Filter binlogs, selectively replicate queries

binlog-filter.rb

#!/usr/bin/env ruby
require "rubygems"
require "bundler/setup"

require "sequel"

# DB Config
options = {
  :database => "development",
  :host => "localhost",
  :user => "root",
  :socket => "/tmp/mysql.sock"}

DB = Sequel.mysql2(options)

TABLES = %w{foos bars}

def tables_regex
  @tables_regex ||= TABLES.join("|")
end

def query_regexes
  @query_regexes ||= [
    /\/\*.*\*\/ INSERT INTO `(#{tables_regex})` \((.*)\) VALUES\((.*)\)/,
    /\/\*.*\*\/ UPDATE `(#{tables_regex})` SET (.*)/]
end

def binlog_filenames 
  Dir.glob("/usr/local/var/mysql/*binlog*.0*")
end

def read_lines_from(filename)
  `mysqlbinlog #{filename}`.split("\n")
end

def read_binlog_lines
  binlog_filenames.map {|f| read_lines_from(f) }.flatten
end

def get_matches_from(log_lines)
  log_lines.select {|l| query_regexes.any? {|r| l.match(r)}}
end

def execute_query(q)
  begin
    DB.run(q)
  rescue => e
    logger.warn(e)
  end
end

def execute_query_statements(queries)
  queries.each {|q| execute_query(q) }
end

execute_query_statements(get_matches_from(read_binlog_lines))