splitclient-rb 3.3.0 → 4.0.0.pre.rc1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 63050dc6d8209d6fdd050462bbafabc650e74fc7
-  data.tar.gz: cbb40be59c858289ae351c6115e147a7062362d1
+  metadata.gz: c6868ad86ef9ff027cff7dbdb58cfa973c7d02d2
+  data.tar.gz: 254804356acd6e57dedd0f01fcec6549374651e7
 SHA512:
-  metadata.gz: 3e9be02c6bb231fce90773edac53d39af03c634012a39dd96ac6480009836c9e34102b8651b8b55f6449a967a0a06292b89b4186ada8e62ac81b3c158956d1de
-  data.tar.gz: a1875dd25ac3602f5660224dd5a890a86c058820c08a716e32c9c29bbdedc35ea26fce9e47977fa5224a382f93cea3140973e0bf8b21be99b4f9f5d6c82ff7be
+  metadata.gz: 2d6f2d5097557c0eda262290c9d80a6a031e1c2b7f4acab2dc3fb163113136241435ccbbe137cd5d5e63f37d46dcea2c9514d0269a792f3ff7f4ea9359db8bc3
+  data.tar.gz: 7a164d6e280b8223108d418392d7051ef2b9f7b3d78389aea46d772180b35af68264cdc3e0943f2634c1ccac0a4b1cfa9fbaf243befc3058a39426193d8b0f0d

data/CHANGES.txt

@@ -1,3 +1,7 @@
+4.0.0
+- Add support for murmur3 algo
+- Optimize memory usage
+
 3.3.0
 - Add support for traffic allocation
 

data/Detailed-README.md

@@ -467,6 +467,14 @@ This will generate a file gemspec with the right version, then:
 gem push splitclient-rb-<VERSION>.gem
 ```
 
+## Benchmarking
+
+To benchmark hashing algorithms (currently we're using MurmurHash) you'll need to run:
+
+```bash
+bundle exec rake benchmark_hashing_algorithm
+```
+
 ## Contributing
 
 Bug reports and pull requests are welcome on GitHub at https://github.com/splitio/ruby-client.

data/NEWS

@@ -1,3 +1,8 @@
+4.0.0
+
+Add support for murmur3 hashing algorithm
+Optimize gem memory usage
+
 3.3.0
 
 Add support for traffic allocation

data/lib/cache/adapters/memory_adapters/map_adapter.rb

@@ -6,7 +6,7 @@ module SplitIoClient
       module MemoryAdapters
         # Memory adapter implementation, which stores everything inside thread-safe Map
         class MapAdapter
-          def initialize(_ = nil)
+          def initialize
             @map = Concurrent::Map.new
           end
 

data/lib/engine/evaluator/splitter.rb

@@ -1,110 +1,118 @@
+require 'digest/murmurhash'
+
 module SplitIoClient
   # Misc class in charge of providing hash functions and
   # determination of treatment based on concept of buckets
   # based on provided key
   #
   class Splitter < NoMethodError
-
-    #
-    # Checks if the partiotion size is 100%
-    #
-    # @param partitions [object] array of partitions
-    #
-    # @return [boolean] true if partition is 100% false otherwise
-    def self.hundred_percent_one_treatment?(partitions)
-      if partitions.size != 1
-        return false
+    class << self
+      #
+      # Checks if the partiotion size is 100%
+      #
+      # @param partitions [object] array of partitions
+      #
+      # @return [boolean] true if partition is 100% false otherwise
+      def hundred_percent_one_treatment?(partitions)
+        (partitions.size != 1) ? false : (partitions.first.size == 100)
       end
-      return (partitions.first).size == 100
-    end
 
+      #
+      # gets the appropriate treatment based on id, seed and partition value
+      #
+      # @param id [string] user key
+      # @param seed [number] seed for the user key
+      # @param partitions [object] array of partitions
+      #
+      # @return traetment [object] treatment value
+      def get_treatment(id, seed, partitions, legacy_algo)
+        legacy = (legacy_algo == 1 || legacy_algo == nil) ? true : false
 
-    #
-    # gets the appropriate treatment based on id, seed and partition value
-    #
-    # @param id [string] user key
-    # @param seed [number] seed for the user key
-    # @param partitions [object] array of partitions
-    #
-    # @return traetment [object] treatment value
-    def self.get_treatment(id, seed, partitions)
-      if partitions.empty?
