jbparse/tests/kanjidic2.py

   1 # -*- coding: utf-8 -*-
   2
   3 from __future__ import absolute_import
   4
   5 import unittest, time
   6 from jbparse import kanjidic2
   7
   8 SRC_NAME = "kanjidic2.xml"
   9 SRC_DIR = "../../dicts"
  10
  11 SRC_NAME = "/".join((SRC_DIR, SRC_NAME))
  12
  13 class Kanjidic2Test(unittest.TestCase):
  14
  15     def setUp(self):
  16         self.parser = kanjidic2.Kanjidic2Parser(SRC_NAME)
  17
  18     def test_single_kanji_search(self):
  19         """KANJIDIC2: Search for single kanji"""
  20         query = u"食"
  21         l = [entry for entry in self.parser.search(query)]
  22         self.assertEqual(len(l), 1)
  23         self.assertEqual(query, l[0].literal)
  24
  25     def test_multi_kanji_search(self):
  26         """KANJIDIC2: Search for multiple kanji at once"""
  27         query = u"上下"
  28         l = [entry for entry in self.parser.search(query)]
  29         self.assertEqual(len(l), 2)
  30         for char in query:
  31             self.assertTrue(char in [entry.literal for entry in l])
  32
  33     def test_mixed_search(self):
  34         """KANJIDIC2: Search query with kanji and non-kanji characters"""
  35         query = u"天気はいいから、散歩しましょう。  Right?"
  36         l = [entry for entry in self.parser.search(query)]
  37         self.assertEqual(len(l), 4)
  38         for char in u"天気散歩":
  39             self.assertTrue(char in [entry.literal for entry in l])
  40
  41     def test_caching(self):
  42         """KANJIDIC2: Check that caching is working"""
  43         self.assertFalse(self.parser.cache)
  44         t = time.time()
  45         self.test_single_kanji_search()
  46         first_t = time.time() - t
  47
  48         self.assertTrue(self.parser.cache)
  49         t = time.time()
  50         self.test_single_kanji_search()
  51         second_t = time.time() - t
  52
  53         print "\n\tFirst query time:  %f" % first_t
  54         print "\tSecond query time: %f" % second_t
  55         self.assertTrue(second_t <= first_t)
  56
  57     def test_no_cache(self):
  58         """KANJIDIC2: Check that parser works without caching."""
  59         self.parser = kanjidic2.Kanjidic2Parser(SRC_NAME, use_cache=False)
  60
  61         self.assertFalse(self.parser.cache)
  62         t = time.time()
  63         self.test_single_kanji_search()
  64         first_t = time.time() - t
  65
  66         self.assertFalse(self.parser.cache)
  67         t = time.time()
  68         self.test_single_kanji_search()
  69         second_t = time.time() - t
  70
  71         print "\n\tFirst query time:  %f" % first_t
  72         print "\tSecond query time: %f" % second_t
  73
  74     def tearDown(self):
  75         self.parser = None
  76
  77 if __name__ == "__main__":
  78     unittest.main()