photo_culler.rb

require "instagram"
require "restclient"
require "twitter"
require 'koala'

hashtag = "danpluslaura"
INSTAGRAM_ACCESS_TOKEN = "############"
INSTAGRAM_CLIENT_ID = "###########"
INSTAGRAM_CLIENT_SECRET = "###########"

Instagram.configure do |config|
  config.client_id = INSTAGRAM_CLIENT_ID 
  config.client_secret = INSTAGRAM_CLIENT_SECRET
end

instagram_client = Instagram.client(:access_token => INSTAGRAM_ACCESS_TOKEN)
tags = instagram_client.tag_search(hashtag)
urls = Array.new
if (!tags.empty?)
  for media_item in instagram_client.tag_recent_media(tags[0].name)
    urls << media_item.images.standard_resolution.url
  end
end

twitter_client = Twitter::REST::Client.new do |config|
  config.consumer_key        = "############"
  config.consumer_secret     = "###############"
  config.access_token        = "##################"
  config.access_token_secret = "######################"
end

for tweet in twitter_client.search("#" + hashtag + " -rt", {include_entities: "1"})
  if tweet.entities?
  	if !tweet.attrs[:entities][:media].nil?
  	  if tweet.attrs[:entities][:media][0][:type].eql? "photo"
  		urls << tweet.attrs[:entities][:media][0][:media_url]
  	  end
  	end
  end
end

facebook_urls = Hash.new
oauth_access_token = "############################"
facebook_client = Koala::Facebook::API.new(oauth_access_token)
for post in facebook_client.search(hashtag)
  if post.has_key?("picture")
    facebook_urls["#{post["id"]}"] = post["picture"]
  end
end

facebook_urls.each do |id, url|
  if (File.exists?(Dir.pwd + "/#{id}"))
  	puts "file #{id} already exists from facebook"
  	next
  end
  puts "getting file #{id} from facebook"
  image = RestClient.get(url)
  path = Dir.pwd + "/#{id}"
  File.open(path, 'w') {|f| f.write(image)}
end

pattern = /^((http[s]?|ftp):\/)?\/?([^:\/\s]+)((\/\w+)*\/)([\w\-\.]+[^#?\s]+)(.*)?(#[\w\-]+)?$/

urls.each_with_index do |url, idx|
  split_url = url.match(pattern)

  if (File.exists?(Dir.pwd + "/#{split_url[6]}"))
  	puts "file #{split_url[6]} already exists"
  	next
  end

  puts "getting file #{split_url[6]}"
  image = RestClient.get(url)

  path = Dir.pwd + "/#{split_url[6]}"
  File.open(path, 'w') {|f| f.write(image) }
end