/// <summary>
/// Load a stemmer table from an inputstream.
/// </summary>
public static Trie Load(Stream stemmerTable)
{
DataInputStream @in = null;
try
{
@in = new DataInputStream(stemmerTable);
string method = @in.ReadUTF().ToUpperInvariant();
if (method.IndexOf('M') < 0)
{
return new Trie(@in);
}
else
{
return new MultiTrie2(@in);
}
}
finally
{
@in.Dispose();
}
}
private static void WriteAndReadAString()
{
// Write out a string whose UTF-8 encoding is quite possibly
// longer than 65535 bytes
int length = Random().nextInt(A_NUMBER_NEAR_65535) + 1;
MemoryStream baos = new MemoryStream();
StringBuilder testBuffer = new StringBuilder();
for (int i = 0; i < length; i++)
{
testBuffer.append((char)Random().Next());
}
string testString = testBuffer.toString();
DataOutputStream dos = new DataOutputStream(baos);
dos.WriteUTF(testString);
// Corrupt the data to produce malformed characters
byte[] testBytes = baos.ToArray();
int dataLength = testBytes.Length;
int corruptions = Random().nextInt(MAX_CORRUPTIONS_PER_CYCLE);
for (int i = 0; i < corruptions; i++)
{
int index = Random().nextInt(dataLength);
testBytes[index] = (byte)Random().Next();
}
// Pay special attention to mangling the end to produce
// partial characters at end
testBytes[dataLength - 1] = (byte)Random().Next();
testBytes[dataLength - 2] = (byte)Random().Next();
// Attempt to decode the bytes back into a String
MemoryStream bais = new MemoryStream(testBytes);
DataInputStream dis = new DataInputStream(bais);
dis.ReadUTF();
}
private static void WriteAndReadAString()
{
// Write out a string whose UTF-8 encoding is quite possibly
// longer than 65535 bytes
int length = Random().nextInt(A_NUMBER_NEAR_65535) + 1;
MemoryStream baos = new MemoryStream();
StringBuilder testBuffer = new StringBuilder();
for (int i = 0; i < length; i++)
testBuffer.append((char)Random().Next());
string testString = testBuffer.toString();
DataOutputStream dos = new DataOutputStream(baos);
dos.WriteUTF(testString);
// Corrupt the data to produce malformed characters
byte[] testBytes = baos.ToArray();
int dataLength = testBytes.Length;
int corruptions = Random().nextInt(MAX_CORRUPTIONS_PER_CYCLE);
for (int i = 0; i < corruptions; i++)
{
int index = Random().nextInt(dataLength);
testBytes[index] = (byte)Random().Next();
}
// Pay special attention to mangling the end to produce
// partial characters at end
testBytes[dataLength - 1] = (byte)Random().Next();
testBytes[dataLength - 2] = (byte)Random().Next();
// Attempt to decode the bytes back into a String
MemoryStream bais = new MemoryStream(testBytes);
DataInputStream dis = new DataInputStream(bais);
dis.ReadUTF();
}
internal static Trie LoadTrie(string path)
{
Trie trie;
using (DataInputStream @is = new DataInputStream(
new FileStream(path, FileMode.Open, FileAccess.Read)))
{
string method = @is.ReadUTF().ToUpperInvariant();
if (method.IndexOf('M') < 0)
{
trie = new Trie(@is);
}
else
{
trie = new MultiTrie(@is);
}
}
return trie;
}