func TestGenerateTreesLowest(t *testing.T) {
_, reduced := symbols.Get([]byte("banana"))
data := []uint16{'\x03', '\x00', '\x03', '\x00', '\x01', '\x04'}
freqs := rle2.GetFrequencies(reduced, data)
trees, selections := GenerateTrees(freqs, data)
if len(trees) < 2 {
t.Error("Not enough huffman trees generated")
}
if len(trees) > 6 {
t.Error("Too many huffman trees generated")
}
if len(selections) != 1 {
t.Error("The wrong number of huffman tree selections was returned")
}
}
func TestTreeCodeLength(t *testing.T) {
_, reduced := symbols.Get([]byte("banana"))
data := []uint16{'\x03', '\x00', '\x03', '\x00', '\x01', '\x04'}
freqs := rle2.GetFrequencies(reduced, data)
lowestlen := 0
tree := NewTree(freqs)
for i, code := range tree.Codes {
if i == 0 || code.Len < lowestlen {
lowestlen = code.Len
}
}
if lowestlen != tree.Codes['\x03'].Len {
t.Error("The lowest code-length isn't the most used symbol")
}
}
func TestGenerateTreesMultipleSelections(t *testing.T) {
_, reduced := symbols.Get([]byte("banana"))
data := []uint16{'\x03', '\x00', '\x03', '\x00', '\x01', '\x04'}
base := data
for len(data) <= 200 {
data = append(data, base...)
}
freqs := rle2.GetFrequencies(reduced, data)
trees, selections := GenerateTrees(freqs, data)
if len(trees) < 2 {
t.Error("Not enough huffman trees generated")
}
if len(trees) > 6 {
t.Error("Too many huffman trees generated")
}
if len(selections) != 5 {
t.Error("The wrong number of huffman tree selections was returned")
}
}
// WriteBlock compresses the content buffered and writes
// a block to the bit writer given.
func (b *block) WriteBlock(bw *bits.Writer) error {
rleData := b.runs.Encode()
syms, reducedSyms := symbols.Get(rleData)
// BWT step.
bwtData := make([]byte, len(rleData))
bwtidx := bwt.Transform(bwtData, rleData)
// MTF step.
mtfData := bwtData
mtf.Transform(reducedSyms, mtfData, bwtData)
// RLE2 step.
rle2Data := rle2.Encode(reducedSyms, mtfData)
freqs := rle2.GetFrequencies(reducedSyms, rle2Data)
// Setup the huffman trees required to encode rle2Data.
trees, selections := huffman.GenerateTrees(freqs, rle2Data)
// Get the MTF encoded huffman tree selections.
treeSelectionSymbols := make(symbols.ReducedSet, len(trees))
for i := range trees {
treeSelectionSymbols[i] = byte(i)
}
treeSelectionBytes := make([]byte, len(selections))
for i, selection := range selections {
treeSelectionBytes[i] = byte(selection)
}
mtf.Transform(treeSelectionSymbols, treeSelectionBytes, treeSelectionBytes)
// Write the block header.
bw.WriteBits(48, blockMagic)
bw.WriteBits(32, uint64(b.crc))
bw.WriteBits(1, 0)
// Write the contents that build the decoding steps.
bw.WriteBits(24, uint64(bwtidx))
b.writeSymbolBitmaps(bw, syms)
bw.WriteBits(3, uint64(len(trees)))
bw.WriteBits(15, uint64(len(selections)))
b.writeTreeSelections(bw, treeSelectionBytes)
b.writeTreeCodes(bw, trees)
// Write the encoded contents, using the huffman trees generated
// switching them out every 50 symbols.
encoded := 0
idx := 0
tree := trees[selections[idx]]
for _, b := range rle2Data {
if encoded == huffman.TreeSelectionLimit {
encoded = 0
idx++
tree = trees[selections[idx]]
}
code := tree.Codes[b]
bw.WriteBits(uint(code.Len), code.Bits)
encoded++
}
return bw.Err()
}