// different options for the same field
public virtual void TestMixedOptions()
{
int numFields = TestUtil.NextInt(Random(), 1, 3);
var docFactory = new RandomDocumentFactory(this, numFields, 10);
foreach (var options1 in ValidOptions())
{
foreach (var options2 in ValidOptions())
{
if (options1 == options2)
{
continue;
}
using (Directory dir = NewDirectory())
{
using (var writer = new RandomIndexWriter(Random(), dir))
{
RandomDocument doc1 = docFactory.NewDocument(numFields, 20, options1);
RandomDocument doc2 = docFactory.NewDocument(numFields, 20, options2);
writer.AddDocument(AddId(doc1.ToDocument(), "1"));
writer.AddDocument(AddId(doc2.ToDocument(), "2"));
using (IndexReader reader = writer.Reader)
{
int doc1ID = DocID(reader, "1");
AssertEquals(doc1, reader.GetTermVectors(doc1ID));
int doc2ID = DocID(reader, "2");
AssertEquals(doc2, reader.GetTermVectors(doc2ID));
}
}
}
}
}
}
public virtual void TestRandom()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt32(Random, 1, 3), TestUtil.NextInt32(Random, 10, 50), RandomOptions());
}
using (Directory dir = NewDirectory())
using (RandomIndexWriter writer = new RandomIndexWriter(Random, dir, ClassEnvRule.similarity, ClassEnvRule.timeZone))
{
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
}
using (IndexReader reader = writer.GetReader())
{
for (int i = 0; i < numDocs; ++i)
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
} // reader.Dispose();
}// writer.Dispose();, dir.Dispose();
}
public void RandomDocumentTest(int minNumParagraphs, int maxNumParagraphs, int minNumSentencesPerParagraph, int maxNumSentencesPerParagraph, string documentSubject)
{
RandomNumber rn = new RandomNumber();
RandomDocument rd = new RandomDocument();
int numParagraphs = 0;
string doc = string.Empty;
try
{
_msg.Length = 0;
_msg.Append("RandomDocumentTest started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
_str.Length = 0;
numParagraphs = rn.GenerateRandomInt(minNumParagraphs, maxNumParagraphs);
doc = rd.GenerateDocument(numParagraphs, minNumSentencesPerParagraph, maxNumSentencesPerParagraph, documentSubject);
_str.Append(doc);
Program._messageLog.WriteLine(_str.ToString());
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... RandomDocumentTest finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
public virtual void TestRandom()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt(Random(), 1, 3), TestUtil.NextInt(Random(), 10, 50), RandomOptions());
}
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
}
IndexReader reader = writer.Reader;
for (int i = 0; i < numDocs; ++i)
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
reader.Dispose();
writer.Dispose();
dir.Dispose();
}
public void GenerateRandomBook()
{
RandomDocument randDoc = new RandomDocument();
int numChaptersToGenerate = 3;
int numParagraphsPerChapter = 10;
int numSentencesPerParagraph = 5;
string[] chapterTitles = null;
try
{
_msg.Length = 0;
_msg.Append("GenerateRandomBook started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
numChaptersToGenerate = Convert.ToInt32(_frm.txtNumChaptersToOutput.Text);
numParagraphsPerChapter = Convert.ToInt32(_frm.txtNumParagraphsToOutput.Text);
numSentencesPerParagraph = Convert.ToInt32(_frm.txtNumSentencesToOutput.Text);
if (numParagraphsPerChapter < 5)
{
numParagraphsPerChapter = 5;
}
if (numSentencesPerParagraph < 5)
{
numSentencesPerParagraph = 5;
}
chapterTitles = new string[2];
chapterTitles[0] = "First Chapter Is This";
chapterTitles[1] = "The Second Chapter Makes an Appearance";
string book = randDoc.GenerateBook("Test Program Book", "*** THE END ***", numChaptersToGenerate, chapterTitles, numParagraphsPerChapter - 3, numParagraphsPerChapter + 3, numSentencesPerParagraph - 2, numSentencesPerParagraph + 3);
_msg.Length = 0;
_msg.Append(book);
Program._messageLog.WriteLine(_msg.ToString());
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... GenerateRandomBook finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
public virtual void TestLotsOfFields()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5000, 10);
