A very fast string matching algorithm for smallalphabets and long patterns (

We are interested in the exact string matching problem which consists of searching for all the occurrences of a pattern of length m in a text of length n. Both the pattern and the text are built over an alphabet of size. We present three versions of an exact string matching algorithm. They use a new shifting technique. The rst version is straightforward and easy to implement. The second version is linear in the worst case, an improvement over the rst. The main result is the third algorithm. It is very fast in practice for small alphabet and long patterns. Asymptotically, it performs O(log m(m + n=(m ? log m))) inspections of text symbols in the average case. This compares favorably with many other string searching algorithms.