-        return Treatments::CONTROL
-      end
+        if partitions.empty?
+          return Treatments::CONTROL
+        end
 
-      if hundred_percent_one_treatment?(partitions)
-        return (partitions.first).treatment
-      end
+        if hundred_percent_one_treatment?(partitions)
+          return (partitions.first).treatment
+        end
 
-      return get_treatment_for_key(bucket(hash(id, seed)), partitions)
-    end
+        return get_treatment_for_key(bucket(count_hash(id, seed, legacy_algo)), partitions)
+      end
 
-    #
-    # returns a hash value for the give key, sedd pair
-    #
-    # @param key [string] user key
-    # @param seed [number] seed for the user key
-    #
-    # @return hash [string] hash value
-    def self.hash(key, seed)
-      h = 0
-      for i in 0..key.length-1
-        h = to_int32(31 * h + key[i].ord)
+      # returns a hash value for the give key, seed pair
+      #
+      # @param key [String] user key
+      # @param seed [Fixnum] seed for the user key
+      #
+      # @return hash [String] hash value
+      def count_hash(key, seed, legacy)
+        legacy ? legacy_hash(key, seed) : murmur_hash(key, seed)
       end
-      h^seed
-    end
 
-    #
-    # misc method to convert ruby number to int 32 since overflow is handled different to java
-    #
-    # @param number [number] ruby number value
-    #
-    # @return [int] returns the int 32 value of the provided number
-    def self.to_int32(number)
-      begin
-        sign = number < 0 ? -1 : 1
-        abs = number.abs
-        return 0 if abs == 0 || abs == Float::INFINITY
-      rescue
-        return 0
+      def murmur_hash(key, seed)
+        Digest::MurmurHash3_x86_32.rawdigest(key, [seed].pack('L'))
       end
 
-      pos_int = sign * abs.floor
-      int_32bit = pos_int % 2**32
+      def legacy_hash(key, seed)
+        h = 0
 
-      return int_32bit - 2**32 if int_32bit >= 2**31
-      int_32bit
-    end
+        for i in 0..key.length-1
+          h = to_int32(31 * h + key[i].ord)
+        end
 
-    #
-    # returns the treatment for a bucket given the partitions
-    #
-    # @param bucket [number] bucket value
-    # @param parittions [object] array of partitions
-    #
-    # @return treatment [treatment] treatment value for this bucket and partitions
-    def self.get_treatment_for_key(bucket, partitions)
-      buckets_covered_thus_far = 0
-      partitions.each do |p|
-        unless p.is_empty?
-          buckets_covered_thus_far += p.size
-          return p.treatment if buckets_covered_thus_far >= bucket
+        h^seed
+      end
+
+      #
+      # misc method to convert ruby number to int 32 since overflow is handled different to java
+      #
+      # @param number [number] ruby number value
+      #
+      # @return [int] returns the int 32 value of the provided number
+      def to_int32(number)
+        begin
+          sign = number < 0 ? -1 : 1
+          abs = number.abs
+          return 0 if abs == 0 || abs == Float::INFINITY
+        rescue
+          return 0
         end
+
+        pos_int = sign * abs.floor
+        int_32bit = pos_int % 2**32
+
+        return int_32bit - 2**32 if int_32bit >= 2**31
+        int_32bit
       end
 
-      return Treatments::CONTROL
-    end
+      #
+      # returns the treatment for a bucket given the partitions
+      #
+      # @param bucket [number] bucket value
+      # @param parittions [object] array of partitions
+      #
+      # @return treatment [treatment] treatment value for this bucket and partitions
+      def get_treatment_for_key(bucket, partitions)
+        buckets_covered_thus_far = 0
+        partitions.each do |p|
+          unless p.is_empty?