foreach (Options options in ValidOptions())
{
using Directory dir = NewDirectory();
using RandomIndexWriter writer = new RandomIndexWriter(Random, dir);
RandomDocument doc = docFactory.NewDocument(AtLeast(100), 5, options);
writer.AddDocument(doc.ToDocument());
using IndexReader reader = writer.GetReader();
AssertEquals(doc, reader.GetTermVectors(0));
}
}
public virtual void TestMerge()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
int numDeletes = Random.Next(numDocs);
HashSet <int?> deletes = new HashSet <int?>();
while (deletes.Count < numDeletes)
{
deletes.Add(Random.Next(numDocs));
}
foreach (Options options in ValidOptions())
{
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt32(Random, 1, 3), AtLeast(10), options);
}
using (Directory dir = NewDirectory())
using (RandomIndexWriter writer = new RandomIndexWriter(Random, dir, ClassEnvRule.similarity, ClassEnvRule.timeZone))
{
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
if (Rarely())
{
writer.Commit();
}
}
foreach (int delete in deletes)
{
writer.DeleteDocuments(new Term("id", "" + delete));
}
// merge with deletes
writer.ForceMerge(1);
using (IndexReader reader = writer.GetReader())
{
for (int i = 0; i < numDocs; ++i)
{
if (!deletes.Contains(i))
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
}
} // reader.Dispose();
}// writer.Dispose();, dir.Dispose();
}
}
public void RandomBookTest(string bookTitle, int numChapters, string pChapterHeadings,
int minNumParagraphsPerChapter, int maxNumParagraphsPerChapter,
int minNumSentencesPerParagraph, int maxNumSentencesPerParagraph)
{
RandomNumber rn = new RandomNumber();
RandomDocument rd = new RandomDocument();
string theEndText = "Finis ... The End";
string book = string.Empty;
string[] chapterHeadings = null;
try
{
_msg.Length = 0;
_msg.Append("RandomBookTest started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
if (pChapterHeadings.Trim().Length > 0)
{
chapterHeadings = pChapterHeadings.Split(new string[] { Environment.NewLine }, StringSplitOptions.None);
}
if (chapterHeadings == null)
{
book = rd.GenerateBook(bookTitle, numChapters, minNumParagraphsPerChapter, maxNumParagraphsPerChapter, minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
}
else
{
book = rd.GenerateBook(bookTitle, theEndText, numChapters, chapterHeadings,
minNumParagraphsPerChapter, maxNumParagraphsPerChapter, minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
}
Program._messageLog.WriteLine(book);
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... RandomBookTest finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
// only one doc with vectors
public virtual void TestRareVectors()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 10, 20);
foreach (Options options in ValidOptions())
{
int numDocs = AtLeast(200);
int docWithVectors = Random.Next(numDocs);
Document emptyDoc = new Document();
using (Directory dir = NewDirectory())
using (RandomIndexWriter writer = new RandomIndexWriter(Random, dir, ClassEnvRule.similarity, ClassEnvRule.timeZone))
{
RandomDocument doc = docFactory.NewDocument(TestUtil.NextInt32(Random, 1, 3), 20, options);
for (int i = 0; i < numDocs; ++i)
{
if (i == docWithVectors)
{
writer.AddDocument(AddId(doc.ToDocument(), "42"));
}
else
{
writer.AddDocument(emptyDoc);
}
}
using (IndexReader reader = writer.GetReader())
{
int docWithVectorsID = DocID(reader, "42");
for (int i = 0; i < 10; ++i)
{
int docID = Random.Next(numDocs);
Fields fields = reader.GetTermVectors(docID);
if (docID == docWithVectorsID)
{
AssertEquals(doc, fields);
}
else
{
Assert.IsNull(fields);
}
}
Fields fields_ = reader.GetTermVectors(docWithVectorsID);
AssertEquals(doc, fields_);
} // reader.Dispose();
}// writer.Dispose();, dir.Dispose();
}
}
// run random tests from different threads to make sure the per-thread clones
// don't share mutable data
public virtual void TestClone()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
foreach (Options options in ValidOptions())
{
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt32(Random, 1, 3), AtLeast(10), options);
}
AtomicObject <Exception> exception = new AtomicObject <Exception>();
