quandl_data 1.3.2 → 1.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: fa65494367e714daf18edb29fb3fc3e216c265a9
-  data.tar.gz: 4def3b647bd730baefd225c93a4468810629ba94
+  metadata.gz: b218225ec95c6b4ab6d1a6aca2e0123c723cbc05
+  data.tar.gz: 1b7650907d85479968f9e9ef5d53ff27476843df
 SHA512:
-  metadata.gz: 7b597c31ad9d69bbdd77452e762aeec79f51706f283b3d267d5f7abd7d797d62f4b7747dcb5938e8d40ed6957a29eaaa1cd93f3878d71a26f9c679b47b5662a1
-  data.tar.gz: ab90dcb57829b15f69b798c947b770e10fcf56e836f3a8eba8defbc603dd9d7171560d12ebd9fe7e3d7da6c0bfb552249a734f7d4b1b201340470ca42f292da8
+  metadata.gz: f11b51e91b3f270084abe1e7e4c017b0ac1261db46bd4f6d3e574c83552f40990741cd4de08010894960317c4f70edb59dfe0c000c8ed37601b099b4e3a58a9b
+  data.tar.gz: 8e3bf9e9cd563ab2e3a0798bf45ee240a75f8274be95c6d83c6839396121baca32584cdc84b01b7d05d8d18db6f50d01f103c1e3c743d3fc9ee37cf132f9d90d

data/UPGRADE.md

@@ -1,3 +1,9 @@
+## 1.3.3
+
+* integrate Babelfish.clean as part of Quandl::Data.clean method.
+* Remove Quandl::Babelfish::Data
+
+
 ## 1.3.2
 
 * readd quandl_babelfish as a dependency

data/lib/quandl/data.rb

@@ -8,7 +8,6 @@ require "active_support/core_ext/object"
 require "quandl/operation"
 
 require 'quandl/babelfish'
-require 'quandl/babelfish/data'
 
 require 'quandl/error/unknown_date_format'
 

data/lib/quandl/data/attributes.rb

@@ -4,29 +4,16 @@ module Attributes
   
   extend ActiveSupport::Concern
   
-  module ClassMethods
-    
-    def with_type(type)
-      type = type.to_s.downcase.to_sym
-      case type
-      when :dirty, :babelfish then Quandl::Babelfish::Data
-      else 
-        Quandl::Data 
-      end
-    end
-    
-  end
-  
   def initialize(*args)
     self.pristine_data = args.first
     self.attributes = args.extract_options!
   end
   
   def headers=(value)
-    # only assign arrays
     @headers = value.flatten if value.kind_of?(Array)
   end
   def headers
+    ensure_data_is_cleaned
     @headers
   end
   def headers?

data/lib/quandl/data/cleaning.rb

@@ -13,24 +13,59 @@ module Cleaning
   def cleaned?
     cleaned == true
   end
+  
+  def ensure_data_is_cleaned
+    data_array unless cleaned?
+  end
 
   protected
   
   def clean(data)
+    # check if data is dirty
+    requires_cleaning = ensure_data_requires_cleaning(data)
+    # short ciruit unless data is dirty
+    return requires_cleaning unless requires_cleaning == true
+    # ensure we're dealing with an array
+    data = ensure_data_is_an_array(data)
+    # clean with either format or babelfish
+    known_format?( data ) ? clean_with_format(data) : clean_with_babelfish(data)
+  end
+  
+  def ensure_data_requires_cleaning(data)
     # skip cleaning if already clean
     return data if data.kind_of?(Array) && cleaned?
     # Quandl::Data is already clean, but to avoid errors extract internal array
-    return data.data_array if data.kind_of?(Quandl::Data)
+    return data.to_date.to_a if data.kind_of?(Quandl::Data)
     # Return empty array if given empty string, nil, etc.
     return [] if data.blank?
+    # data requires cleaning
+    true
+  end
+  
+  def ensure_data_is_an_array(data)
+    # Hash needs conversion to array
+    data = Quandl::Data::Format.hash_to_array( data )
     # String needs conversion to array
+    data = Quandl::Data::Format.csv_to_array( data )
+    data
+  end
+  
+  def known_format?( data )
+    Format.recognized_date?( data[0][0] )
+  end
+  
+  def clean_with_format(data)
     data = Format.parse( data )
-    # mark data as clean
     cleaned!
-    # return data
     data
   end
-
+  
+  def clean_with_babelfish(data)
+    data, self.headers = Quandl::Babelfish.clean(data)
+    cleaned!
+    data
+  end
+  
   def cleaned!
     self.cleaned = true
   end

data/lib/quandl/data/format.rb

@@ -57,6 +57,12 @@ class Format
       data
     end
     
+    def recognized_date?( value )
+      return true if value.is_a?(Date)
+      return true if value.is_a?(String) && value =~ /^[0-9]{4}\-[0-9]{2}\-[0-9]{2}$/
+      false
+    end
+    
     def date_format?(data)
       value = data[0][0]
       # julian date?