+            buckets_covered_thus_far += p.size
+            return p.treatment if buckets_covered_thus_far >= bucket
+          end
+        end
 
-    #
-    # returns bucket value for the given hash value
-    #
-    # @param hash_value [string] hash value
-    #
-    # @return bucket [number] bucket number
-    def self.bucket(hash_value)
-      (hash_value.abs % 100) + 1
-    end
+        return Treatments::CONTROL
+      end
 
+      #
+      # returns bucket value for the given hash value
+      #
+      # @param hash_value [string] hash value
+      #
+      # @return bucket [number] bucket number
+      def bucket(hash_value)
+        (hash_value.abs % 100) + 1
+      end
+    end
   end
-
 end

data/lib/engine/parser/split_treatment.rb

@@ -23,6 +23,7 @@ module SplitIoClient
         def match(split, keys, attributes)
           in_rollout = false
           key = keys[:bucketing_key] ? keys[:bucketing_key] : keys[:matching_key]
+          legacy_algo = (split[:algo] == 1 || split[:algo] == nil) ? true : false
 
           split[:conditions].each do |c|
             condition = SplitIoClient::Condition.new(c)
@@ -31,7 +32,7 @@ module SplitIoClient
 
             if !in_rollout && condition.type == SplitIoClient::Condition::TYPE_ROLLOUT
               if split[:trafficAllocation] < 100
-                bucket = Splitter.bucket(Splitter.hash(key, split[:trafficAllocationSeed].to_i))
+                bucket = Splitter.bucket(Splitter.count_hash(key, split[:trafficAllocationSeed].to_i, legacy_algo))
 
                 if bucket >= split[:trafficAllocation]
                   return treatment(Models::Label::NOT_IN_SPLIT, @default_treatment, split[:changeNumber])
@@ -42,7 +43,8 @@ module SplitIoClient
             end
 
             if matcher_type(condition).match?(keys[:matching_key], attributes)
-              result = Splitter.get_treatment(key, split[:seed], condition.partitions)
+              key = keys[:bucketing_key] ? keys[:bucketing_key] : keys[:matching_key]
+              result = Splitter.get_treatment(key, split[:seed], condition.partitions, split[:algo])
 
               if result.nil?
                 return treatment(Models::Label::NO_RULE_MATCHED, @default_treatment, split[:changeNumber])

data/lib/splitclient-rb/clients/split_client.rb

@@ -1,6 +1,5 @@
 module SplitIoClient
   class SplitClient
-
     #
     # Creates a new split client instance that connects to split.io API.
     #
@@ -79,7 +78,7 @@ module SplitIoClient
 
         store_impression(
           split_name, matching_key, bucketing_key,
-          { treatment: SplitIoClient::Treatments::CONTROL, label: Models::Label::EXCEPTION },
+          { treatment: SplitIoClient::Treatments::CONTROL, label: Engine::Models::Label::EXCEPTION },
           store_impressions
         )
 
@@ -98,7 +97,7 @@ module SplitIoClient
 
         store_impression(
           split_name, matching_key, bucketing_key,
-          { treatment: SplitIoClient::Treatments::CONTROL, label: Models::Label::EXCEPTION },
+          { treatment: SplitIoClient::Treatments::CONTROL, label: Engine::Models::Label::EXCEPTION },
           store_impressions
         )
 
@@ -141,5 +140,11 @@ module SplitIoClient
         treatment_label_change_number[:treatment]
       end
     end
+
+    private
+
+    def split_treatment
+      @split_treatment ||= SplitIoClient::Engine::Parser::SplitTreatment.new(@segments_repository)
+    end
   end
 end

data/lib/splitclient-rb/split_config.rb

@@ -185,12 +185,7 @@ module SplitIoClient
     def self.init_cache_adapter(adapter, data_structure, redis_url = nil, impressions_queue_size = nil)
       case adapter
       when :memory
-        # takes :memory_adapter (symbol) and returns MemoryAdapter (string)
-        adapter = SplitIoClient::Cache::Adapters::MemoryAdapters.const_get(
