Languedoc Changeset - f82e9a5b1c2c

Changeset - f82e9a5b1c2c

Parent rev.

Child rev.

[Not reviewed]

default

0 1 1

Laman - 7 months ago 2024-09-30 22:17:00

added more tests

2 files changed with 48 insertions and 1 deletions:

tests/test_predict.py

tests/test_train.py

0 comments (0 inline, 0 general)

tests/test_predict.py

➞

Show inline comments

 import unittest
 from unittest import TestCase
 from languedoc.predict import preprocess, extract_kgram_counts, extract_ngram_counts, rank_ngram_counts, \
 	extract_ranked_ngrams, Sample, identify
+	extract_ranked_ngrams, Sample, identify, sample_text
 class TestPredict(TestCase):
 	def test_preprocess(self):
 		self.assertEqual(preprocess("abc"), " abc ")
 		self.assertEqual(preprocess("A  b.c"), " a b c ")
 		self.assertEqual(preprocess("1% "), " ")
 		self.assertEqual(preprocess("Глава ĚŠČŘŽ"), " глава ěščřž ")
 	@unittest.skip
 	def test_sample_text(self):
 		pass
 	def test_extract_kgram_counts(self):
 		text = "abbbabb"
 		self.assertEqual(extract_kgram_counts(text, 1), {"a": 2, "b": 5})
 		self.assertEqual(extract_kgram_counts(text, 2), {"ab": 2, "bb": 3, "ba": 1})
 	def test_extract_ngram_counts(self):

tests/test_train.py

➞

Show inline comments

@@ new file 100644 @@
 import unittest
 from unittest import TestCase
 from languedoc.train import merge_ngram_freqs, cross_validate, SampleSet
 class TestTrain(TestCase):
 	def test_merge_ngram_freqs(self):
 		a = {"a": 3, "b": 1, "c": 4}
 		b = {"b": 1, "c": 5, "d": 9}
 		c = merge_ngram_freqs([a, b])
 		self.assertEqual(c, {"a": 3/8/2, "b": (1/8+1/15)/2, "c": (4/8+5/15)/2, "d": 9/15/2})
 		self.assertEqual(sum(c.values()), 1)
 	@unittest.skip
 	def test_crossvalidate(self):
 		pass
 class TestSampleSet(TestCase):
 	def test_add(self):
 		sample_set = SampleSet("xy")
 		self.assertEqual(sample_set.texts, [])
 		self.assertEqual(sample_set.counts, [])
 		sample_set.add("aaab")
 		self.assertEqual(sample_set.texts, ["aaab"])
 		self.assertEqual(sample_set.counts, [{"a": 3, "b": 1, "aa": 2, "ab": 1, "aaa": 1, "aab": 1}])
 	def test_create_model(self):
 		sample_set = SampleSet("xy")
 		sample_set.add("aaab")
 		sample_set.add("aab")
 		res = sample_set.create_model()
 		self.assertEqual(res.language, "xy")
 		self.assertEqual(res.ranked_ngrams, {"a": 0, "aa": 1, "b": 2, "ab": 3, "aab": 4, "aaa": 5})
 	@unittest.skip
 	def test_generate_tests(self):
 		pass

0 comments (0 inline, 0 general)