static StopAnalyzer()
{
{
var stopWords = new System.String[]{"a", "an", "and", "are", "as", "at", "be", "but", "by", "for", "if", "in", "into", "is", "it", "no", "not", "of", "on", "or", "such", "that", "the", "their", "then", "there", "these", "they", "this", "to", "was", "will", "with"};
var stopSet = new CharArraySet(stopWords.Length, false);
stopSet.AddAll(stopWords);
ENGLISH_STOP_WORDS_SET = CharArraySet.UnmodifiableSet(stopSet);
}
}
/// <summary>
/// Sole constructor. </summary>
public SuggestStopFilter(TokenStream input, CharArraySet stopWords)
: base(input)
{
this.stopWords = stopWords;
this.termAtt = AddAttribute<ICharTermAttribute>();
this.posIncAtt = AddAttribute<IPositionIncrementAttribute>();
this.keywordAtt = AddAttribute<IKeywordAttribute>();
this.offsetAtt = AddAttribute<IOffsetAttribute>();
}
public virtual void TestRehash()
{
CharArraySet cas = new CharArraySet(0, true);
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
cas.Add(TEST_STOP_WORDS[i]);
Assert.AreEqual(TEST_STOP_WORDS.Length, cas.Count);
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
Assert.IsTrue(cas.Contains(TEST_STOP_WORDS[i]));
}
public virtual void TestNonZeroOffset()
{
System.String[] words = new System.String[]{"Hello", "World", "this", "is", "a", "test"};
char[] findme = "xthisy".ToCharArray();
CharArraySet set_Renamed = new CharArraySet(10, true);
for (int i = 0; i < words.Length; i++) { set_Renamed.Add(words[i]); }
Assert.IsTrue(set_Renamed.Contains(findme, 1, 4));
Assert.IsTrue(set_Renamed.Contains(new System.String(findme, 1, 4)));
}
public virtual void TestClear()
{
var set = new CharArraySet(10, true);
for (int i = 0; i < TEST_STOP_WORDS.Length; i++) { set.Add(TEST_STOP_WORDS[i]); }
Assert.AreEqual(TEST_STOP_WORDS.Length, set.Count, "Not all words added");
Assert.Throws<NotSupportedException>(set.Clear, "remove is not supported");
Assert.AreEqual(TEST_STOP_WORDS.Length, set.Count, "Not all words added");
}
public virtual void TestObjectContains()
{
CharArraySet set_Renamed = new CharArraySet(10, true);
System.Int32 val = 1;
set_Renamed.Add((System.Object) val);
Assert.IsTrue(set_Renamed.Contains((System.Object) val));
Assert.IsTrue(set_Renamed.Contains((System.Object) 1));
// test unmodifiable
set_Renamed = CharArraySet.UnmodifiableSet(set_Renamed);
Assert.IsTrue(set_Renamed.Contains((System.Object) val));
Assert.IsTrue(set_Renamed.Contains((System.Object) 1));
}
/// <summary> Construct a token stream filtering the given input.
/// If <code>stopWords</code> is an instance of {@link CharArraySet} (true if
/// <code>makeStopSet()</code> was used to construct the set) it will be directly used
/// and <code>ignoreCase</code> will be ignored since <code>CharArraySet</code>
/// directly controls case sensitivity.
/// <p/>
/// If <code>stopWords</code> is not an instance of {@link CharArraySet},
/// a new CharArraySet will be constructed and <code>ignoreCase</code> will be
/// used to specify the case sensitivity of that set.
///
/// </summary>
/// <param name="input">
/// </param>
/// <param name="stopWords">The set of Stop Words.
/// </param>
/// <param name="ignoreCase">-Ignore case when stopping.
/// </param>
public StopFilter(TokenStream input, System.Collections.Hashtable stopWords, bool ignoreCase) : base(input)
{
if (stopWords is CharArraySet)
{
this.stopWords = (CharArraySet) stopWords;
}
else
{
this.stopWords = new CharArraySet(stopWords.Count, ignoreCase);
foreach (System.String sw in stopWords.Values)
{
this.stopWords.Add(sw);
}
}
}
/// <summary> Construct a token stream filtering the given input.
