Hi Kevin,
I've made a fix for quadratic time caused by substring() in LZW compression. The key to the fix is to use a new overloaded method TST.longestPrefixOf(String query, int startIndex) in LZW.compress() method. Attached please find 3 reference files:
-
LempelZivWelch.java: improved LZW.java
-
TernarySearchTrie: improved TST.java
-
lzw.sh: shell script proves the correctness and performance improvement.
NOTE: existing LZW seems to take forever to compress dickens.txt (~29MB)
Regards,
Donghao
LZW_fix.zip