jben/parsers/tests/edict.py

   1 # -*- coding: utf-8 -*-
   2
   3 import unittest, time
   4 from parsers import edict
   5
   6 SRC_NAME = "edict"
   7 SRC_DIR = "../dicts"
   8
   9 SRC_NAME = "/".join((SRC_DIR, SRC_NAME))
  10
  11 class EdictTest(unittest.TestCase):
  12
  13     def setUp(self):
  14         self.parser = edict.EdictParser(SRC_NAME)
  15
  16     def test_japanese_search(self):
  17         """EDICT: Search for Japanese word/phrase"""
  18         query = u"日本"
  19         l = [entry for entry in self.parser.search(query)]
  20         self.assertTrue(len(l) > 0)
  21
  22     def test_native_search(self):
  23         """EDICT: Search for non-Japanese word/phrase"""
  24         query = u"Japan"
  25         l = [entry for entry in self.parser.search(query)]
  26         self.assertTrue(len(l) > 0)
  27
  28     def test_unparsed(self):
  29         """EDICT: Check for unhandled EDICT fields"""
  30         l = [k for k in self.parser.search(u"")]
  31         unparsed = set()
  32         for key, entry in self.parser.cache.iteritems():
  33             for item in entry.unparsed: unparsed.add(item)
  34         if unparsed:
  35             #self.fail(u"Unhandled fields found: %s" % u", ".join(unparsed))
  36             print u"\n\tWARNING: Unhandled fields found: %s" \
  37                   % u", ".join(unparsed)
  38
  39     def test_caching(self):
  40         """EDICT: Check that caching is working"""
  41         self.assertFalse(self.parser.cache)
  42         t = time.time()
  43         self.test_japanese_search()
  44         first_t = time.time() - t
  45
  46         self.assertTrue(self.parser.cache)
  47         t = time.time()
  48         self.test_japanese_search()
  49         second_t = time.time() - t
  50
  51         print "\n\tFirst query time:  %f" % first_t
  52         print "\tSecond query time: %f" % second_t
  53         self.assertTrue(second_t <= first_t)
  54
  55     def test_no_cache(self):
  56         """EDICT: Check that parser works without caching."""
  57         self.parser = edict.EdictParser(SRC_NAME, use_cache=False)
  58
  59         self.assertFalse(self.parser.cache)
  60         t = time.time()
  61         self.test_japanese_search()
  62         first_t = time.time() - t
  63
  64         self.assertFalse(self.parser.cache)
  65         t = time.time()
  66         self.test_japanese_search()
  67         second_t = time.time() - t
  68
  69         print "\n\tFirst query time:  %f" % first_t
  70         print "\tSecond query time: %f" % second_t
  71
  72     def tearDown(self):
  73         self.parser = None
  74
  75 if __name__ == "__main__":
  76     unittest.main()