running_stat 0.0.1 → 0.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8ec11d286e66a056fc873beecd760b3916008ea2
-  data.tar.gz: 99915adeadbbe43c59c190a5392f46c85f438126
+  metadata.gz: 692e627aba71ce44c80b57d406c85dd69f38edfc
+  data.tar.gz: 0caada43394fd224b1e8932958bbe18ed1fd4815
 SHA512:
-  metadata.gz: dc655fd6ca351e6bf5a3d9fb4331b2e704b22cd0a3551dd914ae025ad558b2a261dad16edbf7964578fc4945a7d3b54da013fc9040e83c8fb2929483f4e1e416
-  data.tar.gz: 5ed907c0cc481824640b8df911556c509616132e991ab1a9fecf347cb54c22398132990906bff31d6a6fac7cb1d5f7848b92509486dc3b9e620043867cd6178d
+  metadata.gz: b173a92b2b32bd8decf100e7c4f4f79b179ff4681839abbb4fcb390db1553d411d9ac552779279ee9d4ac8cef767d133f6780dbce8e8927c8ecb8a8560e8f965
+  data.tar.gz: fe7e30605e41b8e2980fa9760b7f722e33f8bf6f1f7b2dec5a7c56ced7c8dd648e345d4a90ac5b676e624d605582e54b8fdd602c67e10f21e0b0a8d6cd703aee

data/CHANGELOG.md

@@ -2,6 +2,14 @@
 All notable changes to this project will be documented in this file.
 This project adheres to [Semantic Versioning](http://semver.org/).
 
+## [0.1.0] - 2016-01-12
+### Added
+- n/a
+
+### Changed
+- All keys for each metric combined into a single redis hash
+  (this will null out old metrics)
+
 ## [0.0.1] - 2016-01-11
 ### Added
 - Initial release

data/README.md

@@ -8,6 +8,8 @@ RunningStat provides distributed redis-backed data buckets on which various stat
 
 The algorithm used is based on Knuth's TAOCP and is numerically stable; a brief writeup is available on Wikipedia under [Algorithms for calculating online variances](https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Online_algorithm).
 
+RunningStat supports pretty much all versions of ruby, including MRI 1.8.7, 1.9.3, 2.2, and JRuby 1.7/9k.
+
 Disclaimer: Because it uses atomic Lua scripting, RunningStat requires redis 2.6+.
 
 

data/lib/running_stat.rb

@@ -2,6 +2,7 @@ require 'running_stat/version'
 
 require 'redis'
 
+require 'running_stat/redis_backend'
 require 'running_stat/lua/push_datum'
 require 'running_stat/lua/variance'
 require 'running_stat/insufficient_data_error'
@@ -22,26 +23,26 @@ class RunningStat
 
   # Adds a piece of numerical data to the dataset's stats
   def push(datum)
-    redis.eval(Lua::PUSH_DATUM, [count_key, mean_key, sum_sq_diff_key], [Float(datum)])
+    redis.eval(Lua::PUSH_DATUM, [bucket_key], [Float(datum)])
   rescue ArgumentError => e
     raise InvalidDataError.new(e)  # datum was non-numerical
   end
 
   # Returns the number of data points seen, or 0 if the stat does not exist
   def cardinality
-    redis.get(count_key).to_i
+    redis.hget(bucket_key, RedisBackend::COUNT_FIELD).to_i
   end
 
   # Returns the arithmetic mean of data points seen, or 0.0 if non-existent
   def mean
-    redis.get(mean_key).to_f
+    redis.hget(bucket_key, RedisBackend::MEAN_FIELD).to_f
   end
 
   # Returns the sample variance of the dataset so far, or raises
   # an InsufficientDataError if insufficient data (< 2 datapoints)
   # has been pushed
   def variance
-    redis.eval(Lua::VARIANCE, [count_key, sum_sq_diff_key], []).to_f
+    redis.eval(Lua::VARIANCE, [bucket_key], []).to_f
   rescue Redis::CommandError => e
     raise InsufficientDataError.new(e)  # only CommandError possible
   end
@@ -55,7 +56,7 @@ class RunningStat
 
