/// <summary>Compress a slice in memory</summary>
public static int CompressTo(Slice input, CompressedBitmapWriter output)
{
if (input.IsNullOrEmpty)
{
return(0);
}
unsafe
{
fixed(byte *ptr = input.Array)
{
return(CompressToUnsafe(ptr + input.Offset, input.Count, output));
}
}
}
/// <summary>Performs a logical NOT on a compressed bitmaps</summary>
/// <param name="bitmap">Compressed bitmap</param>
/// <param name="size">Minimum logical size of the result (bits in the uncompressed bitmap)</param>
/// <returns>Compressed slice with the result of flipping all the bits in <paramref name="bitmap"/>, containing up to at least <paramref name="size"/> bits.</returns>
/// <remarks>If <paramref name="bitmap"/> is larger than <paramref name="size"/>, then the resulting bitmap will be larger.</remarks>
public static CompressedBitmap Not(this CompressedBitmap bitmap, int size)
{
if (bitmap == null)
{
throw new ArgumentNullException(nameof(bitmap));
}
// there is a high change that the final bitmap will have the same size, with an optional extra filler word at the end
var writer = new CompressedBitmapWriter(bitmap.Count + 1);
int n = 0;
if (bitmap.Count > 0)
{
foreach (var word in bitmap)
{
if (word.IsLiteral)
{
writer.Write(CompressedWord.MakeLiteral((uint)(~word.Literal)));
n += 31;
}
else
{
int fc = word.FillCount;
writer.Write(word.FillBit == 1 ? CompressedWord.ALL_ZEROES : CompressedWord.ALL_ONES, fc);
n += 31 * fc;
}
}
}
if (n < size)
{
writer.Write(CompressedWord.ALL_ONES, size / 31);
int r = size % 31;
if (r > 0)
{
writer.Write((1u << r) - 1);
}
}
return(writer.GetBitmap());
}
/// <summary>Compress a buffer in native memory</summary>
/// <param name="buffer">Pointer to the native memory buffer to compress</param>
/// <param name="count">Number of bytes to compress</param>
/// <param name="output">Where to write the compressed words</param>
/// <returns>Number of extra bits that where output in the last literal word (or none if 0)</returns>
internal static unsafe int CompressToUnsafe(byte *buffer, int count, CompressedBitmapWriter output)
{
// Simplified algorithm:
// 1) read 31 bits from input (BE)
// 2) if not all 0 (or all 1), then output a literal word (MSB set to 0), and jump back to step 1)
// 3) set LENGTH = 1, FILL_BIT = 0 (or 1)
// 4) Peek at next 31 bits, and if they are still all 0 (or 1), increment N, and jump back to step 4)
// 5) output a repeat word, with MSB set to 1, followed by FILL_BIT, and then LENGTH-1 (30 bit), and jump back to step 1)
// Optimizations:
// - for very small inputs (3 bytes or less) we return a single literal word
// - we read 64 bits at a time in the buffer, because it fits nicely in an UInt64 register
var bucket = new UncompressedWordReader(buffer, count);
uint word;
while ((word = bucket.Read()) != UncompressedWordReader.NotEnough)
{
output.Write(word);
}
// if there are remaining bits, they are padded with 0 and written as a literal
int bits = bucket.Bits;
if (bits > 0)
{
//note: MSB will already be 0
word = bucket.ReadLast();
output.Write(word);
}
// write the header
output.Pack();
return(bits);
}
internal static CompressedBitmap CompressedBinaryExpression([NotNull] CompressedBitmap left, [NotNull] CompressedBitmap right, LogicalOperation op)
{
Contract.Requires(left != null && right != null && op != LogicalOperation.And && Enum.IsDefined(typeof(LogicalOperation), op));
var writer = new CompressedBitmapWriter();
using (var liter = left.GetEnumerator())
using (var riter = right.GetEnumerator())
{
int ln = 0; // remaining count of current word in left
int rn = 0; // remaining count of current word in right
int lw = 0; // value of current word in left (if ln > 0)
int rw = 0; // value of current word in right (if rn > 0)
const int DONE = -1;
while (true)
{
if (ln == 0)
{
if (!liter.MoveNext())
{ // left is done
if (op == LogicalOperation.And || rn == DONE)
{ // no need to continue
break;
}
// continue with right until it's done
ln = DONE;
lw = 0;
continue;
}
ln = liter.Current.WordCount;
lw = liter.Current.WordValue;
}
if (rn == 0)
{
if (!riter.MoveNext())
{ // right is done
if (op == LogicalOperation.And || ln == DONE)
{ // no need to continue
break;
}
// continue with left until it's done
rn = DONE;
rw = 0;
}
rn = riter.Current.WordCount;
rw = riter.Current.WordValue;
}
if (ln == DONE)
{ // copy right
writer.Write((uint)rw, rn);
rn = 0;
}
else if (rn == DONE)
{ // copy left
writer.Write((uint)lw, ln);
ln = 0;
}
else
{ // merge left & right
int n = Math.Min(ln, rn);
switch (op)
{
case LogicalOperation.And: writer.Write((uint)(lw & rw), n); break;
case LogicalOperation.AndNot: writer.Write((uint)(lw & ~rw), n); break;
case LogicalOperation.Or: writer.Write((uint)(lw | rw), n); break;
case LogicalOperation.OrNot: writer.Write((uint)(lw | ~rw), n); break;
case LogicalOperation.Xor: writer.Write((uint)(lw ^ rw), n); break;
case LogicalOperation.XorNot: writer.Write((uint)(lw ^ ~rw), n); break;
default: throw new InvalidOperationException();
}
ln -= n;
