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])
def testSplitterAdaptorNofold(self): lexicon = Lexicon(Splitter()) wids = lexicon.sourceToWordIds("CATS and dogs") wids = lexicon.termToWordIds("cats and dogs") self.assertEqual(wids, [0, 2, 3])
def testOnePipelineElement(self): lexicon = Lexicon(Splitter(), StupidPipelineElement("dogs", "fish")) wids = lexicon.sourceToWordIds("cats and dogs") wids = lexicon.termToWordIds("fish") self.assertEqual(wids, [3])
def testMissingTermToWordIds(self): lexicon = Lexicon(Splitter()) wids = lexicon.sourceToWordIds("cats and dogs") wids = lexicon.termToWordIds("boxes") self.assertEqual(wids, [0])
def testTermToWordIds(self): lexicon = Lexicon(Splitter()) wids = lexicon.sourceToWordIds("cats and dogs") wids = lexicon.termToWordIds("dogs") self.assertEqual(wids, [3])