Abstract: In a data processing system, data from an input buffer is compressed to produce a compressed data set in a compressed data buffer, wherein the compressed data set has a Lempel-Ziv compressed data format having raw-byte tokens and string tokens. The string tokens in the compressed data set each have a predetermined number of bits, thereby facilitating efficient, high-speed parsing of tokens during subsequent decompression of the compressed data set. In some embodiments, the raw-byte tokens and the string tokens have the same predetermined number of bits.

Abstract: In a data processing system having a history buffer for storing previously decompressed data and an output buffer having at least N bytes, wherein N is greater than 1, a group of input tokens is received from an input buffer. Each input token is either a raw-byte token or a string token, wherein each string token includes a pointer and a length. Thereafter, the process outputs, to the output buffer, decompressed bytes taken from each of the raw-byte tokens within the group of input tokens. The process then determines whether each of the pointers in the string tokens points to data previously stored in the history buffer or points to data within one of the raw-byte tokens within the group of input tokens, which have not yet been stored in the history buffer. For each of the pointers that points to data stored in the history buffer, decompressed bytes pointed to in the history buffer by the pointer and the length are output to the output buffer.