riakpb 0.1.4 → 0.1.5

data/lib/riak/map_reduce.rb

@@ -0,0 +1,218 @@
+require 'riak'
+
+module Riak
+  # Class for invoking map-reduce jobs using the HTTP interface.
+  class MapReduce
+    include Util::Translation
+    # @return [Array<[bucket,key]>,String] The bucket/keys for input to the job, or the bucket (all keys).
+    # @see #add
+    attr_accessor :inputs
+
+    # @return [Array<Phase>] The map and reduce phases that will be executed
+    # @see #map
+    # @see #reduce
+    # @see #link
+    attr_accessor :query
+
+    # Creates a new map-reduce job.
+    # @param [Client] client the Riak::Client interface
+    # @yield [self] helpful for initializing the job
+    def initialize(client)
+      @client, @inputs, @query = client, [], []
+      yield self if block_given?
+    end
+
+    # Add or replace inputs for the job.
+    # @overload add(bucket)
+    #   Run the job across all keys in the bucket.  This will replace any other inputs previously added.
+    #   @param [String, Bucket] bucket the bucket to run the job on
+    # @overload add(bucket,key)
+    #   Add a bucket/key pair to the job.
+    #   @param [String,Bucket] bucket the bucket of the object
+    #   @param [String] key the key of the object
+    # @overload add(object)
+    #   Add an object to the job (by its bucket/key)
+    #   @param [Key] object the object to add to the inputs
+    # @overload add(bucket, key, keydata)
+    #   @param [String,Bucket] bucket the bucket of the object
+    #   @param [String] key the key of the object
+    #   @param [String] keydata extra data to pass along with the object to the job
+    # @return [MapReduce] self
+    def add(*params)
+      params = params.dup.flatten
+      case params.size
+      when 1
+        p = params.first
+        case p
+        when Riak::Bucket
+          @inputs = p.name
+        when Riak::Key
+          @inputs << p.to_input
+        when String
+          @inputs = p
+        end
+      when 2..3
+        bucket = params.shift
+        bucket = bucket.name if Riak::Bucket === bucket
+        @inputs << params.unshift(bucket)
+      end
+      self
+    end
+    alias :<< :add
+    alias :include :add
+
+    # Add a map phase to the job.
+    # @overload map(function)
+    #   @param [String, Array] function a Javascript function that represents the phase, or an Erlang [module,function] pair
+    # @overload map(function?, options)
+    #   @param [String, Array] function a Javascript function that represents the phase, or an Erlang [module, function] pair
+    #   @param [Hash] options extra options for the phase (see {Phase#initialize})
+    # @return [MapReduce] self
+    # @see Phase#initialize
+    def map(*params)
+      options = params.extract_options!
+      @query << Phase.new({:type => :map, :function => params.shift}.merge(options))
+      self
+    end
+
+    # Add a reduce phase to the job.
+    # @overload reduce(function)
+    #   @param [String, Array] function a Javascript function that represents the phase, or an Erlang [module,function] pair
+    # @overload reduce(function?, options)
+    #   @param [String, Array] function a Javascript function that represents the phase, or an Erlang [module, function] pair
+    #   @param [Hash] options extra options for the phase (see {Phase#initialize})
+    # @return [MapReduce] self
+    # @see Phase#initialize
+    def reduce(*params)
+      options = params.extract_options!
+      @query << Phase.new({:type => :reduce, :function => params.shift}.merge(options))
+      self
+    end
+
+    # Add a link phase to the job. Link phases follow links attached to objects automatically (a special case of map).
+    # @param [Hash] params represents the types of links to follow
+    # @return [MapReduce] self
+    def link(params={})
+      bucket          ||= params[:bucket]
+      tag             ||= params[:tag]
+      keep              = params[:keep] || false
+      function          = {}
+
+      function[:bucket] = bucket  unless bucket.nil?
+      function[:tag]    = tag     unless tag.nil?
+
+      @query << Phase.new({:type => :link, :function => function, :keep => keep})
+
+      return(self)
+    end
+    alias :walk :link
+
+    # Sets the timeout for the map-reduce job.
+    # @param [Fixnum] value the job timeout, in milliseconds
+    def timeout(value)
+      @timeout = value
+    end
+
+    # Convert the job to JSON for submission over the HTTP interface.
+    # @return [String] the JSON representation
+    def to_json(options={})
+      hash = {"inputs" => inputs, "query" => query.map(&:as_json)}
