redjohn · February 7, 2012 02:11 · redjohn · Feb 7, 2012 · gburd · Mar 26, 2012
diff --git a/Gemfile b/Gemfile
 source :rubygems

 gem 'riak-client'
diff --git a/gen_riak_data.rb b/gen_riak_data.rb
 require 'json'

 [[:first, 2000, 9001], [:second, 50_000, 9002]].each do |record|
  urls = []
  1.upto record[1] do |i|
    url = {
      audit_id: record[2],
      url: "http://example.com/#{i.to_s(16)}",
      loadtime: 700 + rand(200),
      attempts: 1 + rand(3),
      objects: 1 + rand(300),
      processing_error: rand(1) == 1 ? true : false,
      status_code: [200, 301, 302, 404, 500].shuffle.first,
      depth: 1 + rand(4),
    }
    urls << url
  end
  File.open("#{record[0]}.json", 'w') {|f| f.write(urls.to_json) }
 end
diff --git a/load_riak_data.rb b/load_riak_data.rb
 require 'riak'

 client = Riak::Client.new(nodes: [{ host: 'localhost', http_port: 8091 }])

 bucket = client.bucket('urls')
 bucket.n_value = 1

 audits = []
 [[:first, 9001], [:second, 9002]].each do |record|
  urls = JSON.parse(File.read("#{record[0]}.json"))
  audit_id = record[1]
  urls.each do |url|
    obj = bucket.new
    obj.data = url
    obj.indexes = { 'audit_id_int' => Set.new([audit_id]) }
    obj.store
  end
 end
diff --git a/query.js b/query.js
 {
    "inputs": {
        "bucket": "urls",
        "index": "audit_id_int",
        "key": 9001
    },
    "query": [
        {
            "map": {
                "language": "javascript",
                "name": "Riak.mapValuesJson",
            }
        },
        {
            "reduce": {
                "language": "javascript",
                "source": "function(values) {
                    return values.map(function(item) {
                        return (parseInt(item, 10) || item === 0) ? item : item.loadtime;
                    });
                }",
            }
        }
    ],
    "timeout": 120000
 }
	require 'json'

	[[:first, 2000, 9001], [:second, 50_000, 9002]].each do \|record\|
	urls = []
	1.upto record[1] do \|i\|
	url = {
	audit_id: record[2],
	url: "http://example.com/#{i.to_s(16)}",
	loadtime: 700 + rand(200),
	attempts: 1 + rand(3),
	objects: 1 + rand(300),
	processing_error: rand(1) == 1 ? true : false,
	status_code: [200, 301, 302, 404, 500].shuffle.first,
	depth: 1 + rand(4),
	}
	urls << url
	end
	File.open("#{record[0]}.json", 'w') {\|f\| f.write(urls.to_json) }
	end
	require 'riak'

	client = Riak::Client.new(nodes: [{ host: 'localhost', http_port: 8091 }])

	bucket = client.bucket('urls')
	bucket.n_value = 1

	audits = []
	[[:first, 9001], [:second, 9002]].each do \|record\|
	urls = JSON.parse(File.read("#{record[0]}.json"))
	audit_id = record[1]
	urls.each do \|url\|
	obj = bucket.new
	obj.data = url
	obj.indexes = { 'audit_id_int' => Set.new([audit_id]) }
	obj.store
	end
	end
	{
	"inputs": {
	"bucket": "urls",
	"index": "audit_id_int",
	"key": 9001
	},
	"query": [
	{
	"map": {
	"language": "javascript",
	"name": "Riak.mapValuesJson",
	}
	},
	{
	"reduce": {
	"language": "javascript",
	"source": "function(values) {
	return values.map(function(item) {
	return (parseInt(item, 10) \|\| item === 0) ? item : item.loadtime;
	});
	}",
	}
	}
	],
	"timeout": 120000
	}