using (Directory dir = NewDirectory())
using (RandomIndexWriter writer = new RandomIndexWriter(Random, dir, ClassEnvRule.similarity, ClassEnvRule.timeZone))
{
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
}
using (IndexReader reader = writer.GetReader())
{
for (int i = 0; i < numDocs; ++i)
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
ThreadClass[] threads = new ThreadClass[2];
for (int i = 0; i < threads.Length; ++i)
{
threads[i] = new ThreadAnonymousInnerClassHelper(this, numDocs, docs, reader, exception, i);
}
foreach (ThreadClass thread in threads)
{
thread.Start();
}
foreach (ThreadClass thread in threads)
{
thread.Join();
}
} // reader.Dispose();
}// writer.Dispose();, dir.Dispose();
Assert.IsNull(exception.Value, "One thread threw an exception");
}
}
public virtual void TestLotsOfFields()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 500, 10);
foreach (Options options in ValidOptions())
{
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, ClassEnvRule.Similarity, ClassEnvRule.TimeZone);
RandomDocument doc = docFactory.NewDocument(AtLeast(100), 5, options);
writer.AddDocument(doc.ToDocument());
IndexReader reader = writer.Reader;
AssertEquals(doc, reader.GetTermVectors(0));
reader.Dispose();
writer.Dispose();
dir.Dispose();
}
}
public void GenerateRandomChapter()
{
RandomDocument randDoc = new RandomDocument();
int numParagraphsToGenerate = 3;
int numSentencesPerParagraph = 10;
try
{
_msg.Length = 0;
_msg.Append("GenerateRandomChapter started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
numParagraphsToGenerate = Convert.ToInt32(_frm.txtNumParagraphsToOutput.Text);
numSentencesPerParagraph = Convert.ToInt32(_frm.txtNumSentencesToOutput.Text);
if (numSentencesPerParagraph < 5)
{
numSentencesPerParagraph = 5;
}
string chapter = randDoc.GenerateChapter("Test Program Chapter", numParagraphsToGenerate, numSentencesPerParagraph - 2, numSentencesPerParagraph + 3);
_msg.Length = 0;
_msg.Append(chapter);
Program._messageLog.WriteLine(_msg.ToString());
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... GenerateRandomChapter finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
public void RandomParagraphsTest(int minNumParagraphs, int maxNumParagraphs, int minNumSentencesPerParagraph, int maxNumSentencesPerParagraph)
{
RandomNumber rn = new RandomNumber();
RandomDocument rd = new RandomDocument();
int numParagraphs = 0;
int numSentences = 0;
string paragraph = string.Empty;
try
{
_msg.Length = 0;
_msg.Append("RandomParagraphsTest started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
_str.Length = 0;
numParagraphs = rn.GenerateRandomInt(minNumParagraphs, maxNumParagraphs);
for (int p = 0; p < numParagraphs; p++)
{
numSentences = rn.GenerateRandomInt(minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
paragraph = rd.GenerateParagraph(numSentences);
_str.Append(paragraph);
_str.Append(Environment.NewLine);
_str.Append(Environment.NewLine);
}
Program._messageLog.WriteLine(_str.ToString());
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... RandomParagraphsTest finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
public virtual void TestHighFreqs()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 3, 5);
foreach (Options options in ValidOptions())
{
if (options == Options.NONE)
{
continue;
}
using Directory dir = NewDirectory();
using RandomIndexWriter writer = new RandomIndexWriter(Random, dir);
RandomDocument doc = docFactory.NewDocument(TestUtil.NextInt32(Random, 1, 2), AtLeast(20000),
options);
writer.AddDocument(doc.ToDocument());
using IndexReader reader = writer.GetReader();
AssertEquals(doc, reader.GetTermVectors(0));
}
}
public virtual void TestHighFreqs()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 3, 5);
foreach (Options options in ValidOptions())
{
if (options == Options.NONE)
{
continue;
}
using (Directory dir = NewDirectory())
using (RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, ClassEnvRule.Similarity, ClassEnvRule.TimeZone))
{
RandomDocument doc = docFactory.NewDocument(TestUtil.NextInt(Random(), 1, 2), AtLeast(20000),