+      hash['timeout'] = @timeout.to_i if @timeout
+      ActiveSupport::JSON.encode(hash, options)
+    end
+
+    # Executes this map-reduce job.
+    # @return [Array<Array>] similar to link-walking, each element is an array of results from a phase where "keep" is true. If there is only one "keep" phase, only the results from that phase will be returned.
+    def run
+      response = @client.map_reduce_request(to_json, "application/json")
+#      ActiveSupport::JSON.decode(response[:body])
+    end
+
+    # Represents an individual phase in a map-reduce pipeline. Generally you'll want to call
+    # methods of {MapReduce} instead of using this directly.
+    class Phase
+      include Util::Translation
+      # @return [Symbol] the type of phase - :map, :reduce, or :link
+      attr_accessor :type
+
+      # @return [String, Array<String, String>, Hash, WalkSpec] For :map and :reduce types, the Javascript function to run (as a string or hash with bucket/key), or the module + function in Erlang to run. For a :link type, a {Riak::WalkSpec} or an equivalent hash.
+      attr_accessor :function
+
+      # @return [String] the language of the phase's function - "javascript" or "erlang". Meaningless for :link type phases.
+      attr_accessor :language
+
+      # @return [Boolean] whether results of this phase will be returned
+      attr_accessor :keep
+
+      # @return [Array] any extra static arguments to pass to the phase
+      attr_accessor :arg
+
+      # Creates a phase in the map-reduce pipeline
+      # @param [Hash] options options for the phase
+      # @option options [Symbol] :type one of :map, :reduce, :link
+      # @option options [String] :language ("javascript") "erlang" or "javascript"
+      # @option options [String, Array, Hash] :function In the case of Javascript, a literal function in a string, or a hash with :bucket and :key. In the case of Erlang, an Array of [module, function].  For a :link phase, a hash including any of :bucket, :tag or a WalkSpec.
+      # @option options [Boolean] :keep (false) whether to return the results of this phase
+      # @option options [Array] :arg (nil) any extra static arguments to pass to the phase
+      def initialize(options={})
+        self.type = options[:type]
+        self.language = options[:language] || "javascript"
+        self.function = options[:function]
+        self.keep = options[:keep] || false
+        self.arg = options[:arg]
+      end
+
+      def type=(value)
+        raise ArgumentError, t("invalid_phase_type") unless value.to_s =~ /^(map|reduce|link)$/i
+        @type = value.to_s.downcase.to_sym
+      end
+
+      def function=(value)
+        case value
+        when Array
+          raise ArgumentError, t("module_function_pair_required") unless value.size == 2
+          @language = "erlang"
+        when Hash
+          raise ArgumentError, t("stored_function_invalid") unless type == :link || value.has_key?(:bucket) && value.has_key?(:key)
+          @language = "javascript"
+        when String
+          @language = "javascript"
+        else
+          raise ArgumentError, t("invalid_function_value", :value => value.inspect)
+        end
+        @function = value
+      end
+
+      # Converts the phase to JSON for use while invoking a job.
+      # @return [String] a JSON representation of the phase
+      def to_json(options=nil)
+        ActiveSupport::JSON.encode(as_json, options)
+      end
+
+      # Converts the phase to its JSON-compatible representation for job invocation.
+      # @return [Hash] a Hash-equivalent of the phase
+      def as_json(options=nil)
+        obj = case type
+              when :map, :reduce
+                defaults = {:language => language, :keep => keep}
+                case function
+                when Hash
+                  defaults.merge(function)
+                when String
+                  if function =~ /\s*function/
+                    defaults.merge("source" => function)
+                  else
+                    defaults.merge("name" => function)
+                  end
+                when Array
+                  defaults.merge("module" => function[0], "function" => function[1])
+                end
+              when :link
+                function.merge({:keep => keep})
+              end
+        obj["arg"] = arg if arg
+        { type => obj }
+      end
+    end
+  end
+end

data/lib/riak.rb

@@ -9,7 +9,7 @@ require 'base64'
 require 'riak/client_pb'
 
 module Riak
-  VERSION = '0.1.4'
+  VERSION = '0.1.5'
 
