public virtual void TestUnmodifiableSet()
{
var set = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
set.AddAll(TEST_STOP_WORDS);
set.Add(Convert.ToInt32(1));
int size = set.size();
set = CharArraySet.UnmodifiableSet(set);
assertEquals("Set size changed due to unmodifiableSet call", size, set.size());
foreach (var stopword in TEST_STOP_WORDS)
{
assertTrue(set.contains(stopword));
}
assertTrue(set.contains(Convert.ToInt32(1)));
assertTrue(set.contains("1"));
assertTrue(set.contains(new[] { '1' }));
try
{
CharArraySet.UnmodifiableSet(null);
fail("can not make null unmodifiable");
}
catch (System.NullReferenceException)
{
// expected
}
}
public virtual void TestUnmodifiableSet()
{
var set = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
set.UnionWith(TEST_STOP_WORDS);
set.Add(Convert.ToInt32(1));
int size = set.size();
set = CharArraySet.UnmodifiableSet(set);
assertEquals("Set size changed due to unmodifiableSet call", size, set.size());
foreach (var stopword in TEST_STOP_WORDS)
{
assertTrue(set.Contains(stopword));
}
assertTrue(set.Contains(Convert.ToInt32(1)));
assertTrue(set.Contains("1"));
assertTrue(set.Contains(new[] { '1' }));
try
{
CharArraySet.UnmodifiableSet(null);
fail("can not make null unmodifiable");
}
catch (ArgumentNullException) // NOTE: In .NET we throw an ArgumentExcpetion, not a NullReferenceExeption
{
// expected
}
}
/// <summary>
/// Reads stopwords from a stopword list in Snowball format.
/// <para>
/// The snowball format is the following:
/// <list type="bullet">
/// <item>Lines may contain multiple words separated by whitespace.</item>
/// <item>The comment character is the vertical line (|).</item>
/// <item>Lines may contain trailing comments.</item>
/// </list>
/// </para>
/// </summary>
/// <param name="reader"> <see cref="TextReader"/> containing a Snowball stopword list </param>
/// <param name="result"> the <see cref="CharArraySet"/> to fill with the readers words </param>
/// <returns> the given <see cref="CharArraySet"/> with the reader's words </returns>
public static CharArraySet GetSnowballWordSet(TextReader reader, CharArraySet result)
{
try
{
string line = null;
while ((line = reader.ReadLine()) != null)
{
int comment = line.IndexOf('|');
if (comment >= 0)
{
line = line.Substring(0, comment);
}
string[] words = line.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries).Select(w => w.Trim()).ToArray();
foreach (var word in words)
{
if (word.Length > 0)
{
result.Add(word);
}
}
}
}
finally
{
IOUtils.Close(reader);
}
return(result);
}
/// <summary>
/// Reads stopwords from a stopword list in Snowball format.
/// <para>
/// The snowball format is the following:
/// <list type="bullet">
/// <item><description>Lines may contain multiple words separated by whitespace.</description></item>
/// <item><description>The comment character is the vertical line (|).</description></item>
/// <item><description>Lines may contain trailing comments.</description></item>
/// </list>
/// </para>
/// </summary>
/// <param name="reader"> <see cref="TextReader"/> containing a Snowball stopword list </param>
/// <param name="result"> the <see cref="CharArraySet"/> to fill with the readers words </param>
/// <returns> the given <see cref="CharArraySet"/> with the reader's words </returns>
public static CharArraySet GetSnowballWordSet(TextReader reader, CharArraySet result)
{
try
{
string line = null;
while ((line = reader.ReadLine()) != null)
{
int comment = line.IndexOf('|');
if (comment >= 0)
{
line = line.Substring(0, comment);
}
string[] words = WHITESPACE.Split(line).TrimEnd();
foreach (var word in words)
{
if (word.Length > 0)
{
result.Add(word);
}
}
}
}
finally
{
IOUtils.Dispose(reader);
}
return(result);
}
public virtual void TestRehash()
{
CharArraySet cas = new CharArraySet(TEST_VERSION_CURRENT, 0, true);
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
{
cas.Add(TEST_STOP_WORDS[i]);
}
assertEquals(TEST_STOP_WORDS.Length, cas.size());
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
{
assertTrue(cas.Contains(TEST_STOP_WORDS[i]));
}
}
