def calcStringStats(self, values, sink, feedback):
stat = QgsStringStatisticalSummary()
total = 50.0 / len(values) if values else 0
current = 0
for cat, v in values.items():
if feedback.isCanceled():
break
feedback.setProgress(int(current * total) + 50)
stat.calculate(v)
f = QgsFeature()
f.setAttributes(list(cat) + [stat.count(),
stat.countDistinct(),
stat.countMissing(),
stat.count() - stat.countMissing(),
stat.min(),
stat.max(),
stat.minLength(),
stat.maxLength(),
stat.meanLength()
])
sink.addFeature(f, QgsFeatureSink.FastInsert)
current += 1
def testStats(self):
# we test twice, once with values added as a list and once using values
# added one-at-a-time
s = QgsStringStatisticalSummary()
self.assertEqual(s.statistics(), QgsStringStatisticalSummary.All)
strings = ['cc', 'aaaa', 'bbbbbbbb', 'aaaa', 'eeee', '', 'eeee', '', 'dddd']
s.calculate(strings)
s2 = QgsStringStatisticalSummary()
for string in strings:
s2.addString(string)
s2.finalize()
self.assertEqual(s.count(), 9)
self.assertEqual(s2.count(), 9)
self.assertEqual(s.countDistinct(), 6)
self.assertEqual(s2.countDistinct(), 6)
self.assertEqual(set(s.distinctValues()), set(['cc', 'aaaa', 'bbbbbbbb', 'eeee', 'dddd', '']))
self.assertEqual(s2.distinctValues(), s.distinctValues())
self.assertEqual(s.countMissing(), 2)
self.assertEqual(s2.countMissing(), 2)
self.assertEqual(s.min(), 'aaaa')
self.assertEqual(s2.min(), 'aaaa')
self.assertEqual(s.max(), 'eeee')
self.assertEqual(s2.max(), 'eeee')
self.assertEqual(s.minLength(), 0)
self.assertEqual(s2.minLength(), 0)
self.assertEqual(s.maxLength(), 8)
self.assertEqual(s2.maxLength(), 8)
self.assertEqual(s.meanLength(), 3.33333333333333333333333)
self.assertEqual(s2.meanLength(), 3.33333333333333333333333)
#extra check for minLength without empty strings
s.calculate(['1111111', '111', '11111'])
self.assertEqual(s.minLength(), 3)
def testIndividualStats(self):
# tests calculation of statistics one at a time, to make sure statistic calculations are not
# dependent on each other
tests = [{'stat': QgsStringStatisticalSummary.Count, 'expected': 9},
{'stat': QgsStringStatisticalSummary.CountDistinct, 'expected': 6},
{'stat': QgsStringStatisticalSummary.CountMissing, 'expected': 2},
{'stat': QgsStringStatisticalSummary.Min, 'expected': 'aaaa'},
{'stat': QgsStringStatisticalSummary.Max, 'expected': 'eeee'},
{'stat': QgsStringStatisticalSummary.MinimumLength, 'expected': 0},
{'stat': QgsStringStatisticalSummary.MaximumLength, 'expected': 8},
]
s = QgsStringStatisticalSummary()
s3 = QgsStringStatisticalSummary()
for t in tests:
# test constructor
s2 = QgsStringStatisticalSummary(t['stat'])
self.assertEqual(s2.statistics(), t['stat'])
s.setStatistics(t['stat'])
s3.setStatistics(t['stat'])
self.assertEqual(s.statistics(), t['stat'])
strings = ['cc', 'aaaa', 'bbbbbbbb', 'aaaa', 'eeee', '', 'eeee', '', 'dddd']
s.calculate(strings)
s3.reset()
for string in strings:
s3.addString(string)
s3.finalize()
self.assertEqual(s.statistic(t['stat']), t['expected'])
self.assertEqual(s3.statistic(t['stat']), t['expected'])
# display name
self.assertTrue(len(QgsStringStatisticalSummary.displayName(t['stat'])) > 0)
def testStats(self):
s = QgsStringStatisticalSummary()
self.assertEqual(s.statistics(), QgsStringStatisticalSummary.All)
s.calculate(['cc', 'aaaa', 'bbbbbbbb', 'aaaa', 'eeee', '', 'eeee', '', 'dddd'])
self.assertEqual(s.count(), 9)
self.assertEqual(s.countDistinct(), 6)
self.assertEqual(set(s.distinctValues()), set(['cc', 'aaaa', 'bbbbbbbb', 'eeee', 'dddd', '']))
self.assertEqual(s.countMissing(), 2)
self.assertEqual(s.min(), 'aaaa')
self.assertEqual(s.max(), 'eeee')
self.assertEqual(s.minLength(), 0)
self.assertEqual(s.maxLength(), 8)
#extra check for minLength without empty strings
s.calculate(['1111111', '111', '11111'])
self.assertEqual(s.minLength(), 3)
def testIndividualStats(self):
# tests calculation of statistics one at a time, to make sure statistic calculations are not
# dependent on each other
tests = [
