lazy twitter backup script

backuptweets.rb

grep -m 1  * | grep -v ']}'
require 'rubygems'
require 'twitter'
require 'json'

username = "fin" 
last_id = nil
page = 1
numberoftweets = 0
savedtweets = 0
backup_dir = 'mytweets'

while 1 do
  begin
    Twitter.user_timeline(username, :page => page, :count => 200, :include_rts => true, :include_entities => true).each do |tweet|
      t = tweet.attrs
      savethis = t.to_json
      text = t['text']
      last_id = t['id_str']
      numberoftweets+=1
      filename = "#{backup_dir}/#{last_id}.json"
      if not File.exists? filename
        f = File.new filename, 'w'
        f.puts text
        f.puts savethis
        f.close
        savedtweets+=1
      end
    end 
    $stderr.puts "page #{page}, last id #{last_id}, #{numberoftweets} tweets, #{savedtweets} new"
    page += 1
    if page > 20
      break
    end
  rescue => ex
    $stderr.puts "error, retrying"
    $stderr.puts ex.inspect
    sleep 4
  end
end

search the folder using:

$ grep -m 1 WHAT YOU ARE LOOKING FOR * | grep -v ']}'

-m 1 makes sure only the first line (text) of the file is read.
| grep -v ']}' filters matches in markup only. unless you tweet weird character combinations. your fault then.

or if you need to search metadata, feed the output through

$ sed 's/.\text":"([^"]).*"/\1/g'

to only get the text