public virtual void TestCopyCharArraySet()
{
CharArraySet setIngoreCase = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
CharArraySet setCaseSensitive = new CharArraySet(TEST_VERSION_CURRENT, 10, false);
IList <string> stopwords = TEST_STOP_WORDS;
IList <string> stopwordsUpper = new List <string>();
foreach (string @string in stopwords)
{
stopwordsUpper.Add(@string.ToUpperInvariant());
}
setIngoreCase.addAll(TEST_STOP_WORDS);
setIngoreCase.Add(Convert.ToInt32(1));
setCaseSensitive.addAll(TEST_STOP_WORDS);
setCaseSensitive.Add(Convert.ToInt32(1));
CharArraySet copy = CharArraySet.Copy(TEST_VERSION_CURRENT, setIngoreCase);
CharArraySet copyCaseSens = CharArraySet.Copy(TEST_VERSION_CURRENT, setCaseSensitive);
assertEquals(setIngoreCase.size(), copy.size());
assertEquals(setCaseSensitive.size(), copy.size());
assertTrue(copy.containsAll(stopwords));
assertTrue(copy.containsAll(stopwordsUpper));
assertTrue(copyCaseSens.containsAll(stopwords));
foreach (string @string in stopwordsUpper)
{
assertFalse(copyCaseSens.contains(@string));
}
// test adding terms to the copy
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(stopwordsUpper));
assertTrue(copy.containsAll(newWords));
// new added terms are not in the source set
foreach (string @string in newWords)
{
assertFalse(setIngoreCase.contains(@string));
assertFalse(setCaseSensitive.contains(@string));
}
}
// LUCENENET TODO: Add .NET overloads that accept a file name? Or at least a FileInfo object as was done in 3.0.3?
/// <summary>
/// Reads lines from a <see cref="TextReader"/> and adds every line as an entry to a <see cref="CharArraySet"/> (omitting
/// leading and trailing whitespace). Every line of the <see cref="TextReader"/> should contain only
/// one word. The words need to be in lowercase if you make use of an
/// <see cref="Analyzer"/> which uses <see cref="Core.LowerCaseFilter"/> (like <see cref="Standard.StandardAnalyzer"/>).
/// </summary>
/// <param name="reader"> <see cref="TextReader"/> containing the wordlist </param>
/// <param name="result"> the <see cref="CharArraySet"/> to fill with the readers words </param>
/// <returns> the given <see cref="CharArraySet"/> with the reader's words </returns>
public static CharArraySet GetWordSet(TextReader reader, CharArraySet result)
{
try
{
string word = null;
while ((word = reader.ReadLine()) != null)
{
result.Add(word.Trim());
}
}
finally
{
IOUtils.Close(reader);
}
return(result);
}
public virtual void TestObjectContains()
{
CharArraySet set = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
int? val = Convert.ToInt32(1);
set.Add(val);
assertTrue(set.Contains(val));
assertTrue(set.Contains(new int?(1))); // another integer
assertTrue(set.Contains("1"));
assertTrue(set.Contains(new char[] { '1' }));
// test unmodifiable
set = CharArraySet.UnmodifiableSet(set);
assertTrue(set.Contains(val));
assertTrue(set.Contains(new int?(1))); // another integer
assertTrue(set.Contains("1"));
assertTrue(set.Contains(new char[] { '1' }));
}
public virtual void TestObjectContains()
{
CharArraySet set = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
J2N.Numerics.Int32 val = J2N.Numerics.Int32.GetInstance(1);
set.Add(val);
assertTrue(set.Contains(val));
assertTrue(set.Contains(J2N.Numerics.Int32.GetInstance(1))); // another integer
assertTrue(set.Contains("1"));
assertTrue(set.Contains(new char[] { '1' }));
// test unmodifiable
set = CharArraySet.UnmodifiableSet(set);
assertTrue(set.Contains(val));
assertTrue(set.Contains(J2N.Numerics.Int32.GetInstance(1))); // another integer
assertTrue(set.Contains("1"));
assertTrue(set.Contains(new char[] { '1' }));
}
/// <summary>
/// Reads lines from a <see cref="TextReader"/> and adds every non-comment line as an entry to a <see cref="CharArraySet"/> (omitting
/// leading and trailing whitespace). Every line of the <see cref="TextReader"/> should contain only
/// one word. The words need to be in lowercase if you make use of an
/// <see cref="Analyzer"/> which uses <see cref="Core.LowerCaseFilter"/> (like <see cref="Standard.StandardAnalyzer"/>).