   # Resets the stat to reflect an empty dataset
   def flush
-    redis.del(count_key, mean_key, sum_sq_diff_key)
+    redis.del(bucket_key)
   end
 
   private
@@ -64,15 +65,7 @@ class RunningStat
     @redis || Redis.current
   end
 
-  def count_key
-    "#{BASE_KEY}:#{@data_bucket}:count"
-  end
-
-  def mean_key
-    "#{BASE_KEY}:#{@data_bucket}:mean"
-  end
-
-  def sum_sq_diff_key
-    "#{BASE_KEY}:#{@data_bucket}:sum_sq_diff"
+  def bucket_key
+    "#{BASE_KEY}:#{@data_bucket}"
   end
 end

data/lib/running_stat/lua/push_datum.rb

@@ -1,11 +1,10 @@
 require 'digest/sha1'
+require 'running_stat/redis_backend'
 
 class RunningStat
   module Lua
     # Keys:
-    #   count - the dataset cardinality
-    #   mean - the mean of the dataset so far
-    #   m2 - the running sum of the squares of the differences of the dataset
+    #   bucket - the hash corresponding to the dataset metric
     # Arguments:
     #   datum - the number to be added to the dataset
     # Effects:
@@ -13,17 +12,15 @@ class RunningStat
     # Returns:
     #   nothing
     PUSH_DATUM = <<-EOLUA
-      local count_key = KEYS[1]
-      local mean_key = KEYS[2]
-      local m2_key = KEYS[3]
+      local bucket_key = KEYS[1]
       local datum = ARGV[1]
 
-      local mean = tonumber(redis.call("GET", mean_key)) or 0.0
+      local mean = tonumber(redis.call("HGET", bucket_key, "#{RedisBackend::MEAN_FIELD}")) or 0.0
       local delta = datum - mean
 
-      local count = redis.call("INCR", count_key)
-      mean = redis.call("INCRBYFLOAT", mean_key, tostring(delta / count))
-      redis.call("INCRBYFLOAT", m2_key, tostring(delta * (datum - mean)))
+      local count = redis.call("HINCRBY", bucket_key, "#{RedisBackend::COUNT_FIELD}", 1)
+      mean = redis.call("HINCRBYFLOAT", bucket_key, "#{RedisBackend::MEAN_FIELD}", tostring(delta / count))
+      redis.call("HINCRBYFLOAT", bucket_key, "#{RedisBackend::M2_FIELD}", tostring(delta * (datum - mean)))
     EOLUA
 
     PUSH_DATUM_SHA1 = Digest::SHA1.hexdigest(PUSH_DATUM).freeze

data/lib/running_stat/lua/variance.rb

@@ -1,11 +1,11 @@
 require 'digest/sha1'
+require 'running_stat/redis_backend'
 require 'running_stat/insufficient_data_error'
 
 class RunningStat
   module Lua
     # Keys:
-    #   count - the dataset cardinality
-    #   m2 - the running sum of the squares of the differences of the dataset
+    #   bucket - the hash corresponding to the dataset metric
     # Arguments:
     #   nothing
     # Effects:
@@ -13,14 +13,14 @@ class RunningStat
     # Returns:
     #   the sample variance of the dataset, as a stringified float
     VARIANCE = <<-EOLUA
-      local count_key = KEYS[1]
-      local m2_key = KEYS[2]
+      local bucket_key = KEYS[1]
 
-      local count = tonumber(redis.call("GET", count_key)) or 0
+      local values = redis.call("HMGET", bucket_key, "#{RedisBackend::COUNT_FIELD}", "#{RedisBackend::M2_FIELD}")
+      local count = tonumber(values[1]) or 0
       if count < 2 then
         return redis.error_reply("#{InsufficientDataError::ERROR_STRING}")
       else
-        local m2 = tonumber(redis.call("GET", m2_key)) or 0.0
+        local m2 = tonumber(values[2]) or 0.0
         return tostring(m2 / (count - 1))
       end
     EOLUA

data/lib/running_stat/redis_backend.rb

@@ -0,0 +1,10 @@
+class RunningStat
+  module RedisBackend
+    # fields within the metric's redis hash
+    COUNT_FIELD = :c  # the dataset cardinality
+    MEAN_FIELD = :m  # the arithmetic mean of the dataset
+    M2_FIELD = :m2  # the running sum of the squares of the differences of the dataset
+
+    # TODO: implement global redis set/get
+  end
+end

data/lib/running_stat/version.rb

@@ -1,3 +1,3 @@
 class RunningStat
-  VERSION = '0.0.1'
+  VERSION = '0.1.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: running_stat
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.1.0
 platform: ruby
 authors:
 - Anuj Das
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2016-01-11 00:00:00.000000000 Z
+date: 2016-01-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: redis
@@ -89,6 +89,7 @@ files:
 - lib/running_stat/invalid_data_error.rb
 - lib/running_stat/lua/push_datum.rb
 - lib/running_stat/lua/variance.rb
+- lib/running_stat/redis_backend.rb
 - lib/running_stat/version.rb
 - running_stat.gemspec
 homepage: https://www.github.com/anujdas/running_stat