options);
writer.AddDocument(doc.ToDocument());
using (IndexReader reader = writer.Reader)
AssertEquals(doc, reader.GetTermVectors(0));
}
}
}
protected virtual void AssertEquals(RandomDocument doc, Fields fields)
{
// compare field names
Assert.AreEqual(doc is null, fields is null);
Assert.AreEqual(doc.fieldNames.Length, fields.Count);
ISet <string> fields1 = new JCG.HashSet <string>();
ISet <string> fields2 = new JCG.HashSet <string>();
for (int i = 0; i < doc.fieldNames.Length; ++i)
{
fields1.Add(doc.fieldNames[i]);
}
foreach (string field in fields)
{
fields2.Add(field);
}
Assert.IsTrue(fields1.SetEquals(fields2));
for (int i = 0; i < doc.fieldNames.Length; ++i)
{
AssertEquals(doc.tokenStreams[i], doc.fieldTypes[i], fields.GetTerms(doc.fieldNames[i]));
}
}
protected internal virtual void AssertEquals(RandomDocument doc, Fields fields)
{
// compare field names
Assert.AreEqual(doc == null, fields == null);
Assert.AreEqual(doc.FieldNames.Length, fields.Size);
HashSet <string> fields1 = new HashSet <string>();
HashSet <string> fields2 = new HashSet <string>();
for (int i = 0; i < doc.FieldNames.Length; ++i)
{
fields1.Add(doc.FieldNames[i]);
}
foreach (string field in fields)
{
fields2.Add(field);
}
Assert.IsTrue(fields1.SetEquals(fields2));
for (int i = 0; i < doc.FieldNames.Length; ++i)
{
AssertEquals(doc.TokenStreams[i], doc.FieldTypes[i], fields.Terms(doc.FieldNames[i]));
}
}
public void GenerateRandomParagraphs()
{
RandomDocument randDoc = new RandomDocument();
int numParagraphsToGenerate = 3;
int numSentencesPerParagraph = 10;
try
{
_msg.Length = 0;
_msg.Append("GenerateRandomParagraphs started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
numParagraphsToGenerate = Convert.ToInt32(_frm.txtNumParagraphsToOutput.Text);
numSentencesPerParagraph = Convert.ToInt32(_frm.txtNumSentencesToOutput.Text);
for (int p = 0; p < numParagraphsToGenerate; p++)
{
string paragraph = randDoc.GenerateParagraph(numSentencesPerParagraph);
_msg.Length = 0;
_msg.Append(paragraph);
Program._messageLog.WriteLine(_msg.ToString());
}
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... GenerateRandomParagraphs finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
// [Test] // LUCENENET NOTE: For now, we are overriding this test in every subclass to pull it into the right context for the subclass
// run random tests from different threads to make sure the per-thread clones
// don't share mutable data
public virtual void TestClone()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
foreach (Options options in ValidOptions())
{
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt(Random(), 1, 3), AtLeast(10), options);
}
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, ClassEnvRule.Similarity, ClassEnvRule.TimeZone);
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
}
IndexReader reader = writer.Reader;
for (int i = 0; i < numDocs; ++i)
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
AtomicObject<Exception> exception = new AtomicObject<Exception>();
ThreadClass[] threads = new ThreadClass[2];
for (int i = 0; i < threads.Length; ++i)
{
threads[i] = new ThreadAnonymousInnerClassHelper(this, numDocs, docs, reader, exception, i);
}
foreach (ThreadClass thread in threads)
{
thread.Start();
}
foreach (ThreadClass thread in threads)
{
thread.Join();
}
reader.Dispose();
writer.Dispose();
dir.Dispose();
Assert.IsNull(exception.Value, "One thread threw an exception");
}
}
// [Test] // LUCENENET NOTE: For now, we are overriding this test in every subclass to pull it into the right context for the subclass
public virtual void TestMerge()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
int numDeletes = Random().Next(numDocs);
HashSet<int?> deletes = new HashSet<int?>();
while (deletes.Count < numDeletes)
{
deletes.Add(Random().Next(numDocs));
}
foreach (Options options in ValidOptions())
{
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt(Random(), 1, 3), AtLeast(10), options);
}
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, ClassEnvRule.Similarity, ClassEnvRule.TimeZone);
