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 void TestWrite()
{
MemoryStream stream = new MemoryStream();
// FileIndex should support a stream starting at any point
stream.Seek(10, SeekOrigin.Begin);
PostingListEncoder decoder = new PostingListEncoder();
PerformWrite(stream);
BinaryReader reader = new BinaryReader(stream);
// Term count
Assert.AreEqual(2, reader.ReadInt64());
long ptr1 = reader.ReadInt64();
long ptr2 = reader.ReadInt64();
Assert.AreEqual(stream.Position, ptr1);
Assert.AreEqual("aTerm", reader.ReadString());
IList<Posting> postings = decoder.read(reader);
Assert.AreEqual(2, postings.Count);
Assert.AreEqual(new Posting(DocA, 1), postings[0]);
Assert.AreEqual(new Posting(DocB, 1), postings[1]);
Assert.AreEqual(stream.Position, ptr2);
Assert.AreEqual("bTerm", reader.ReadString());
postings = decoder.read(reader);
Assert.AreEqual(3, postings.Count);
Assert.AreEqual(new Posting(DocA, 1), postings[0]);
Assert.AreEqual(new Posting(DocZ, 1), postings[1]);
Assert.AreEqual(new Posting(DocT, 1), postings[2]);
}
public void TestReadWrite()
{
MemoryStream stream = new MemoryStream();
BinaryWriter writer = new BinaryWriter(stream);
BinaryReader reader = new BinaryReader(stream);
PostingListEncoder encoder = new PostingListEncoder();
List<Posting> postings = new List<Posting>();
postings.Add(new Posting(DocA, 23));
postings.Add(new Posting(DocB, 23));
encoder.write(writer, postings);
stream.Seek(0, SeekOrigin.Begin);
IList<Posting> readPostings = encoder.read(reader);
Assert.AreEqual(postings.Count, readPostings.Count);
for (int postingIndex = 0; postingIndex < postings.Count; postingIndex++)
{
Assert.AreEqual(postings[postingIndex].Frequency, readPostings[postingIndex].Frequency);
Assert.AreEqual(postings[postingIndex].DocumentId, readPostings[postingIndex].DocumentId);
}
}
public string FlushToFile()
{
string filename = Path.GetTempFileName();
SortedList<string, SortedList<long, Posting>> orderedEntries = index.Entries;
using (FileStream fs = File.Open(filename, FileMode.Append))
{
PostingListEncoder encoder = new PostingListEncoder();
BinaryWriter writer = new BinaryWriter(fs);
writer.Write((Int32)orderedEntries.Count);
foreach (KeyValuePair<string, SortedList<long, Posting>> termPostingsPair in orderedEntries)
{
writer.Write((string)termPostingsPair.Key);
encoder.write(writer, termPostingsPair.Value.Values);
}
}
this.index = new MemoryIndex();
return filename;
}
/// <remarks> Keeps file open until all postings have been read</remarks>
public IEnumerable<PostingList> Read(string filepath)
{
using (FileStream stream = File.Open(filepath, FileMode.Open))
{
PostingListEncoder decoder = new PostingListEncoder();
BinaryReader reader = new BinaryReader(stream);
int termCount = reader.ReadInt32();
for (int termIndex = 0; termIndex < termCount; termIndex++)
{
string term = reader.ReadString();
yield return new PostingList(term, decoder.read(reader));
}
}
}