rn -= n;
}
}
}
return writer.GetBitmap();
}
public static CompressedBitmap Not([NotNull] this CompressedBitmap bitmap, int size)
{
if (bitmap == null) throw new ArgumentNullException("bitmap");
// there is a high change that the final bitmap will have the same size, with an optional extra filler word at the end
var writer = new CompressedBitmapWriter(bitmap.Count + 1);
int n = 0;
if (bitmap.Count > 0)
{
foreach (var word in bitmap)
{
if (word.IsLiteral)
{
writer.Write(CompressedWord.MakeLiteral((uint)(~word.Literal)));
n += 31;
}
else
{
int fc = word.FillCount;
writer.Write(word.FillBit == 1 ? CompressedWord.ALL_ZEROES : CompressedWord.ALL_ONES, fc);
n += 31 * fc;
}
}
}
if (n < size)
{
writer.Write(CompressedWord.ALL_ONES, size / 31);
int r = size % 31;
if (r > 0) writer.Write((1u << r) - 1);
}
return writer.GetBitmap();
}
/// <summary>Compress a buffer in native memory</summary>
/// <param name="buffer">Pointer to the native memory buffer to compress</param>
/// <param name="count">Number of bytes to compress</param>
/// <param name="output">Where to write the compressed words</param>
/// <returns>Number of extra bits that where output in the last literal word (or none if 0)</returns>
internal static unsafe int CompressToUnsafe(byte* buffer, int count, CompressedBitmapWriter output)
{
// Simplified algorithm:
// 1) read 31 bits from input (BE)
// 2) if not all 0 (or all 1), then output a literal word (MSB set to 0), and jump back to step 1)
// 3) set LENGTH = 1, FILL_BIT = 0 (or 1)
// 4) Peek at next 31 bits, and if they are still all 0 (or 1), increment N, and jump back to step 4)
// 5) output a repeat word, with MSB set to 1, followed by FILL_BIT, and then LENGTH-1 (30 bit), and jump back to step 1)
// Optimisations:
// - for very small inputs (3 bytes or less) we return a single literal word
// - we read 64 bits at a time in the buffer, because it fits nicely in an UInt64 register
var bucket = new UncompressedWordReader(buffer, count);
uint word;
while ((word = bucket.Read()) != UncompressedWordReader.NotEnough)
{
output.Write(word);
}
// if there are reamining bits, they are padded with 0 and written as a literal
int bits = bucket.Bits;
if (bits > 0)
{
//note: MSB will already be 0
word = bucket.ReadLast();
output.Write(word);
}
// write the header
output.Pack();
return bits;
}
/// <summary>Compress a slice in memory</summary>
public static int CompressTo(Slice input, CompressedBitmapWriter output)
{
if (input.IsNullOrEmpty) return 0;
unsafe
{
fixed (byte* ptr = input.Array)
{
return CompressToUnsafe(ptr + input.Offset, input.Count, output);
}
}
}
/// <summary>Performs a binary operation between two compressed bitmaps</summary>
/// <param name="left">First compressed bitmap</param>
/// <param name="right">Second compressed bitmap</param>
/// <param name="op">Type of operation to perform (And, Or, Xor, ...)</param>
/// <returns>Compressed slice with the result of boolean expression <paramref name="left"/> AND <paramref name="right"/></returns>
internal static CompressedBitmap CompressedBinaryExpression(CompressedBitmap left, CompressedBitmap right, LogicalOperation op)
{
Contract.Requires(left != null && right != null && /*op != LogicalOperation.And &&*/ Enum.IsDefined(typeof(LogicalOperation), op));
var writer = new CompressedBitmapWriter();
using (var liter = left.GetEnumerator())
using (var riter = right.GetEnumerator())
{
int ln = 0; // remaining count of current word in left
int rn = 0; // remaining count of current word in right
uint lw = 0; // value of current word in left (if ln > 0)
uint rw = 0; // value of current word in right (if rn > 0)
const int DONE = -1;
while (true)
{
if (ln == 0)
{
if (!liter.MoveNext())
{ // left is done
if (op == LogicalOperation.And || rn == DONE)
{ // no need to continue
break;
}
// continue with right until it's done
ln = DONE;
lw = 0;
continue;
}
ln = liter.Current.WordCount;
lw = liter.Current.WordValue;
}
if (rn == 0)
{
if (!riter.MoveNext())
{ // right is done
if (op == LogicalOperation.And || ln == DONE)
{ // no need to continue
break;
}
// continue with left until it's done
rn = DONE;
rw = 0;
continue;
}
rn = riter.Current.WordCount;
rw = riter.Current.WordValue;
}
if (ln == DONE)
{ // copy right
writer.Write((uint)rw, rn);
rn = 0;
}
else if (rn == DONE)
{ // copy left
writer.Write((uint)lw, ln);
ln = 0;
}
else
{ // merge left & right
int n = Math.Min(ln, rn);
switch (op)
{
case LogicalOperation.And: writer.Write((uint)(lw & rw), n); break;
case LogicalOperation.AndNot: writer.Write((uint)(lw & (~rw & LITERAL_MASK)), n); break;
case LogicalOperation.Or: writer.Write((uint)(lw | rw), n); break;
case LogicalOperation.OrNot: writer.Write((uint)(lw | (~rw & LITERAL_MASK)), n); break;
case LogicalOperation.Xor: writer.Write((uint)(lw ^ rw), n); break;
case LogicalOperation.XorNot: writer.Write((uint)(lw ^ (~rw & LITERAL_MASK)), n); break;
default: throw new InvalidOperationException();
}
ln -= n;
rn -= n;
}
}
}
return(writer.GetBitmap());
}