Need a better algorithmic design

[tomrittervg]

See http://swtch.com/~rsc/regexp/regexp4.html

[tomrittervg]

Other ideas for performance improvement:

• Change the commitwordmap_tbl to have one row per unique word, and the second column is a comma seperated list of commitids.
• Change the commitwordmap_tbl to have commitid,uniquewordid, and a second table uniqueword_tbl containing the uniquewordid and word
• Store nonexistance of words on a per-commit basis; no words between [a-adversary)
• Store the nonexistance of words on a global basis, so if you search for something I know if it doesn't exist and don't have to consult database for anything.