   # Domain objects
   autoload :I18n,             'riak/i18n'

data/load_stocks.rb

@@ -0,0 +1,12 @@
+require 'riak'
+require 'csv'
+
+client = Riak::Client.new
+bucket = client["goog"]
+
+CSV.foreach('goog.csv', :headers => true) do |row|
+  puts row.first[1].to_s
+  key = bucket[row.first[1].to_s]
+  key.content.value = Hash[row.to_a]
+  key.save
+end

data/riakpb.gemspec

@@ -0,0 +1,37 @@
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{riakpb}
+  s.version = "0.1.5"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Scott Gonyea"]
+  s.date = %q{2010-06-16}
+  s.description = %q{riakpb is a protocol buffer client for Riak--the distributed database by Basho.  Connects via Protocol Buffers instead of REST.}
+  s.email = %q{me@inherentlylame.com}
+  s.files = ["goog.csv", "History.txt", "lib/riak/bucket.rb", "lib/riak/client/rpc.rb", "lib/riak/client.rb", "lib/riak/client_pb.rb", "lib/riak/failed_exchange.rb", "lib/riak/failed_request.rb", "lib/riak/i18n.rb", "lib/riak/key.rb", "lib/riak/locale/en.yml", "lib/riak/map_reduce.rb", "lib/riak/riak_content.rb", "lib/riak/sibling_error.rb", "lib/riak/util/decode.rb", "lib/riak/util/encode.rb", "lib/riak/util/message_code.rb", "lib/riak/util/translation.rb", "lib/riak.rb", "load_stocks.rb", "Manifest.txt", "Rakefile", "README.rdoc", "riakpb.gemspec", "script/console", "script/destroy", "script/generate", "spec/riak/bucket_spec.rb", "spec/riak/client_spec.rb", "spec/riak/key_spec.rb", "spec/riak/map_reduce_spec.rb", "spec/riak/riak_content_spec.rb", "spec/riak/rpc_spec.rb", "spec/spec_helper.rb"]
+  s.homepage = %q{http://github.com/aitrus/riak-pbclient}
+  s.require_paths = ["lib"]
+  s.rubygems_version = %q{1.3.6}
+  s.summary = %q{riakpb is a protocol buffer client for Riak--the distributed database by Basho.}
+  s.test_files = ["spec/riak/bucket_spec.rb", "spec/riak/client_spec.rb", "spec/riak/key_spec.rb", "spec/riak/map_reduce_spec.rb", "spec/riak/riak_content_spec.rb", "spec/riak/rpc_spec.rb", "spec/spec_helper.rb"]
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 3
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+      s.add_development_dependency(%q<rspec>, ["~> 2.0.0.beta.9"])
+      s.add_runtime_dependency(%q<activesupport>, [">= 2.3.5"])
+      s.add_runtime_dependency(%q<protobuf>, [">= 0.4.4"])
+    else
+      s.add_dependency(%q<rspec>, ["~> 2.0.0.beta.9"])
+      s.add_dependency(%q<activesupport>, [">= 2.3.5"])
+      s.add_dependency(%q<protobuf>, [">= 0.4.4"])
+    end
+  else
+    s.add_dependency(%q<rspec>, ["~> 2.0.0.beta.9"])
+    s.add_dependency(%q<activesupport>, [">= 2.3.5"])
+    s.add_dependency(%q<protobuf>, [">= 0.4.4"])
+  end
+end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 1
-  - 4
-  version: 0.1.4
+  - 5
+  version: 0.1.5
 platform: ruby
 authors: 
 - Scott Gonyea
@@ -18,21 +18,23 @@ date: 2010-06-16 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
-  name: rubyforge
+  name: rspec
   prerelease: false
   requirement: &id001 !ruby/object:Gem::Requirement 
     requirements: 
-    - - ">="
+    - - ~>
       - !ruby/object:Gem::Version 
         segments: 
         - 2
         - 0
-        - 4
-        version: 2.0.4
+        - 0
+        - beta
+        - 9
+        version: 2.0.0.beta.9
   type: :development
   version_requirements: *id001
 - !ruby/object:Gem::Dependency 
-  name: hoe
+  name: activesupport
   prerelease: false
   requirement: &id002 !ruby/object:Gem::Requirement 
     requirements: 
@@ -40,60 +42,75 @@ dependencies:
       - !ruby/object:Gem::Version 
         segments: 
         - 2
-        - 6
-        - 0
-        version: 2.6.0
-  type: :development
+        - 3
+        - 5
+        version: 2.3.5
+  type: :runtime
   version_requirements: *id002
-description: This is a Ruby client for Riak, using protocol buffers instead of REST.  It offers some benefit in terms of speed and it abstracts Buckets/Keys differently than does the REST client.  Different != Better.
-email: 
-- me@sgonyea.com
+- !ruby/object:Gem::Dependency 
+  name: ruby_protobuf
+  prerelease: false
+  requirement: &id003 !ruby/object:Gem::Requirement 
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 0
+        - 4
+        - 4
+        version: 0.4.4
+  type: :runtime
+  version_requirements: *id003
+description: riakpb is a protocol buffer client for Riak--the distributed database by Basho.  Connects via Protocol Buffers instead of REST.
+email: me@inherentlylame.com
 executables: []
 