data/lib/quandl/data/version.rb

@@ -1,5 +1,5 @@
 module Quandl
 class Data
-  VERSION = "1.3.2"
+  VERSION = "1.3.3"
 end
 end

data/spec/lib/quandl/data_spec.rb

@@ -41,4 +41,22 @@ describe Quandl::Data do
     its("#{o1}!"){ should eq subject }
   end
   
+  context "babelfish" do
+  
+    let(:csv) { "Date, Column 1, Column 2, C3, C4, C5, C6, C7\n 2012-03-07,,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-06,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-05,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-04,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-02-29,,69.75,69.75,69.75,0.0,0.0,0.0\n2012-02-28,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n" }
+  
+    let(:data){ Quandl::Data.new(csv) }
+    subject{ data }
+  
+    let(:expected_data){ [['2012-03-07',nil,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-06',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-05',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-04',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-02-29',nil,69.75,69.75,69.75,0.0,0.0,0.0],['2012-02-28',69.75,69.75,69.75,69.75,0.0,0.0,0.0]] }
+    let(:expected_headers){ ['Date', 'Column 1', 'Column 2', 'C3', 'C4', 'C5', 'C6', 'C7'] }
+  
+    its(:headers){ should eq expected_headers }  
+    its('clone.headers'){ should eq expected_headers }
+    its('to_jd.headers'){ should eq expected_headers }
+  
+    its(:to_date_str){ should eq expected_data }
+  
+  end
+  
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: quandl_data
 version: !ruby/object:Gem::Version
-  version: 1.3.2
+  version: 1.3.3
 platform: ruby
 authors:
 - Blake Hilscher
@@ -153,7 +153,6 @@ files:
 - README.md
 - Rakefile
 - UPGRADE.md
-- lib/quandl/babelfish/data.rb
 - lib/quandl/data.rb
 - lib/quandl/data/attributes.rb
 - lib/quandl/data/cleaning.rb
@@ -167,7 +166,6 @@ files:
 - lib/quandl/fabricate/data.rb
 - quandl_data.gemspec
 - spec/lib/fabricate/data_spec.rb
-- spec/lib/quandl/babelfish/data_spec.rb
 - spec/lib/quandl/data/enumerator_spec.rb
 - spec/lib/quandl/data/format_spec.rb
 - spec/lib/quandl/data/operations_spec.rb
@@ -199,7 +197,6 @@ specification_version: 4
 summary: For interfacing with data
 test_files:
 - spec/lib/fabricate/data_spec.rb
-- spec/lib/quandl/babelfish/data_spec.rb
 - spec/lib/quandl/data/enumerator_spec.rb
 - spec/lib/quandl/data/format_spec.rb
 - spec/lib/quandl/data/operations_spec.rb

data/lib/quandl/babelfish/data.rb

@@ -1,37 +0,0 @@
-module Quandl
-module Babelfish
-
-class Data < Quandl::Data
-  
-  def initialize(*args)
-    super(*args)
-    # clean data on initialize
-    self.data_array
-  end
-  
-  
-  protected
-  
-  def clean(data)
-    # skip cleaning if already clean
-    return data if data.kind_of?(Array) && cleaned?
-    # Quandl::Data is already clean, but to avoid errors extract internal array
-    return data.data_array if data.kind_of?(Quandl::Data)
-    # Return empty array if given empty string, nil, etc.
-    return [] if data.blank?
-    # Hash needs conversion to array
-    data = Quandl::Data::Format.hash_to_array( data )
-    # String needs conversion to array
-    data = Quandl::Data::Format.csv_to_array( data )
-    # Babelfish cleaner
-    data, self.headers = Quandl::Babelfish.clean(data)
-    # mark data as clean
-    cleaned!
-    # return data
-    data
-  end
-  
-end
-
-end
-end

data/spec/lib/quandl/babelfish/data_spec.rb

@@ -1,19 +0,0 @@
-require 'spec_helper'
-
-describe Quandl::Babelfish::Data do
-  
-  let(:csv) { "Date, Column 1, Column 2, C3, C4, C5, C6, C7\n 2012-03-07,,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-06,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-05,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-03-04,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n2012-02-29,,69.75,69.75,69.75,0.0,0.0,0.0\n2012-02-28,69.75,69.75,69.75,69.75,0.0,0.0,0.0\n" }
-  
-  let(:data){ Quandl::Babelfish::Data.new(csv) }
-  subject{ data }
-  
-  let(:expected_data){ [['2012-03-07',nil,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-06',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-05',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-03-04',69.75,69.75,69.75,69.75,0.0,0.0,0.0],['2012-02-29',nil,69.75,69.75,69.75,0.0,0.0,0.0],['2012-02-28',69.75,69.75,69.75,69.75,0.0,0.0,0.0]] }
-  let(:expected_headers){ ['Date', 'Column 1', 'Column 2', 'C3', 'C4', 'C5', 'C6', 'C7'] }
-  
-  its(:headers){ should eq expected_headers }  
-  its('clone.headers'){ should eq expected_headers }
-  its('to_jd.headers'){ should eq expected_headers }
-  
-  its(:to_date_str){ should eq expected_data }
-  
-end