/// If <c>stopWords</c> is an instance of <see cref="CharArraySet" /> (true if
/// <c>makeStopSet()</c> was used to construct the set) it will be directly used
/// and <c>ignoreCase</c> will be ignored since <c>CharArraySet</c>
/// directly controls case sensitivity.
/// <p/>
/// If <c>stopWords</c> is not an instance of <see cref="CharArraySet" />,
/// a new CharArraySet will be constructed and <c>ignoreCase</c> will be
/// used to specify the case sensitivity of that set.
/// </summary>
/// <param name="enablePositionIncrements">true if token positions should record the removed stop words</param>
/// <param name="input">Input TokenStream</param>
/// <param name="stopWords">A Set of strings or strings or char[] or any other ToString()-able set representing the stopwords</param>
/// <param name="ignoreCase">if true, all words are lower cased first</param>
public StopFilter(bool enablePositionIncrements, TokenStream input, ISet<string> stopWords, bool ignoreCase)
: base(input)
{
if (stopWords is CharArraySet)
{
this.stopWords = (CharArraySet) stopWords;
}
else
{
this.stopWords = new CharArraySet(stopWords.Count, ignoreCase);
this.stopWords.AddAll(stopWords);
}
this.enablePositionIncrements = enablePositionIncrements;
termAtt = AddAttribute<ITermAttribute>();
posIncrAtt = AddAttribute<IPositionIncrementAttribute>();
}
public virtual void TestClear()
{
CharArraySet set_Renamed = new CharArraySet(10, true);
for (int i = 0; i < TEST_STOP_WORDS.Length; i++) { set_Renamed.Add(TEST_STOP_WORDS[i]); }
Assert.AreEqual(TEST_STOP_WORDS.Length, set_Renamed.Count, "Not all words added");
try
{
set_Renamed.Clear();
Assert.Fail("remove is not supported");
}
catch (System.NotSupportedException e)
{
// expected
Assert.AreEqual(TEST_STOP_WORDS.Length, set_Renamed.Count, "Not all words added");
}
}
public virtual void TestUnmodifiableSet()
{
CharArraySet set_Renamed = new CharArraySet(10, true);
set_Renamed.AddAll(new System.Collections.ArrayList(TEST_STOP_WORDS));
int size = set_Renamed.Count;
set_Renamed = CharArraySet.UnmodifiableSet(set_Renamed);
Assert.AreEqual(size, set_Renamed.Count, "Set size changed due to UnmodifiableSet call");
try
{
CharArraySet.UnmodifiableSet(null);
Assert.Fail("can not make null unmodifiable");
}
catch (System.NullReferenceException e)
{
// expected
}
}
/// <summary> Constructs a filter which removes words from the input
/// TokenStream that are named in the array of words.
/// </summary>
public StopFilter(TokenStream in_Renamed, System.String[] stopWords, bool ignoreCase) : base(in_Renamed)
{
this.stopWords = (CharArraySet) MakeStopSet(stopWords, ignoreCase);
}
static DefaultSetHolder()
{
try
{
DEFAULT_SET = WordlistLoader.GetSnowballWordSet(
IOUtils.GetDecodingReader(typeof(SnowballFilter), typeof(SnowballFilter).Namespace + "." + DEFAULT_STOPWORD_FILE, Encoding.UTF8),
#pragma warning disable 612, 618
LuceneVersion.LUCENE_CURRENT);
#pragma warning restore 612, 618
}
catch (IOException)
{
// default set should always be present as it is part of the
// distribution (JAR)
throw new Exception("Unable to load default stopword set");
}
}
/// <summary>
/// Builds an analyzer with the given stop words
/// </summary>
/// <param name="matchVersion">
/// lucene compatibility version </param>
/// <param name="stopwords">
/// a stopword set </param>
/// <param name="stemExclusionSet">
/// a stemming exclusion set </param>
public GermanAnalyzer(LuceneVersion matchVersion, CharArraySet stopwords, CharArraySet stemExclusionSet)
: base(matchVersion, stopwords)
{
exclusionSet = CharArraySet.UnmodifiableSet(CharArraySet.Copy(matchVersion, stemExclusionSet));
}
/// <summary>
/// Builds an analyzer with the given stop words
/// </summary>
/// <param name="matchVersion">
/// lucene compatibility version </param>
/// <param name="stopwords">
/// a stopword set </param>
public GermanAnalyzer(LuceneVersion matchVersion, CharArraySet stopwords)
: this(matchVersion, stopwords, CharArraySet.EMPTY_SET)
{
}
/// <summary> Constructs a filter which removes words from the input
/// TokenStream that are named in the array of words.
