File src/0dev.org/predictor/predictor.go artifact d4a72555aa part of check-in 67794bdf14

// Package predictor implements the predictor compression/decompression algorithm
// as specified by RFC1978 - PPP Predictor Compression Protocol
package predictor

import (
	"io"
)

type context struct {
	table [65536]byte
	hash  uint16
	input []byte
}

// Returns a closure over the provided writer that compresses data when called.
//
// It can buffer data as the predictor mandates 8-byte blocks with a header.
// A call with no data will force a flush.
func Compressor(writer io.Writer) func([]byte) error {
	var ctx context
	ctx.input = make([]byte, 8)

	// Forward declaration as it is required for recursion
	var write func(data []byte) error

	write = func(data []byte) error {
		var (
			err          error
			blockSize    int = 8
			bufferLength int = len(ctx.input)
		)

		// Force a flush if we are called with no data to write
		if len(data) == 0 {
			// We can't have more than 7 bytes in the buffer so this is safe
			blockSize = len(ctx.input)
			goto write
		}

		// Check if there are pending bytes in the buffer
		if bufferLength > 0 && bufferLength < 8 {

			// Check whether we have enough bytes for a complete block
			if len(data) > 8-bufferLength {
				// Fill the buffer ...
				ctx.input = append(ctx.input, data[:8-bufferLength]...)
				// ... and recurse, calling ourselves with the full buffer
				err = write(ctx.input)
				if err != nil {
					return err
				}

				// Clear the buffer
				ctx.input = ctx.input[:0]

				// Handle remaining bytes, if any
				var remaining []byte = data[8-bufferLength:]
				if len(remaining) > 0 {
					// Recurse, calling ourselves with the rest of the bytes
					err = write(data[8-bufferLength:])
					if err != nil {
						return err
					}
				}
			} else {
				// Add the insufficient data to the buffer and return
				ctx.input = append(ctx.input, data...)
				return nil
			}
		}

	write:
		var buf []byte = make([]byte, 1, blockSize+1)
		for block := 0; block < len(data)/blockSize; block++ {
			for i := 0; i < blockSize; i++ {
				var current byte = data[(block*blockSize)+i]
				if ctx.table[ctx.hash] == current {
					// Guess was right - don't output
					buf[0] |= 1 << uint(i)
				} else {
					// Guess was wrong, output char
					ctx.table[ctx.hash] = current
					buf = append(buf, current)
				}
				ctx.hash = (ctx.hash << 4) ^ uint16(current)
			}
			_, err = writer.Write(buf)
			if err != nil {
				return err
			}

			// Reset the flags and buffer for the next iteration
			buf[0] ^= buf[0]
			buf = buf[:1]
		}
		return nil
	}

	return write
}