{"stat": QgsStringStatisticalSummary.Count, "expected": 9},
{"stat": QgsStringStatisticalSummary.CountDistinct, "expected": 6},
{"stat": QgsStringStatisticalSummary.CountMissing, "expected": 2},
{"stat": QgsStringStatisticalSummary.Min, "expected": "aaaa"},
{"stat": QgsStringStatisticalSummary.Max, "expected": "eeee"},
{"stat": QgsStringStatisticalSummary.MinimumLength, "expected": 0},
{"stat": QgsStringStatisticalSummary.MaximumLength, "expected": 8},
{"stat": QgsStringStatisticalSummary.MeanLength, "expected": 3.3333333333333335},
]
s = QgsStringStatisticalSummary()
s3 = QgsStringStatisticalSummary()
for t in tests:
# test constructor
s2 = QgsStringStatisticalSummary(t["stat"])
self.assertEqual(s2.statistics(), t["stat"])
s.setStatistics(t["stat"])
s3.setStatistics(t["stat"])
self.assertEqual(s.statistics(), t["stat"])
strings = ["cc", "aaaa", "bbbbbbbb", "aaaa", "eeee", "", "eeee", "", "dddd"]
s.calculate(strings)
s3.reset()
for string in strings:
s3.addString(string)
s3.finalize()
self.assertEqual(s.statistic(t["stat"]), t["expected"])
self.assertEqual(s3.statistic(t["stat"]), t["expected"])
# display name
self.assertTrue(len(QgsStringStatisticalSummary.displayName(t["stat"])) > 0)
def testIndividualStats(self):
# tests calculation of statistics one at a time, to make sure statistic calculations are not
# dependent on each other
tests = [{'stat': QgsStringStatisticalSummary.Count, 'expected': 9},
{'stat': QgsStringStatisticalSummary.CountDistinct, 'expected': 6},
{'stat': QgsStringStatisticalSummary.CountMissing, 'expected': 2},
{'stat': QgsStringStatisticalSummary.Min, 'expected': 'aaaa'},
{'stat': QgsStringStatisticalSummary.Max, 'expected': 'eeee'},
{'stat': QgsStringStatisticalSummary.MinimumLength, 'expected': 0},
{'stat': QgsStringStatisticalSummary.MaximumLength, 'expected': 8},
{'stat': QgsStringStatisticalSummary.MeanLength, 'expected': 3.3333333333333335},
]
s = QgsStringStatisticalSummary()
s3 = QgsStringStatisticalSummary()
for t in tests:
# test constructor
s2 = QgsStringStatisticalSummary(t['stat'])
self.assertEqual(s2.statistics(), t['stat'])
s.setStatistics(t['stat'])
s3.setStatistics(t['stat'])
self.assertEqual(s.statistics(), t['stat'])
strings = ['cc', 'aaaa', 'bbbbbbbb', 'aaaa', 'eeee', '', 'eeee', '', 'dddd']
s.calculate(strings)
s3.reset()
for string in strings:
s3.addString(string)
s3.finalize()
self.assertEqual(s.statistic(t['stat']), t['expected'])
self.assertEqual(s3.statistic(t['stat']), t['expected'])
# display name
self.assertTrue(len(QgsStringStatisticalSummary.displayName(t['stat'])) > 0)
def testStats(self):
# we test twice, once with values added as a list and once using values
# added one-at-a-time
s = QgsStringStatisticalSummary()
self.assertEqual(s.statistics(), QgsStringStatisticalSummary.All)
strings = [
'cc', 'aaaa', 'bbbbbbbb', 'aaaa', 'eeee', '', 'eeee', '', 'dddd'
]
s.calculate(strings)
s2 = QgsStringStatisticalSummary()
for string in strings:
s2.addString(string)
s2.finalize()
self.assertEqual(s.count(), 9)
self.assertEqual(s2.count(), 9)
self.assertEqual(s.countDistinct(), 6)
self.assertEqual(s2.countDistinct(), 6)
self.assertEqual(set(s.distinctValues()),
set(['cc', 'aaaa', 'bbbbbbbb', 'eeee', 'dddd', '']))
self.assertEqual(s2.distinctValues(), s.distinctValues())
self.assertEqual(s.countMissing(), 2)
self.assertEqual(s2.countMissing(), 2)
self.assertEqual(s.min(), 'aaaa')
self.assertEqual(s2.min(), 'aaaa')
self.assertEqual(s.max(), 'eeee')
self.assertEqual(s2.max(), 'eeee')
self.assertEqual(s.minLength(), 0)
self.assertEqual(s2.minLength(), 0)
self.assertEqual(s.maxLength(), 8)
self.assertEqual(s2.maxLength(), 8)
self.assertEqual(s.meanLength(), 3.33333333333333333333333)
self.assertEqual(s2.meanLength(), 3.33333333333333333333333)
#extra check for minLength without empty strings
s.calculate(['1111111', '111', '11111'])
self.assertEqual(s.minLength(), 3)