/// </summary>
/// <param name="enablePositionIncrements">true if token positions should record the removed stop words
/// </param>
/// <param name="in">input TokenStream
/// </param>
/// <param name="stopWords">array of stop words
/// </param>
/// <param name="ignoreCase">true if case is ignored
/// </param>
/// <deprecated> Use {@link #StopFilter(boolean, TokenStream, Set, boolean)} instead.
/// </deprecated>
public StopFilter(bool enablePositionIncrements, TokenStream in_Renamed, System.String[] stopWords, bool ignoreCase)
: base(in_Renamed)
{
this.stopWords = (CharArraySet) MakeStopSet(stopWords, ignoreCase);
this.enablePositionIncrements = enablePositionIncrements;
Init();
}
public virtual void TestUnmodifiableSet()
{
CharArraySet set_Renamed = new CharArraySet(10, true);
set_Renamed.AddAll(new List<string>(TEST_STOP_WORDS));
int size = set_Renamed.Count;
set_Renamed = CharArraySet.UnmodifiableSet(set_Renamed);
Assert.AreEqual(size, set_Renamed.Count, "Set size changed due to UnmodifiableSet call");
Assert.Throws<ArgumentNullException>(() => CharArraySet.UnmodifiableSet(null), "can not make null unmodifiable");
}
protected internal CharArraySetEnumerator(CharArraySet creator)
{
_Creator = creator;
}
private void InitBlock(CharArraySet enclosingInstance)
{
this.enclosingInstance = enclosingInstance;
}
/// <summary>
/// Sole constructor. </summary>
public SuggestStopFilter(TokenStream input, CharArraySet stopWords)
: base(input)
{
this.stopWords = stopWords;
}
/// <summary> </summary>
/// <param name="stopWords">A List of Strings or char[] or any other toString()-able list representing the stopwords </param>
/// <param name="ignoreCase">if true, all words are lower cased first</param>
/// <returns>A Set (<see cref="CharArraySet"/>)containing the words</returns>
public static ISet<string> MakeStopSet(IList<object> stopWords, bool ignoreCase)
{
var stopSet = new CharArraySet(stopWords.Count, ignoreCase);
foreach(var word in stopWords)
stopSet.Add(word.ToString());
return stopSet;
}
/// <summary></summary>
/// <param name="stopWords">An array of stopwords</param>
/// <param name="ignoreCase">If true, all words are lower cased first.</param>
/// <returns> a Set containing the words</returns>
public static ISet<string> MakeStopSet(string[] stopWords, bool ignoreCase)
{
var stopSet = new CharArraySet(stopWords.Length, ignoreCase);
stopSet.AddAll(stopWords);
return stopSet;
}
private static ISet<string> GetStopWords()
{
int portalId;
string cultureCode;
var searchDoc = Thread.GetData(Thread.GetNamedDataSlot(Constants.TlsSearchInfo)) as SearchDocument;
if (searchDoc == null)
{
portalId = 0; // default
cultureCode = Thread.CurrentThread.CurrentCulture.Name;
}
else
{
portalId = searchDoc.PortalId;
cultureCode = searchDoc.CultureCode;
if (string.IsNullOrEmpty(cultureCode))
{
var portalInfo = PortalController.Instance.GetPortal(portalId);
if (portalInfo != null)
cultureCode = portalInfo.DefaultLanguage;
}
}
var stops = StopAnalyzer.ENGLISH_STOP_WORDS_SET;
var searchStopWords = SearchHelper.Instance.GetSearchStopWords(portalId, cultureCode);
if (searchStopWords != null && !string.IsNullOrEmpty(searchStopWords.StopWords))
{
//TODO Use cache from InternalSearchController
var strArray = searchStopWords.StopWords.Split(',');
var set = new CharArraySet(strArray.Length, false);
set.AddAll(strArray);
stops = CharArraySet.UnmodifiableSet(set);
}
return stops;
}
/// <summary> </summary>
/// <param name="stopWords">
/// </param>
/// <param name="ignoreCase">If true, all words are lower cased first.
