Scan log files and track number of subscribers over time. (for dumping into excel)

scan.rb

# Run through all zip files to get log files

# Process log files => events

# Process events =>  (event + initial_state) => new_state  => to state stream

# 

require 'rubygems'
require 'zip'
require 'date'

$TIMESTAMP = "(?:^[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}:[0-9]{2},[0-9]{3})"


def log_to_events(buffer, &block)
    buffer.scan(/#{$TIMESTAMP}(?:(?!#{$TIMESTAMP}).)+/m, &block)
end

def update(state, event) # => new_state
    return add_subscriber(state, event) if ev_is_subscription(event)    
    return remove_subscriber(state, event) if ev_is_unsubscription(event)
    state
end

def initial_state
    []
end

def ev_is_subscript_change(event)
    event =~ /MetadataUpdated/ && (ev_is_subscription(event) || ev_is_unsubscription(event))
end

def ev_is_subscription(event)
    event =~ /Add subscriber./
end

def ev_is_unsubscription(event)
    event =~ /Remove subscriber./
end

def ev_subscriber(event)
    m = event.match(/net.tcp:\/\/([^:]*):/)
    m && m[1]
end

def ev_timestamp(event)
    m = event.match(/#{$TIMESTAMP}/)
    m && DateTime.parse(m[0])
end

def add_subscriber(state, event)
    (state + [ev_subscriber(event)]).uniq
end

def remove_subscriber(state, event)
    state.reject{|e| e == ev_subscriber(event)}
end

def process_file(buffer, state)
    events = log_to_events(buffer)
    @index ||= 1;
    events.select{|e| ev_is_subscript_change(e)}.inject(state){|st,ev| new_state = update(st, ev); puts "#{ev_timestamp(ev)}\t=DATEVALUE(MID(A#{@index},1,10))+TIMEVALUE(MID(A#{@index},12,8))\t#{new_state.length}"; @index += 1; @state = new_state; new_state}
end

def for_each_zip(path)
    Dir.glob(path+"/*.Hosting.exe.*.zip") do |file|
        Zip::File.open(file) do |zipfile|
            zipfile.glob "*exe.log*" do |file|
                yield zipfile.read(file)
            end
        end
    end    
end

@state = initial_state

for_each_zip('.../Server/Archive'){|f| process_file(f, @state) }

This was a quick hacky solution, but it worked. I'm posting it so I don't loose it in-case I one-day need it again.