/// </summary>
/// <param name="reader"> <see cref="TextReader"/> containing the wordlist </param>
/// <param name="comment"> The string representing a comment. </param>
/// <param name="result"> the <see cref="CharArraySet"/> to fill with the readers words </param>
/// <returns> the given <see cref="CharArraySet"/> with the reader's words </returns>
public static CharArraySet GetWordSet(TextReader reader, string comment, CharArraySet result)
{
try
{
string word = null;
while ((word = reader.ReadLine()) != null)
{
if (word.StartsWith(comment, StringComparison.Ordinal) == false)
{
result.Add(word.Trim());
}
}
}
finally
{
IOUtils.Close(reader);
}
return(result);
}
public virtual void TestMethods()
{
CharArrayMap <int?> cm = new CharArrayMap <int?>(TEST_VERSION_CURRENT, 2, false);
Dictionary <string, int?> hm = new Dictionary <string, int?>();
hm["foo"] = 1;
hm["bar"] = 2;
cm.PutAll(hm);
assertEquals(hm.Count, cm.Count);
hm["baz"] = 3;
cm.PutAll(hm);
assertEquals(hm.Count, cm.Count);
// LUCENENET: Need to cast here - no implicit conversion.
CharArraySet cs = cm.Keys as CharArraySet;
int n = 0;
foreach (string o in cs)
{
assertTrue(cm.ContainsKey(o));
char[] co = o.ToCharArray();
assertTrue(cm.ContainsKey(co, 0, co.Length));
n++;
}
assertEquals(hm.Count, n);
assertEquals(hm.Count, cs.Count);
assertEquals(cm.Count, cs.Count);
cs.Clear();
assertEquals(0, cs.Count);
assertEquals(0, cm.Count);
try
{
cs.Add("test");
fail("keySet() allows adding new keys");
}
catch (System.NotSupportedException)
{
// pass
}
cm.PutAll(hm);
assertEquals(hm.Count, cs.Count);
assertEquals(cm.Count, cs.Count);
// LUCENENET: Need to cast here - no implicit conversion
IEnumerator <KeyValuePair <string, int?> > iter1 = (IEnumerator <KeyValuePair <string, int?> >)cm.EntrySet().GetEnumerator();
n = 0;
while (iter1.MoveNext())
{
KeyValuePair <string, int?> entry = iter1.Current;
object key = entry.Key;
int? val = entry.Value;
assertEquals(cm.Get(key), val);
// LUCENENET: Need a cast to get to this method because it is not part of the IEnumerator<T> interface
((CharArrayMap <int?> .EntryIterator)iter1).SetValue(val * 100);
assertEquals(val * 100, (int)cm.Get(key));
n++;
}
assertEquals(hm.Count, n);
cm.Clear();
cm.PutAll(hm);
assertEquals(cm.size(), n);
CharArrayMap <int?> .EntryIterator iter2 = cm.EntrySet().GetEnumerator() as CharArrayMap <int?> .EntryIterator;
n = 0;
while (iter2.MoveNext())
{
var keyc = iter2.Current.Key;
int?val = iter2.Current.Value;
assertEquals(hm[keyc], val);
iter2.SetValue(val * 100);
assertEquals(val * 100, (int)cm.Get(keyc));
n++;
}
assertEquals(hm.Count, n);
cm.EntrySet().Clear();
assertEquals(0, cm.size());
assertEquals(0, cm.EntrySet().size());
assertTrue(cm.Count == 0);
}
public virtual void TestModifyOnUnmodifiable()
{
CharArraySet set = new CharArraySet(TEST_VERSION_CURRENT, 10, true);
set.UnionWith(TEST_STOP_WORDS);
int size = set.size();
set = CharArraySet.UnmodifiableSet(set);
assertEquals("Set size changed due to unmodifiableSet call", size, set.size());
string NOT_IN_SET = "SirGallahad";
assertFalse("Test String already exists in set", set.Contains(NOT_IN_SET));
try
{
set.Add(NOT_IN_SET.ToCharArray());
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
try
{
set.add(NOT_IN_SET);
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
try
{
set.Add(new StringBuilder(NOT_IN_SET));
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
try
{
set.clear();
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Changed unmodifiable set", set.contains(NOT_IN_SET));
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
try
{
set.add(NOT_IN_SET);
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
// NOTE: This results in a StackOverflow exception. Since this is not a public member of CharArraySet,
// but an extension method for the test fixture (which apparently has a bug), this test is non-critical
//// This test was changed in 3.1, as a contains() call on the given Collection using the "correct" iterator's
//// current key (now a char[]) on a Set<String> would not hit any element of the CAS and therefor never call
//// remove() on the iterator
//try
//{
// set.removeAll(new CharArraySet(TEST_VERSION_CURRENT, TEST_STOP_WORDS, true));
// fail("Modified unmodifiable set");
//}
//catch (Exception e) when (e.IsUnsupportedOperationException())
//{
// // expected
// assertEquals("Size of unmodifiable set has changed", size, set.size());
//}
#region LUCENENET Added for better .NET support
// This test was added for .NET to check the Remove method, since the extension method
// above fails to execute.
