# search, search / replace, search / execute / replace

# This example reads a web server log file, extracts all the hits which
# are arrivals from Google searches, and tablulates them. It's a demonstration
# of regular expressions from our Ruby course, and it also shows a hash of
# arrays, sorting, and other features.

arrivals = {}

for lyne in (File.new "ac_20120221").readlines
        parts = lyne.split
        calledfor = parts[6]
        referer = parts[10]
        # puts "#{calledfor} - #{referer}"

        if referer =~ /\.google\..*[?&]q=([^&"]*)/
                # puts "#{calledfor} search arrival for #{$1}"
                query = $1

        # Following two lines would be good as a function - URL decoding

                query.tr! '+', ' '
                query = query.gsub(/%(..)/) { "%c" % $1.to_i(16) }

                # puts "#{calledfor} search arrival for #{query}"

        # Make me an empty container if this is the first time Google has
        # sent someone here

                if arrivals[calledfor] == nil
                        arrivals[calledfor] = []
                end

        # Add in the query that was used into that empty container

                if query == ""
                        # print "#{$.} lyne"
                        # To add - act on empty queries (what are they?)
                else
                        arrivals[calledfor].push query
                end

        end

end

# List out all the pages that were arrived at, and a count of arrival numbers

for landingplace in arrivals.keys
        numberofarrivals = arrivals[landingplace].length
        puts "#{numberofarrivals} to #{landingplace}"
        end

# What were all the requests for popular pages?

for landingplace in arrivals.keys
        numberofarrivals = arrivals[landingplace].length
        if numberofarrivals > 30
                arrivals[landingplace].sort!
                puts "Arrived at #{landingplace} from"
                for arr_string in arrivals[landingplace]
                        puts " .......... #{arr_string}"
                end
        end
end

puts "Analysis was of #{$.} lines"