void ChooseIddc(FastqReader reader, BinaryWriter writer)
{
if (!encodeIds)
{
iddc = new PlaceholderIdGenerator();
return;
}
//@SRX000571_SRR002321.54856271 080226_CMLIVERKIDNEY_0007:8:330:23:135 length=36
Regex sra = new Regex(@"^(@[^.]+\.)\d+\s([\S]+)(?:\d+:){3}\d+.*$", RegexOptions.Singleline);
Regex length = new Regex(@"^.+length=\d+$", RegexOptions.Singleline);
//@HWUSI-EAS627_1:3:1:0:370/1 (or /2)
//@BILLIEHOLIDAY_3_FC30G08AAXX:1:1:0:1966
Regex encode = new Regex(@"^(@[\S]+)(?:\d+:){3}\d+(\/[12])*$", RegexOptions.Singleline);
String id = reader.GetID(0);
Match sraMatch = sra.Match(id);
Match encodeMatch = encode.Match(id);
if (sraMatch.Success) //type 0
{
Match lengthMatch = length.Match(id);
writer.Write(0);
iddc = new SraIdDeCompresser(reader, writer, sraMatch, lengthMatch.Success);
}
else if (encodeMatch.Success) //type 2
{
writer.Write(2);
iddc = new EncodeIdDeCompresser(reader, writer, encodeMatch);
}
else //type 1
{
writer.Write(1);
iddc = new PlainIdDeCompresser(reader, writer);
}
}
/* Encodes IDs starting at the given index (id) until "buffer is full"
* or the fastq file ends and writes the result in the given BinaryWriter.
* Updates id according to its advancements.
*/
public void EncodeId(ref int id)
{
// should check if "mode" is right (ie. reader && writer != null)
// but we avoid doing so for efficiency
//the first byte starts with 11 if we are encoding an ID
byte first = (byte)64;
int b = 0;
StringBuilder ids = new StringBuilder();
if (idContinuation.Length != 0)
{
ids.Append(idContinuation);
b += idContinuation.Length;
idContinuation = new StringBuilder();
}
//we assume that a continuation will never be longer
//than BUFFER
while (reader.HasIDLeft(id, 1) && b < ID_BUFFER)
{
string currentId = reader.GetID(id);
b += currentId.Length;
if (b > ID_BUFFER)
{
//continuation
ids.Append(currentId.Substring(0, ID_BUFFER - (b - currentId.Length)));
idContinuation.Append(currentId.Substring(ID_BUFFER - (b - currentId.Length)));
b = ID_BUFFER;
}
else
{
ids.Append(currentId);
}
id++;
//here method to deal with known ID's structure
}
//we use ascii encoding, so 1 char = 1 byte
if (b == ID_BUFFER)
{
writer.Write(first);
writer.Write(ae.GetBytes(ids.ToString()));
}
else if (b < ID_BUFFER)
{
//mark smaller buffer
first += (byte)32; //we have to tell the decoder that we have a block with a length
//different than BUFFER
writer.Write(first);
writer.Write(b);
writer.Write(ae.GetBytes(ids.ToString()));
}
}
/* Encodes IDs starting at the given index (id) until "buffer is full"
* or the fastq file ends and writes the result in the given BinaryWriter.
* Updates id according to its advancements.
*/
public void EncodeId(ref int id)
{
idBuffer.Seek(0, SeekOrigin.Begin);
encodedId.Seek(0, SeekOrigin.Begin);
// should check if "mode" is right (ie. reader && writer != null)
// but we avoid doing so for efficiency
//the first byte starts with 11 if we are encoding an ID
byte first = (byte)64;
int b = 0;
if (continuationLength != 0)
{
encodedId.Write(idContinuation, 0, continuationLength);
b += continuationLength;
writtenContinuation = continuationLength;
continuationLength = 0;
}
//we assume that a continuation will never be longer
//than BUFFER
while (reader.HasIDLeft(id, 1) && b < ID_BUFFER)
{
//encodedId.Seek(0, SeekOrigin.Begin);
string[] currentId = reader.GetID(id).Split(separators);
if (currentId.Length < 5)
{
throw new Exception("invalid ID format");
}
//1 2 3 4
//XXX TODO check if you can use only idBuffer and not also encodedID
//the first number that has to be encoded
for (int i = 0; i < 4; i++)
{
encodedId.Write(Convert.ToUInt16(currentId[i + 1]));
// we skip the first item
}
b += ENCODED_ID_LENGTH;
byte[] buffer = ((MemoryStream)encodedId.BaseStream).GetBuffer();
if (b > ID_BUFFER)
{
//continuation
continuationLength = b - ID_BUFFER;
int firstExceedingByte = ENCODED_ID_LENGTH - continuationLength;
for (int i = 0; i < continuationLength; i++)
{
idContinuation[i] = buffer[firstExceedingByte + i];
}
idBuffer.Write(buffer, 0, firstExceedingByte);
//we have to write firstExceedingByte bytes as the count argument
b = ID_BUFFER;
}
else
{
//XXX dopo continuation non scrive primi byte per l'uint?248.1
idBuffer.Write(buffer, 0, ENCODED_ID_LENGTH + writtenContinuation);
}
id++;
encodedId.Seek(0, SeekOrigin.Begin);
writtenContinuation = 0;
}
if (b == ID_BUFFER)
{
writer.Write(first);
writer.Write(((MemoryStream)idBuffer.BaseStream).GetBuffer(), 0, b);
}
else if (b < ID_BUFFER)
{
//mark smaller buffer
first += (byte)32; //we have to tell the decoder that we have a block with a length
//different than BUFFER
writer.Write(first);
writer.Write(b);
writer.Write(((MemoryStream)idBuffer.BaseStream).GetBuffer(), 0, b);
}
}
