public virtual void TestCopyJDKSet()
{
ISet <string> set = new JCG.HashSet <string>();
IList <string> stopwords = TEST_STOP_WORDS;
IList <string> stopwordsUpper = new List <string>();
foreach (string @string in stopwords)
{
stopwordsUpper.Add(@string.ToUpperInvariant());
}
set.addAll(TEST_STOP_WORDS);
CharArraySet copy = CharArraySet.Copy(TEST_VERSION_CURRENT, set);
assertEquals(set.Count, copy.size());
assertEquals(set.Count, copy.size());
assertTrue(copy.containsAll(stopwords));
foreach (string @string in stopwordsUpper)
{
assertFalse(copy.contains(@string));
}
IList <string> newWords = new List <string>();
foreach (string @string in stopwords)
{
newWords.Add(@string + "_1");
}
copy.addAll(newWords);
assertTrue(copy.containsAll(stopwords));
assertTrue(copy.containsAll(newWords));
// new added terms are not in the source set
foreach (string @string in newWords)
{
assertFalse(set.Contains(@string));
}
}
static TestRandomChains()
{
try
{
brokenConstructors[typeof(LimitTokenCountFilter).GetConstructor(new Type[] { typeof(TokenStream), typeof(int) })] = ALWAYS;
brokenConstructors[typeof(LimitTokenCountFilter).GetConstructor(new Type[] { typeof(TokenStream), typeof(int), typeof(bool) })] = new PredicateAnonymousInnerClassHelper2();
brokenConstructors[typeof(LimitTokenPositionFilter).GetConstructor(new Type[] { typeof(TokenStream), typeof(int) })] = ALWAYS;
brokenConstructors[typeof(LimitTokenPositionFilter).GetConstructor(new Type[] { typeof(TokenStream), typeof(int), typeof(bool) })] = new PredicateAnonymousInnerClassHelper3();
foreach (Type c in new Type[] {
// TODO: can we promote some of these to be only
// offsets offenders?
// doesn't actual reset itself:
typeof(CachingTokenFilter),
// Not broken, simulates brokenness:
typeof(CrankyTokenFilter),
// Not broken: we forcefully add this, so we shouldn't
// also randomly pick it:
typeof(ValidatingTokenFilter)
})
{
foreach (ConstructorInfo ctor in c.GetConstructors())
{
brokenConstructors[ctor] = ALWAYS;
}
}
}
catch (Exception e)
{
throw new Exception(e.Message, e);
}
try
{
foreach (Type c in new Type[] {
typeof(ReversePathHierarchyTokenizer),
typeof(PathHierarchyTokenizer),
// TODO: it seems to mess up offsets!?
typeof(WikipediaTokenizer),
// TODO: doesn't handle graph inputs
typeof(CJKBigramFilter),
// TODO: doesn't handle graph inputs (or even look at positionIncrement)
typeof(HyphenatedWordsFilter),
// TODO: LUCENE-4983
typeof(CommonGramsFilter),
// TODO: doesn't handle graph inputs
typeof(CommonGramsQueryFilter),
// TODO: probably doesnt handle graph inputs, too afraid to try
typeof(WordDelimiterFilter)
})
{
foreach (ConstructorInfo ctor in c.GetConstructors())
{
brokenOffsetsConstructors[ctor] = ALWAYS;
}
}
}
catch (Exception e)
{
throw new Exception(e.Message, e);
}
allowedTokenizerArgs = new JCG.HashSet <Type>(IdentityEqualityComparer <Type> .Default);
allowedTokenizerArgs.addAll(argProducers.Keys);
allowedTokenizerArgs.Add(typeof(TextReader));
allowedTokenizerArgs.Add(typeof(AttributeSource.AttributeFactory));
allowedTokenizerArgs.Add(typeof(AttributeSource));
allowedTokenFilterArgs = new JCG.HashSet <Type>(IdentityEqualityComparer <Type> .Default);
allowedTokenFilterArgs.addAll(argProducers.Keys);
allowedTokenFilterArgs.Add(typeof(TokenStream));
// TODO: fix this one, thats broken:
allowedTokenFilterArgs.Add(typeof(CommonGramsFilter));
allowedCharFilterArgs = new JCG.HashSet <Type>(IdentityEqualityComparer <Type> .Default);
allowedCharFilterArgs.addAll(argProducers.Keys);
allowedCharFilterArgs.Add(typeof(TextReader));
}