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
if (Rarely())
{
writer.Commit();
}
}
foreach (int delete in deletes)
{
writer.DeleteDocuments(new Term("id", "" + delete));
}
// merge with deletes
writer.ForceMerge(1);
IndexReader reader = writer.Reader;
for (int i = 0; i < numDocs; ++i)
{
if (!deletes.Contains(i))
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
}
reader.Dispose();
writer.Dispose();
dir.Dispose();
}
}
// [Test] // LUCENENET NOTE: For now, we are overriding this test in every subclass to pull it into the right context for the subclass
public virtual void TestRandom()
{
RandomDocumentFactory docFactory = new RandomDocumentFactory(this, 5, 20);
int numDocs = AtLeast(100);
RandomDocument[] docs = new RandomDocument[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = docFactory.NewDocument(TestUtil.NextInt(Random(), 1, 3), TestUtil.NextInt(Random(), 10, 50), RandomOptions());
}
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, ClassEnvRule.Similarity, ClassEnvRule.TimeZone);
for (int i = 0; i < numDocs; ++i)
{
writer.AddDocument(AddId(docs[i].ToDocument(), "" + i));
}
IndexReader reader = writer.Reader;
for (int i = 0; i < numDocs; ++i)
{
int docID = DocID(reader, "" + i);
AssertEquals(docs[i], reader.GetTermVectors(docID));
}
reader.Dispose();
writer.Dispose();
dir.Dispose();
}
protected internal virtual void AssertEquals(RandomDocument doc, Fields fields)
{
// compare field names
Assert.AreEqual(doc == null, fields == null);
Assert.AreEqual(doc.FieldNames.Length, fields.Size);
HashSet<string> fields1 = new HashSet<string>();
HashSet<string> fields2 = new HashSet<string>();
for (int i = 0; i < doc.FieldNames.Length; ++i)
{
fields1.Add(doc.FieldNames[i]);
}
foreach (string field in fields)
{
fields2.Add(field);
}
Assert.IsTrue(fields1.SetEquals(fields2));
for (int i = 0; i < doc.FieldNames.Length; ++i)
{
AssertEquals(doc.TokenStreams[i], doc.FieldTypes[i], fields.Terms(doc.FieldNames[i]));
}
}
public void RandomChaptersTest(int numChapters, int minNumParagraphsPerChapter, int maxNumParagraphsPerChapter,
int minNumSentencesPerParagraph, int maxNumSentencesPerParagraph,
string pChapterHeadings)
{
RandomNumber rn = new RandomNumber();
RandomDocument rd = new RandomDocument();
int numParagraphs = 0;
string chap = string.Empty;
string[] chapterHeadings = null;
int chapHeadingInx = -1;
int chapHeadingMaxInx = -1;
try
{
_msg.Length = 0;
_msg.Append("RandomChaptersTest started ...\r\n");
Program._messageLog.WriteLine(_msg.ToString());
if (pChapterHeadings.Trim().Length > 0)
{
chapterHeadings = pChapterHeadings.Split(new string[] { Environment.NewLine }, StringSplitOptions.None);
chapHeadingMaxInx = chapterHeadings.Length - 1;
}
chapHeadingInx = -1;
for (int ch = 0; ch < numChapters; ch++)
{
numParagraphs = rn.GenerateRandomInt(minNumParagraphsPerChapter, maxNumParagraphsPerChapter);
if (chapterHeadings == null)
{
_str.Length = 0;
_str.Append("Chapter ");
_str.Append((ch + 1).ToString());
_str.Append(Environment.NewLine);
_str.Append(Environment.NewLine);
chap = rd.GenerateChapter(numParagraphs, minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
_str.Append(chap);
chap = _str.ToString();
}
else
{
chapHeadingInx++;
if (chapHeadingInx > chapHeadingMaxInx)
{
//More chapters are being generated than there are defined chapter headings
_str.Length = 0;
_str.Append("Chapter ");
_str.Append((ch + 1).ToString());
chap = rd.GenerateChapter(_str.ToString(), numParagraphs, minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
}
else
{
chap = rd.GenerateChapter(chapterHeadings[chapHeadingInx], numParagraphs, minNumSentencesPerParagraph, maxNumSentencesPerParagraph);
}
}
Program._messageLog.WriteLine(chap);
}
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
Program._messageLog.WriteLine(_msg.ToString());
AppMessages.DisplayErrorMessage(_msg.ToString(), _saveErrorMessagesToAppLog);
}
finally
{
_msg.Length = 0;
_msg.Append("\r\n... RandomChaptersTest finished.");
Program._messageLog.WriteLine(_msg.ToString());
}
}
/// <summary>
/// Routine to create random documents.
