That little class wouldn't be too hard to convert to perl, although I don't
have a perl installation handy to convert it myself...  Maybe if you or
someone you work with is familiar with perl you could convert it...  Or
maybe you could contact the original author (his email address is in the
source listing) and ask him for help.

hi, Fabio,

about a year ago i re-wrote our website's fulltext search and had very
similar questions. seems to me it would be a great open source
(mini-)project: doesn't need to be very big, but still takes more than half
an hour.

one question, the first replier to your question talked about perl, but i
didn't find any mention of perl in your question. what language are you
implementing in?

i implemented in C#. not sure whether my result is of much use to you, but i
can sketch how i proceeded. i basically support quotes, AND, OR, NEAR, and
NOT (which i promote to "AND NOT" if it isn't already written that way). i
eliminate naked (unquoted) noise words, and i eliminate a quoted chunk if it
contains only noise words. if 2 terms appear without an operator between
them, i treat them as ANDed -- which to me is what "net intuition" expects.

my Contains "getter" (C# term) begins as follows to give you an idea.

cheers,

Tim

/// <summary>

/// for many reasons it would be grand to support a real search grammar
(last and perhaps least because we could convert search terms to canonical
form for tagging)

/// but we don't yet (and our indexing has other problems such as not
allowing hyphens in words (thus we can't find wal-mart and g-8-gipfel)

/// so here is our poor man's quasi-grammar

/// </summary>

static Regex reBinarySearchOperators = new Regex(@"^(AND|OR|NEAR)$",
RegexOptions.IgnoreCase);

static Regex reUnarySearchOperators = new Regex(@"^(NOT)$",
RegexOptions.IgnoreCase);

public string Contains {

get {

if (IsEmpty) {