public void TestWrite()
{
PostingListEncoder decoder = new PostingListEncoder();
SpimiBlockWriter writer = new SpimiBlockWriter();
writer.AddPosting("bTerm", DocA);
writer.AddPosting("aTerm", DocA);
writer.AddPosting("aTerm", DocB);
string filePath = writer.FlushToFile();
using (FileStream file = File.Open(filePath, FileMode.Open))
{
BinaryReader reader = new BinaryReader(file);
Assert.AreEqual(2, reader.ReadInt32());
Assert.AreEqual("aTerm", reader.ReadString());
IList<Posting> postings = new List<Posting>();
postings.Add(new Posting(DocA, 1));
postings.Add(new Posting(DocB, 1));
IList<Posting> readPostings = decoder.read(reader);
for (int i = 0; i < postings.Count; i++ )
{
readPostings[i].Equals(postings[i]);
}
Assert.AreEqual("bTerm", reader.ReadString());
readPostings = decoder.read(reader);
Assert.AreEqual(new Posting(DocA, 1), readPostings[0]);
}
}
public SpimiIndexer(ILexer lexer)
{
this.lexer = lexer;
this.blockReader = new SpimiBlockReader();
this.blockWriter = new SpimiBlockWriter();
this.fileIndexWriter = new FileIndexWriter();
}
public SpimiIndexer(ILexer lexer, IParser parser, Stream indexStream, Stream metadata)
{
this.lexer = lexer;
this.parser = parser;
this.termIndexBlockWriter = new SpimiBlockWriter();
this.indexStream = indexStream;
this.metadataWriter = new CollectionMetadataWriter(metadata);
}
public void TestMerge()
{
List<IEnumerator<PostingList>> postingLists
= new List<IEnumerator<PostingList>>();
SpimiBlockReader reader = new SpimiBlockReader();
SpimiBlockWriter writer = new SpimiBlockWriter();
writer.AddPosting("aTerm", "aDoc");
writer.AddPosting("aTerm", "bDoc");
writer.AddPosting("bTerm", "aDoc");
postingLists.Add(reader.Read(writer.FlushToFile()).GetEnumerator());
writer = new SpimiBlockWriter();
writer.AddPosting("cTerm", "dDoc");
writer.AddPosting("aTerm", "zDoc");
writer.AddPosting("bTerm", "aDoc");
postingLists.Add(reader.Read(writer.FlushToFile()).GetEnumerator());
using (IEnumerator<PostingList> merged = reader.BeginBlockMerge(postingLists).GetEnumerator())
{
Assert.IsTrue(merged.MoveNext());
PostingList postingList = merged.Current;
Assert.AreEqual("aTerm", postingList.Term);
Assert.AreEqual(3, postingList.Postings.Count);
Assert.AreEqual("aDoc", postingList.Postings[0]);
Assert.AreEqual("bDoc", postingList.Postings[1]);
Assert.AreEqual("zDoc", postingList.Postings[2]);
Assert.IsTrue(merged.MoveNext());
postingList = merged.Current;
Assert.AreEqual("bTerm", postingList.Term);
Assert.AreEqual(1, postingList.Postings.Count);
Assert.AreEqual("aDoc", postingList.Postings[0]);
Assert.IsTrue(merged.MoveNext());
postingList = merged.Current;
Assert.AreEqual("cTerm", postingList.Term);
Assert.AreEqual(1, postingList.Postings.Count);
Assert.AreEqual("dDoc", postingList.Postings[0]);
Assert.IsFalse(merged.MoveNext());
}
}
public void TestRead()
{
SpimiBlockWriter writer = new SpimiBlockWriter();
writer.AddPosting("aTerm", "aDoc");
writer.AddPosting("aTerm", "bDoc");
writer.AddPosting("bTerm", "aDoc");
string filePath = writer.FlushToFile();
SpimiBlockReader reader = new SpimiBlockReader();
IEnumerable<PostingList> postingLists = reader.Read(filePath);
IEnumerator<PostingList> enumerator = postingLists.GetEnumerator();
Assert.AreEqual(true, enumerator.MoveNext());
Assert.AreEqual("aTerm", enumerator.Current.Term);
Assert.AreEqual("aDoc", enumerator.Current.Postings[0]);
Assert.AreEqual("bDoc", enumerator.Current.Postings[1]);
Assert.AreEqual(true, enumerator.MoveNext());
Assert.AreEqual("bTerm", enumerator.Current.Term);
Assert.AreEqual("aDoc", enumerator.Current.Postings[0]);
}
public void TestWrite()
{
SpimiBlockWriter writer = new SpimiBlockWriter();
writer.AddPosting("bTerm", "aDoc");
writer.AddPosting("aTerm", "aDoc");
writer.AddPosting("aTerm", "bDoc");
string filePath = writer.FlushToFile();
using (FileStream file = File.Open(filePath, FileMode.Open))
{
BinaryReader reader = new BinaryReader(file);
Assert.AreEqual(2, reader.ReadInt32());
Assert.AreEqual("aTerm", reader.ReadString());
Assert.AreEqual(2, reader.ReadInt32());
Assert.AreEqual("aDoc", reader.ReadString());
Assert.AreEqual("bDoc", reader.ReadString());
Assert.AreEqual("bTerm", reader.ReadString());
Assert.AreEqual(1, reader.ReadInt32());
Assert.AreEqual("aDoc", reader.ReadString());
}
}
public void TestRead()
{
SpimiBlockWriter writer = new SpimiBlockWriter();
writer.AddPosting("aTerm", DocA);
writer.AddPosting("aTerm", DocB);
writer.AddPosting("bTerm", DocA);
string filePath = writer.FlushToFile();
SpimiBlockReader reader = new SpimiBlockReader();
IEnumerator<PostingList> postingLists = reader.Read(filePath).GetEnumerator();
Assert.AreEqual(true, postingLists.MoveNext());
Assert.AreEqual("aTerm", postingLists.Current.Term);
Assert.AreEqual(2, postingLists.Current.Postings.Count);
Assert.AreEqual(new Posting(DocA, 1), postingLists.Current.Postings[0]);
Assert.AreEqual(new Posting(DocB, 1), postingLists.Current.Postings[1]);
Assert.AreEqual(true, postingLists.MoveNext());
Assert.AreEqual("bTerm", postingLists.Current.Term);
Assert.AreEqual(new Posting(DocA, 1), postingLists.Current.Postings[0]);
}