Thilanka Marasinghe wrote:
> I was wondering if theres anything out there that can take web page full of
> text and search it for particular phrases I give it.. say for example I
> have A web page or several linked webpages.. I could say find "how to
> initiate a class".. or something like that.. This might seem likea silly
> question. also if I were to attempt something like that Would I have to use
> ruby rails?..
> 
> 
> thanks thilankka
> 

Well, there are two parts to that question. One is how to get the body 
of a webpage in a string,

require 'net/http'
str = Net::HTTP.get URI.parse('http://www.ruby-lang.org/en/')

and the second is how to search within that string.

str.scan(/.*ruby.*/) do |match|
	puts "This line contained the word \"ruby\" in lower case:"
	puts match
end

This type of program is called a "screen scraper", and there are some 
tools/libraries that aid in their creation, if you want something 
fancier than what I outlined above. WWW::Mechanize is the classic 
(ported from Perl, I think), but the one I hear the most about is 
Hpricot (based on Mechanize). These support things like using cookies 
and automatically following redirects, and probably even the ability to 
search a page and strip the HTML out of the result (though I may be 
misremembering--I have only read about these programs). If you still 
want to write your own stuff but get annoyed at the clunkiness and 
obscure errors that Net::HTTP can be, have a look at http-access2 here: 
http://raa.ruby-lang.org/project/http-access2/ .

There are plenty of other screen scraping libraries that I didn't 
mention. Just use google. Your real problem is how to actually do the 
searching. For that, you will need to read up on regular expressions. 
One simple test might be

matches = 0
[/how/, /build/, /initiate|instantiate/, /class/].each do |regex|
	matches += 1 if str =~ regex
end
if matches > 2
	do something with str
end

The point is, you will need to be creative and research and probably 
also ask for help. (The above does something with the string if it 
matches 3 or 4 of the regular expressions in the list.)

Good luck,
Dan