try
{
#pragma warning disable 612, 618
set.Remove(TEST_STOP_WORDS[0]);
#pragma warning restore 612, 618
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
// LUCENENET Specific - added to test .NETified UnionWith method
try
{
set.UnionWith(new[] { NOT_IN_SET });
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
}
#endregion LUCENENET Added for better .NET support
try
{
set.retainAll(new CharArraySet(TEST_VERSION_CURRENT, new [] { NOT_IN_SET }, true));
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertEquals("Size of unmodifiable set has changed", size, set.size());
}
try
{
set.addAll(new[] { NOT_IN_SET });
fail("Modified unmodifiable set");
}
catch (Exception e) when(e.IsUnsupportedOperationException())
{
// expected
assertFalse("Test String has been added to unmodifiable set", set.contains(NOT_IN_SET));
}
for (int i = 0; i < TEST_STOP_WORDS.Length; i++)
{
assertTrue(set.contains(TEST_STOP_WORDS[i]));
}
}
public virtual void TestMethods()
{
CharArrayMap <int?> cm = new CharArrayMap <int?>(TEST_VERSION_CURRENT, 2, false);
//Dictionary<string, int?> hm = new Dictionary<string, int?>();
Dictionary <object, int?> hm = new Dictionary <object, int?>(); // TODO: In .NET, we cannot implicitly convert from string to object using generics
hm["foo"] = 1;
hm["bar"] = 2;
cm.PutAll(hm);
assertEquals(hm.Count, cm.Count);
hm["baz"] = 3;
cm.PutAll(hm);
assertEquals(hm.Count, cm.Count);
// TODO: In .NET we cannot make this conversion implicitly.
CharArraySet cs = cm.Keys as CharArraySet;
int n = 0;
foreach (object o in cs)
{
assertTrue(cm.ContainsKey(o));
char[] co = (char[])o;
assertTrue(cm.ContainsKey(co, 0, co.Length));
n++;
}
assertEquals(hm.Count, n);
assertEquals(hm.Count, cs.Count);
assertEquals(cm.Count, cs.Count);
// TODO: This directly contradicts the TestModifyOnUnmodifiable test,
// where clear is not allowed from the Keys property.
//cs.Clear();
//assertEquals(0, cs.Count);
//assertEquals(0, cm.Count);
try
{
cs.Add("test");
fail("keySet() allows adding new keys");
}
catch (System.NotSupportedException)
{
// pass
}
cm.PutAll(hm);
assertEquals(hm.Count, cs.Count);
assertEquals(cm.Count, cs.Count);
IEnumerator <KeyValuePair <object, int?> > iter1 = IDictionaryExtensions.EntrySet(cm).GetEnumerator();
n = 0;
while (iter1.MoveNext())
{
KeyValuePair <object, int?> entry = iter1.Current;
object key = entry.Key;
int? val = entry.Value;
assertEquals(cm.Get(key), val);
// TODO: In .NET the Value property of KeyValuePair is read-only. Do we need a solution?
//entry.Value = val * 100;
//assertEquals(val * 100, (int)cm.Get(key));
n++;
}
assertEquals(hm.Count, n);
cm.Clear();
cm.PutAll(hm);
assertEquals(cm.size(), n);
CharArrayMap <int?> .EntryIterator iter2 = cm.EntrySet().GetEnumerator() as CharArrayMap <int?> .EntryIterator;
n = 0;
while (iter2.MoveNext())
{
char[] keyc = (char[])iter2.Current.Key;
int? val = iter2.Current.Value;
assertEquals(hm[new string(keyc)], val);
// TODO: In .NET the Value property of KeyValuePair is read-only. Do we need a solution?
//iter2.Value = val * 100;
//assertEquals(val * 100, (int)cm.Get(keyc));
n++;
}
assertEquals(hm.Count, n);
// TODO: In .NET, the EntrySet extension method makes a copy of the data
// so clearing it won't work like this.
cm.EntrySet().clear();
assertEquals(0, cm.size());
assertEquals(0, cm.EntrySet().size());
assertTrue(cm.Count == 0);
}