/// </param>
/// <returns> a Set containing the words
/// </returns>
public static System.Collections.Hashtable MakeStopSet(System.String[] stopWords, bool ignoreCase)
{
CharArraySet stopSet = new CharArraySet(stopWords.Length, ignoreCase);
for (int i = 0; i < stopWords.Length; i++)
{
stopSet.Add(stopWords[i]);
}
return stopSet;
}
/// <summary> </summary>
/// <param name="stopWords">A List of Strings representing the stopwords
/// </param>
/// <param name="ignoreCase">if true, all words are lower cased first
/// </param>
/// <returns> A Set containing the words
/// </returns>
public static System.Collections.Hashtable MakeStopSet(System.Collections.IList stopWords, bool ignoreCase)
{
CharArraySet stopSet = new CharArraySet(stopWords.Count, ignoreCase);
stopSet.AddAll(stopWords);
return stopSet;
}
public ChineseFilter(TokenStream _in)
: base(_in)
{
stopTable = new CharArraySet((IEnumerable<string>)STOP_WORDS, false);
termAtt = AddAttribute<ITermAttribute>();
}
public virtual void TestModifyOnUnmodifiable()
{
//System.Diagnostics.Debugger.Break();
CharArraySet set = new CharArraySet(10, true);
set.AddAll(TEST_STOP_WORDS);
int size = set.Count;
set = CharArraySet.UnmodifiableSet(set);
Assert.AreEqual(size, set.Count, "Set size changed due to UnmodifiableSet call");
System.String NOT_IN_SET = "SirGallahad";
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String already exists in set");
Assert.Throws<NotSupportedException>(() => set.Add(NOT_IN_SET.ToCharArray()), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.Add(NOT_IN_SET), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.Add(new System.Text.StringBuilder(NOT_IN_SET)), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.Clear(), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Changed unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.Add((object)NOT_IN_SET), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.RemoveAll(new List<string>(TEST_STOP_WORDS)), "Modified unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.RetainAll(new List<string>(new[] { NOT_IN_SET })), "Modified unmodifiable set");
Assert.AreEqual(size, set.Count, "Size of unmodifiable set has changed");
Assert.Throws<NotSupportedException>(() => set.AddAll(new List<string>(new[] { NOT_IN_SET })), "Modified unmodifiable set");
Assert.IsFalse(set.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
{
Assert.IsTrue(set.Contains(TEST_STOP_WORDS[i]));
}
}
public virtual void TestModifyOnUnmodifiable()
{
//System.Diagnostics.Debugger.Break();
CharArraySet set_Renamed = new CharArraySet(10, true);
set_Renamed.AddAll(TEST_STOP_WORDS);
int size = set_Renamed.Count;
set_Renamed = CharArraySet.UnmodifiableSet(set_Renamed);
Assert.AreEqual(size, set_Renamed.Count, "Set size changed due to UnmodifiableSet call");
System.String NOT_IN_SET = "SirGallahad";
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String already exists in set");
try
{
set_Renamed.Add(NOT_IN_SET.ToCharArray());
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.Add(NOT_IN_SET);
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.Add(new System.Text.StringBuilder(NOT_IN_SET));
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.Clear();
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Changed unmodifiable set");
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.Add((System.Object) NOT_IN_SET);
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.RemoveAll(new System.Collections.ArrayList(TEST_STOP_WORDS));
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.RetainAll(new System.Collections.ArrayList(new System.String[] { NOT_IN_SET }));
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.AreEqual(size, set_Renamed.Count, "Size of unmodifiable set has changed");
}
try
{
set_Renamed.AddAll(new System.Collections.ArrayList(new System.String[] { NOT_IN_SET }));
Assert.Fail("Modified unmodifiable set");
}
catch (System.NotSupportedException e)
{
// expected
Assert.IsFalse(set_Renamed.Contains(NOT_IN_SET), "Test String has been added to unmodifiable set");
}
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
{
Assert.IsTrue(set_Renamed.Contains(TEST_STOP_WORDS[i]));
}
}
/// <summary> Construct a token stream filtering the given input.
