def main(cls, argv):
engine = WordNetSynonymEngine(argv[1])
text = "The quick brown fox jumps over the lazy dogs"
AnalyzerUtils.displayTokensWithPositions(SynonymAnalyzer(engine), text)
text = "\"Oh, we get both kinds - country AND western!\" - B.B."
AnalyzerUtils.displayTokensWithPositions(SynonymAnalyzer(engine), text)
def main(cls):
query = QueryParser(Version.LUCENE_CURRENT, "content",
cls.synonymAnalyzer).parse('"fox jumps"')
print "\"fox jumps\" parses to ", query.toString("content")
print "From AnalyzerUtils.tokensFromAnalysis: "
AnalyzerUtils.displayTokens(cls.synonymAnalyzer, "\"fox jumps\"")
print ''
def main(cls):
query = QueryParser(Version.LUCENE_CURRENT, "content",
cls.synonymAnalyzer).parse('"fox jumps"')
print "\"fox jumps\" parses to ", query.toString("content")
print "From AnalyzerUtils.tokensFromAnalysis: "
AnalyzerUtils.displayTokens(cls.synonymAnalyzer, "\"fox jumps\"")
print ''
def main(cls, argv):
engine = WordNetSynonymEngine(argv[1])
text = "The quick brown fox jumps over the lazy dogs"
AnalyzerUtils.displayTokensWithPositions(SynonymAnalyzer(engine), text)
text = "\"Oh, we get both kinds - country AND western!\" - B.B."
AnalyzerUtils.displayTokensWithPositions(SynonymAnalyzer(engine), text)
def analyze(cls, text):
print 'Analyzing "%s"' %(text)
for analyzer in cls.analyzers:
name = type(analyzer).__name__
print " %s:" %(name),
AnalyzerUtils.displayTokens(analyzer, text)
print
print
def analyze(cls, text):
print 'Analyzing "%s"' %(text)
for analyzer in cls.analyzers:
name = type(analyzer).__name__
print " %s:" %(name),
AnalyzerUtils.displayTokens(analyzer, text)
print
print
def testHoles(self):
expected = ["one", "enough"]
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one is not enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one is enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one but not enough",
expected)
def testHoles(self):
expected = ["one", "enough"]
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one is not enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one is enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one enough",
expected)
AnalyzerUtils.assertAnalyzesTo(self.stopAnalyzer, "one but not enough",
expected)
def addAliasesToStack(self):
synonyms = self.engine.getSynonyms(self.termAttr.term())
if synonyms is None:
return
current = self.captureState()
for synonym in synonyms:
self.save.restoreState(current)
AnalyzerUtils.setTerm(self.save, synonym)
AnalyzerUtils.setType(self.save, self.TOKEN_TYPE_SYNONYM)
AnalyzerUtils.setPositionIncrement(self.save, 0)
self.synonymStack.append(self.save.captureState())
def addAliasesToStack(self):
synonyms = self.engine.getSynonyms(self.termAttr.term())
if synonyms is None:
return
current = self.captureState()
for synonym in synonyms:
self.save.restoreState(current)
AnalyzerUtils.setTerm(self.save, synonym)
AnalyzerUtils.setType(self.save, self.TOKEN_TYPE_SYNONYM)
AnalyzerUtils.setPositionIncrement(self.save, 0)
self.synonymStack.append(self.save.captureState())
import os, sys, lucene lucene.initVM() sys.path.append(os.path.dirname(os.path.abspath(sys.argv[0]))) from lia.analysis.AnalyzerUtils import AnalyzerUtils AnalyzerUtils.main(sys.argv)
def main(cls):
text = "The quick brown fox jumps over the lazy dogs"
AnalyzerUtils.displayTokensWithPositions(cls.porterAnalyzer, text)
print ''
def testSpanish(self):
analyzer = SnowballAnalyzer(Version.LUCENE_CURRENT, "Spanish")
AnalyzerUtils.assertAnalyzesTo(analyzer, "algoritmos", ["algoritm"])
def testEnglish(self):
analyzer = SnowballAnalyzer(Version.LUCENE_CURRENT, "English")
AnalyzerUtils.assertAnalyzesTo(analyzer, "stemming algorithms",
["stem", "algorithm"])
def testEnglish(self):
analyzer = SnowballAnalyzer(Version.LUCENE_CURRENT, "English")
AnalyzerUtils.assertAnalyzesTo(analyzer, "stemming algorithms",
["stem", "algorithm"])
def testSpanish(self):
analyzer = SnowballAnalyzer(Version.LUCENE_CURRENT, "Spanish")
AnalyzerUtils.assertAnalyzesTo(analyzer, "algoritmos", ["algoritm"])
def main(cls):
AnalyzerUtils.displayTokens(StopAnalyzerFlawed(),
"The quick brown...")
def testStopAnalyzerFlawed(self):
AnalyzerUtils.assertAnalyzesTo(StopAnalyzerFlawed(),
"The quick brown...",
["the", "quick", "brown"])
def testStopAnalyzer2(self):
AnalyzerUtils.assertAnalyzesTo(StopAnalyzer2(),
"The quick brown...",
["quick", "brown"])
import os, sys, lucene lucene.initVM() sys.path.append(os.path.dirname(os.path.abspath(sys.argv[0]))) from lia.analysis.AnalyzerUtils import AnalyzerUtils AnalyzerUtils.main(sys.argv)