-          data_structure.to_s.split('_').collect(&:capitalize).join
-        ).new(impressions_queue_size)
-
-        SplitIoClient::Cache::Adapters::MemoryAdapter.new(adapter)
+        SplitIoClient::Cache::Adapters::MemoryAdapter.new(map_memory_adapter(data_structure, impressions_queue_size))
       when :redis
         begin
           require 'redis'
@@ -202,6 +197,15 @@ module SplitIoClient
       end
     end
 
+    def self.map_memory_adapter(name, queue_size)
+      case name
+      when :map_adapter
+        SplitIoClient::Cache::Adapters::MemoryAdapters::MapAdapter.new
+      when :queue_adapter
+        SplitIoClient::Cache::Adapters::MemoryAdapters::QueueAdapter.new(queue_size)
+      end
+    end
+
     def self.default_mode
       :standalone
     end

data/lib/splitclient-rb/version.rb

@@ -1,3 +1,3 @@
 module SplitIoClient
-  VERSION = '3.3.0'
+  VERSION = '4.0.0-rc1'
 end

data/splitclient-rb.gemspec

@@ -27,6 +27,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "pry"
   spec.add_development_dependency "pry-byebug"
   spec.add_development_dependency "simplecov"
+  spec.add_development_dependency "allocation_stats"
 
   spec.add_runtime_dependency "json", "~> 1.8"
   spec.add_runtime_dependency "thread_safe"
@@ -36,4 +37,5 @@ Gem::Specification.new do |spec|
   spec.add_runtime_dependency "faraday_middleware"
   spec.add_runtime_dependency "net-http-persistent", "<= 2.9.4"
   spec.add_runtime_dependency "redis"
+  spec.add_runtime_dependency "digest-murmurhash"
 end

data/tasks/benchmark_hashing_algorithm.rake

@@ -0,0 +1,51 @@
+require 'benchmark'
+require 'digest/murmurhash'
+
+desc 'Benchmark murmur32 hashing algorithm'
+
+task :benchmark_hashing_algorithm do
+  iterations = 200_000
+  key = SecureRandom.uuid
+
+  Benchmark.bmbm do |x|
+    x.report('MurmurHash1') do
+      iterations.times { Digest::MurmurHash1.rawdigest(key) }
+    end
+
+    x.report('MurmurHash2') do
+      iterations.times { Digest::MurmurHash2.rawdigest(key) }
+    end
+
+    x.report('MurmurHash2A') do
+      iterations.times { Digest::MurmurHash2A.rawdigest(key) }
+    end
+
+    x.report('LegacyHash') do
+      iterations.times { legacy_hash(key, 123) }
+    end
+  end
+end
+
+def legacy_hash(key, seed)
+  h = 0
+  for i in 0..key.length-1
+    h = to_int32(31 * h + key[i].ord)
+  end
+  h^seed
+end
+
+def to_int32(number)
+  begin
+    sign = number < 0 ? -1 : 1
+    abs = number.abs
+    return 0 if abs == 0 || abs == Float::INFINITY
+  rescue
+    return 0
+  end
+
+  pos_int = sign * abs.floor
+  int_32bit = pos_int % 2**32
+
+  return int_32bit - 2**32 if int_32bit >= 2**31
+  int_32bit
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: splitclient-rb
 version: !ruby/object:Gem::Version
-  version: 3.3.0
+  version: 4.0.0.pre.rc1
 platform: ruby
 authors:
 - Split Software
@@ -122,6 +122,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
+- !ruby/object:Gem::Dependency
+  name: allocation_stats
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 - !ruby/object:Gem::Dependency
   name: json
   requirement: !ruby/object:Gem::Requirement
@@ -234,6 +248,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
+- !ruby/object:Gem::Dependency
+  name: digest-murmurhash
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 description: Ruby client for using split SDK.
 email:
 - pato@split.io
@@ -313,6 +341,7 @@ files:
 - splitclient-rb.gemspec
 - splitio.yml.example
 - tasks/benchmark_get_treatment.rake
+- tasks/benchmark_hashing_algorithm.rake
 - tasks/console.rake
 - tasks/rspec.rake
 homepage: https://github.com/splitio/ruby-client
@@ -330,9 +359,9 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">="
+  - - ">"
     - !ruby/object:Gem::Version
-      version: '0'
+      version: 1.3.1
 requirements: []
 rubyforge_project: 
 rubygems_version: 2.5.2