)
totalNumS1, totalLengthS1, avgLegthS1 = conv.stats.getBasicLengthStats(
'S1')
self.assertEqual(totalNumS1, 3)
self.assertEqual(totalLengthS1, 110)
self.assertEqual(avgLegthS1, 110 / 3)
totalNumS2, totalLengthS2, avgLegthS2 = conv.stats.getBasicLengthStats(
'S2')
self.assertEqual(totalNumS2, 3)
self.assertEqual(totalLengthS2, 56)
self.assertEqual(avgLegthS2, 56 / 3)
totalNum, totalLength, avgLegth = conv.stats.getBasicLengthStats()
self.assertEqual(totalNum, 6)
self.assertEqual(totalLength, 166)
self.assertEqual(avgLegth, 166 / 6)
def testConversationGenerator():
conv = conversationGenerator.generateNewConversation(100, "2014.01.30 06:01:57", "2014.12.30 06:01:57", ["s1", "s2"], 3, 20)
mio.printListToFile(conv, os.path.join(mio.getResourcesPath(), "test.txt"))
def test_delayStats(self):
conv = self.getConversation(mio.getResourcesPath() + "\\unittest\\test_delay_conv.txt")
df = conv.stats.generateStats(IConvStats.STATS_NAME_DELAY)
self.assertEqual(df.loc['S1'].values[0], (np.timedelta64(10, 'h')+np.timedelta64(8, 's')))
self.assertEqual(df.loc['S2'].values[0], np.timedelta64(10, 's'))
def testConversationGenerator():
conv = conversationGenerator.generateNewConversation(
100, "2014.01.30 06:01:57", "2014.12.30 06:01:57", ["s1", "s2"], 3, 20)
mio.printListToFile(conv, os.path.join(mio.getResourcesPath(), "test.txt"))
def setUpClass(cls):
cls.conv = PlottingTestCase.getConversation(mio.getResourcesPath() + PlottingTestCase.TEST_FILE)
def setUpClass(cls):
cls.conv = PlottingTestCase.getConversation(mio.getResourcesPath() +
PlottingTestCase.TEST_FILE)