/// </summary>
/// <param name="numRows">Number of data rows to generate containing random sentences.</param>
/// <param name="minNumParagraphs">Minimum number of paragraphs in a data row.</param>
/// <param name="maxNumParagraphs">Maximum number of paragraphs in a data row.</param>
/// <param name="minNumSentencesPerParagraph">Minimum number of sentences per paragraph.</param>
/// <param name="maxNumSentencesPerParagraph">Maximum number of sentences per paragraph.</param>
/// <param name="includeDocumentTitle">If true, a line containing a random phrase that serves as document title is included at the beginning of each random document. If false, no title is generated.</param>
/// <returns>ADO.NET DataTable containing the random documents.</returns>
public DataTable CreateRandomDocumentDataTable(int numRows, string minNumParagraphs, string maxNumParagraphs, string minNumSentencesPerParagraph, string maxNumSentencesPerParagraph, bool includeDocumentTitle)
{
DataTable dt = new DataTable();
RandomNumber rn = new RandomNumber();
PFRandomWord randWord = new PFRandomWord();
RandomDocument randDocument = new RandomDocument();
StringBuilder title = new StringBuilder();
try
{
int minimumNumberOfParagraphs = Convert.ToInt32(minNumParagraphs);
int maximumNumberOfParagraphs = Convert.ToInt32(maxNumParagraphs);
int minimumNumberOfSentencesPerParagraph = Convert.ToInt32(minNumSentencesPerParagraph);
int maximumNumberOfSentencesPerParagraph = Convert.ToInt32(maxNumSentencesPerParagraph);
DataColumn dc = new DataColumn("RandomValue");
dc.DataType = Type.GetType("System.String");
dt.Columns.Add(dc);
for (int i = 0; i < numRows; i++)
{
DataRow dr = dt.NewRow();
title.Length = 0;
if (includeDocumentTitle)
{
int numWordsInTitle = rn.GenerateRandomInt((int)3, (int)7);
for (int t = 0; t < numWordsInTitle; t++)
{
if (t > 0)
{
title.Append(" ");
}
string temp = randWord.GetWord();
temp = CultureInfo.CurrentCulture.TextInfo.ToTitleCase(temp.ToLower());
title.Append(temp);
}
}
int numParagraphs = rn.GenerateRandomInt(minimumNumberOfParagraphs, maximumNumberOfParagraphs);
string str = randDocument.GenerateDocument(numParagraphs, minimumNumberOfSentencesPerParagraph, maximumNumberOfSentencesPerParagraph, title.ToString());
dr[0] = str;
dt.Rows.Add(dr);
}
}
catch (System.Exception ex)
{
_msg.Length = 0;
_msg.Append("Error in CreateRandomDocumentDataTable routine.\r\n");
_msg.Append(AppGlobals.AppMessages.FormatErrorMessage(ex));
throw new System.Exception(_msg.ToString());
}
finally
{
;
}
return(dt);
}