/// If <code>stopWords</code> is an instance of {@link CharArraySet} (true if
/// <code>makeStopSet()</code> was used to construct the set) it will be directly used
/// and <code>ignoreCase</code> will be ignored since <code>CharArraySet</code>
/// directly controls case sensitivity.
/// <p/>
/// If <code>stopWords</code> is not an instance of {@link CharArraySet},
/// a new CharArraySet will be constructed and <code>ignoreCase</code> will be
/// used to specify the case sensitivity of that set.
///
/// </summary>
/// <param name="enablePositionIncrements">true if token positions should record the removed stop words
/// </param>
/// <param name="input">Input TokenStream
/// </param>
/// <param name="stopWords">The set of Stop Words.
/// </param>
/// <param name="ignoreCase">-Ignore case when stopping.
/// </param>
public StopFilter(bool enablePositionIncrements, TokenStream input, System.Collections.Hashtable stopWords, bool ignoreCase)
: base(input)
{
if (stopWords is CharArraySet)
{
this.stopWords = (CharArraySet) stopWords;
}
else
{
this.stopWords = new CharArraySet(stopWords.Count, ignoreCase);
this.stopWords.Add(stopWords);
}
this.enablePositionIncrements = enablePositionIncrements;
Init();
}
/// <summary> Returns an unmodifiable {@link CharArraySet}. This allows to provide
/// unmodifiable views of internal sets for "read-only" use.
///
/// </summary>
/// <param name="set">a set for which the unmodifiable set is returned.
/// </param>
/// <returns> an new unmodifiable {@link CharArraySet}.
/// </returns>
/// <throws> NullPointerException </throws>
/// <summary> if the given set is <code>null</code>.
/// </summary>
public static CharArraySet UnmodifiableSet(CharArraySet set_Renamed)
{
if (set_Renamed == null)
throw new System.NullReferenceException("Given set is null");
/*
* Instead of delegating calls to the given set copy the low-level values to
* the unmodifiable Subclass
*/
return new UnmodifiableCharArraySet(set_Renamed.entries, set_Renamed.ignoreCase, set_Renamed.count);
}
/// <summary> </summary>
/// <param name="stopWords">An array of stopwords
/// </param>
/// <param name="ignoreCase">If true, all words are lower cased first.
/// </param>
/// <returns> a Set containing the words
/// </returns>
public static System.Collections.Hashtable MakeStopSet(System.String[] stopWords, bool ignoreCase)
{
CharArraySet stopSet = new CharArraySet(stopWords.Length, ignoreCase);
stopSet.AddAll(new System.Collections.ArrayList(stopWords));
return stopSet;
}
internal CharArraySetIterator(CharArraySet enclosingInstance)
{
InitBlock(enclosingInstance);
GoNext();
}
public StopFilter(bool enablePositionIncrements, TokenStream in_Renamed, string[] stopWords, bool ignoreCase) : base(in_Renamed)
{
this.stopWords = (CharArraySet)MakeStopSet(stopWords, ignoreCase);
this.enablePositionIncrements = enablePositionIncrements;
Init();
}
