Abstract: A method and apparatus for reconstructing a token from a token fragment (unreconstructed token) wherein at least two indices for an unreconstructed string and three indices for a vocabulary string are iteratively calculated and recorded. Each string has associated with it two indices that record distance moved by successful character comparisons from either end of the strings. In addition, the vocabulary string also has an index that points to the last character in the vocabulary string that was successfully matched to a character in the unreconstructed string during the first comparison step with respect to the initial end of the strings.