remi 0.2.5 → 0.2.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a68e65a6146a37cea022e1f7cbf18574e16905f7
-  data.tar.gz: c944f32038d24deff54396591ca98a64b69ba296
+  metadata.gz: 2c802335b8494b30ff89e4c31b1ce4df34b2fc8a
+  data.tar.gz: b405fbefb668bf07db5bcbb4a0cf6d8550658f13
 SHA512:
-  metadata.gz: c016ce1f58bb37da6214c29263ddee66c3b46c9557539210be25517c2676ffc0abd6a3c3fb5b6406290a350c280a277bce0a4aedb14856266d423169f650eeaf
-  data.tar.gz: 4dc14a6c9653c12dea6ffaaef74ab7d1acc116308ec99e181d6756ae63db8d5a740f6c72e3d8b258f52fce160f8c3c2892405f96c82ce3985a3f11753db690d5
+  metadata.gz: e2380bbb4cc87d67cfdb554763f8614cfcc791ceb3b70f711ed2fb975dcaf142438f2937453734479a43d3735ea61c8bcb4e093e3f380b5ae107e2dba5fe9522
+  data.tar.gz: 56f42c6b3608157959713478fbeee5cea4898238a6af176ab0b632cc0bb75ec5c60c66fe0ad60b45f6483aee09bb2943ddfabe586a88ef223ceaf9349eeded07

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    remi (0.2.4)
+    remi (0.2.6)
       activesupport (~> 4.2)
       bond (~> 0.5)
       cucumber (~> 2.1)

data/lib/remi/project/features/aggregate.feature

@@ -4,20 +4,39 @@ Feature: Tests the aggregate refinement to the Daru library
     Given the job is 'Aggregate'
     And the job source 'Source Data'
     And the job target 'Target Data'
+    And the job target 'Multigroup Target Data'
 
     And the source 'Source Data'
-    And the target 'Target Data'
+
 
   Scenario: The aggregator should find the minimum year for each 'Alpha'
-    Given the following example record for 'Source Data':
-      | Alpha | Year | something |
-      | a     | 2016 | 1 |
-      | a     | 2018 | 1 |
-      | b     | 2016 | 2 |
-      | b     | 2010 | 3 |
-      | a     | 2017 | 4 |
+    Given the target 'Target Data'
+    And the following example record for 'Source Data':
+      | Alpha | Beta | Year |
+      | a     | aa   | 2016 |
+      | a     | aa   | 2018 |
+      | b     | bb   | 2016 |
+      | b     | bb   | 2010 |
+      | a     | ab   | 2017 |
     And the following example record called 'expected result':
       | Alpha | Year |
       | a     | Group a has a minimum value of 2016 |
       | b     | Group b has a minimum value of 2010 |
     Then the target should match the example 'expected result'
+
+
+  Scenario: The aggregator should find the minimum year for each 'Alpha'
+    Given the target 'Multigroup Target Data'
+    And the following example record for 'Source Data':
+      | Alpha | Beta | Year |
+      | a     | aa   | 2016 |
+      | a     | aa   | 2018 |
+      | b     | bb   | 2016 |
+      | b     | bb   | 2010 |
+      | a     | ab   | 2017 |
+    And the following example record called 'expected result':
+      | Alpha | Beta | Year |
+      | a     | aa   | Group ["a", "aa"] has a minimum value of 2016 |
+      | a     | ab   | Group ["a", "ab"] has a minimum value of 2017 |
+      | b     | bb   | Group ["b", "bb"] has a minimum value of 2010 |
+    Then the target should match the example 'expected result'

data/lib/remi/project/jobs/aggregate_job.rb

@@ -6,15 +6,27 @@ class AggregateJob
 
   define_source :source_data, Remi::DataSource::DataFrame
   define_target :target_data, Remi::DataTarget::DataFrame
+  define_target :multigroup_target_data, Remi::DataTarget::DataFrame
 
   define_transform :main, sources: :source_data, targets: :target_data do
-
     mymin = lambda do |field, df, group_key, indicies|
       values = indicies.map { |idx| df.row[idx][field] }
       "Group #{group_key} has a minimum value of #{values.min}"
     end
 
+    # Daru groups don't use the index of the dataframe when returning groups (WTF?).
+    # Instead they return the position of the record in the dataframe.  Here, we
+    # shift the indexes which causes a failure if this artifact is not handled
+    # properly in the aggregate function
+    source_data.df.index = Daru::Index.new(1.upto(source_data.df.size).to_a)
+
     target_data.df = source_data.df.aggregate(by: :alpha, func: mymin.curry.(:year)).detach_index
     target_data.df.vectors = Daru::Index.new([:alpha, :year])
+
+    multigroup_target_data.df = source_data.df.aggregate(by: [:alpha,:beta], func: mymin.curry.(:year)).detach_index
+    multigroup_target_data.df.vectors = Daru::Index.new([:alpha_beta, :year])
+
+
+
   end
 end

data/lib/remi/refinements/daru.rb

@@ -29,8 +29,8 @@ module Remi
         # Example:
         #   df = Daru::DataFrame.new( { a: ['a','a','a','b','b'], year: ['2018','2015','2019', '2014', '2013'] })
         #
-        #   mymin = lambda do |field, df, group_key, indicies|
-        #     values = indicies.map { |idx| df.row[idx][field] }
+        #   mymin = lambda do |field, df, group_key, indices|
+        #     values = indices.map { |idx| df.row[idx][field] }
         #     "Group #{group_key} has a minimum value of #{values.min}"
         #   end
         #
@@ -40,10 +40,14 @@ module Remi
         # Returns a Daru::Vector.
         def aggregate(by:, func:)
           grouped = self.group_by(by)
+          df_indices = self.index.to_a
           ::Daru::Vector.new(
-            grouped.groups.reduce({}) do |h, (key, indicies)|
+            grouped.groups.reduce({}) do |h, (key, indices)|
+              # Daru groups don't use the index of the dataframe when returning groups (WTF?).
+              # Instead they return the position of the record in the dataframe.  Here, we
+              group_df_indices = indices.map { |v| df_indices[v] }
               group_key = key.size == 1 ? key.first : key
-              h[group_key] = func.(self, group_key, indicies)
+              h[group_key] = func.(self, group_key, group_df_indices)
               h
             end
           )

data/lib/remi/version.rb

@@ -1,3 +1,3 @@
 module Remi
-  VERSION = '0.2.5'
+  VERSION = '0.2.6'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: remi
 version: !ruby/object:Gem::Version
-  version: 0.2.5
+  version: 0.2.6
 platform: ruby
 authors:
 - Sterling Paramore