예제 #1
0
 def testThreeElementPipeline(self):
     lexicon = Lexicon(
         Splitter(),
         StopWordPipelineElement({"and": 1}),
         StupidPipelineElement("dogs", "fish"),
         WackyReversePipelineElement("fish"),
     )
     wids = lexicon.sourceToWordIds("cats and dogs")
     wids = lexicon.termToWordIds("hsif")
     self.assertEqual(wids, [2])
예제 #2
0
 def testSplitterAdaptorNofold(self):
     lexicon = Lexicon(Splitter())
     wids = lexicon.sourceToWordIds("CATS and dogs")
     wids = lexicon.termToWordIds("cats and dogs")
     self.assertEqual(wids, [0, 2, 3])
예제 #3
0
 def testOnePipelineElement(self):
     lexicon = Lexicon(Splitter(), StupidPipelineElement("dogs", "fish"))
     wids = lexicon.sourceToWordIds("cats and dogs")
     wids = lexicon.termToWordIds("fish")
     self.assertEqual(wids, [3])
예제 #4
0
 def testMissingTermToWordIds(self):
     lexicon = Lexicon(Splitter())
     wids = lexicon.sourceToWordIds("cats and dogs")
     wids = lexicon.termToWordIds("boxes")
     self.assertEqual(wids, [0])
예제 #5
0
 def testTermToWordIds(self):
     lexicon = Lexicon(Splitter())
     wids = lexicon.sourceToWordIds("cats and dogs")
     wids = lexicon.termToWordIds("dogs")
     self.assertEqual(wids, [3])