 extensions: []
 
-extra_rdoc_files: 
-- History.txt
-- Manifest.txt
+extra_rdoc_files: []
+
 files: 
+- goog.csv
 - History.txt
-- Manifest.txt
-- README.rdoc
-- Rakefile
-- lib/riak.rb
-- lib/riak/client.rb
+- lib/riak/bucket.rb
 - lib/riak/client/rpc.rb
+- lib/riak/client.rb
 - lib/riak/client_pb.rb
-- lib/riak/bucket.rb
-- lib/riak/key.rb
-- lib/riak/riak_content.rb
+- lib/riak/failed_exchange.rb
+- lib/riak/failed_request.rb
 - lib/riak/i18n.rb
+- lib/riak/key.rb
 - lib/riak/locale/en.yml
+- lib/riak/map_reduce.rb
+- lib/riak/riak_content.rb
+- lib/riak/sibling_error.rb
 - lib/riak/util/decode.rb
 - lib/riak/util/encode.rb
 - lib/riak/util/message_code.rb
 - lib/riak/util/translation.rb
-- lib/riak/failed_exchange.rb
-- lib/riak/failed_request.rb
-- lib/riak/sibling_error.rb
+- lib/riak.rb
+- load_stocks.rb
+- Manifest.txt
+- Rakefile
+- README.rdoc
+- riakpb.gemspec
 - script/console
 - script/destroy
 - script/generate
-- spec/spec_helper.rb
-- spec/riak/client_spec.rb
-- spec/riak/rpc_spec.rb
 - spec/riak/bucket_spec.rb
+- spec/riak/client_spec.rb
 - spec/riak/key_spec.rb
-- spec/riak/riak_content_spec.rb
 - spec/riak/map_reduce_spec.rb
+- spec/riak/riak_content_spec.rb
+- spec/riak/rpc_spec.rb
+- spec/spec_helper.rb
 has_rdoc: true
 homepage: http://github.com/aitrus/riak-pbclient
 licenses: []
 
 post_install_message: 
-rdoc_options: 
-- --main
-- README.rdoc
+rdoc_options: []
+
 require_paths: 
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement 
@@ -112,10 +129,16 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: "0"
 requirements: []
 
-rubyforge_project: riakpb
+rubyforge_project: 
 rubygems_version: 1.3.6
 signing_key: 
 specification_version: 3
-summary: riak-pbclient is a protocol buffer client for Riak--the distributed database by Basho.
-test_files: []
-
+summary: riakpb is a protocol buffer client for Riak--the distributed database by Basho.
+test_files: 
+- spec/riak/bucket_spec.rb
+- spec/riak/client_spec.rb
+- spec/riak/key_spec.rb
+- spec/riak/map_reduce_spec.rb
+- spec/riak/riak_content_spec.rb
+- spec/riak/rpc_spec.rb
+- spec/spec_helper.rb