data_package 0.0.1

data/lib/data_package.rb

@@ -0,0 +1,17 @@
+require 'active_support/core_ext/hash'
+
+require "data_package/version"
+
+# Attribute Helpers
+require 'attr_helper'
+
+# Core Specification
+require 'data_package/base'
+require 'data_package/package'
+require 'data_package/field'
+require 'data_package/schema'
+require 'data_package/resource'
+require 'data_package/dialect'
+require 'data_package/person'
+require 'data_package/source'
+require 'data_package/license'

data/spec/attr_helper_spec.rb

@@ -0,0 +1,60 @@
+require 'spec_helper'
+require 'klass_helper'
+
+describe AttrHelper do
+  it "should initialize" do
+    obj = KlassHelper::BaseKlass.new(:name => 'myvalue')
+
+    obj.name.should == 'myvalue'
+
+    obj.attr_required?(:name).should == true
+    obj.attr_required?(:title).should == false
+    obj.attr_required?(:format).should == false
+
+    obj.missing_attributes.collect(&:name).should == [:data, :path, :url, :email, :web]
+    obj.required_attributes.collect(&:name).should == [:name, :data, :path, :url, :email, :web]
+
+    obj.attr_present?(:name).should == true
+    obj.attr_missing?(:name).should == false
+
+    obj.attr_present?(:data).should == false
+    obj.attr_missing?(:data).should == true
+  end
+
+  it "should inherit" do
+    obj = KlassHelper::ChildKlass.new(:name => 'myvalue')
+
+    obj.name.should == 'myvalue'
+
+    obj.attr_required?(:name).should == true
+    obj.attr_required?(:title).should == false
+    obj.attr_required?(:format).should == false
+
+    obj.missing_attributes.collect(&:name).should == [:data, :path, :url, :email, :web]
+    obj.required_attributes.collect(&:name).should == [:name, :data, :path, :url, :email, :web]
+
+    obj.attr_present?(:name).should == true
+    obj.attr_missing?(:name).should == false
+
+    obj.attr_present?(:data).should == false
+    obj.attr_missing?(:data).should == true
+  end
+
+  it "should allow mutual exclusion" do
+    obj = KlassHelper::ChildKlass.new(:name => 'myvalue')
+
+    obj.missing_attributes.collect(&:name).should == [:data, :path, :url, :email, :web]
+    obj.required_attributes.collect(&:name).should == [:name, :data, :path, :url, :email, :web]
+
+    obj.path = 'data.csv'
+    obj.missing_attributes.collect(&:name).should == [:email, :web]
+    obj.required_attributes.collect(&:name).should == [:name, :path, :email, :web]
+  end
+
+  it "should write attributes through the setters" do
+    obj = KlassHelper::ChildKlass.new(:name => 'myvalue', :title => 'mytitle')
+
+    obj.name.should == 'myvalue'
+    obj.title.should == 'mytitlechild'
+  end
+end

data/spec/attr_helpers/base_attr_spec.rb

@@ -0,0 +1,62 @@
+require 'spec_helper'
+
+class ValueObj
+  attr_accessor :name
+
+  def initialize(attrs = {})
+    @name = attrs[:name]
+  end
+
+  def is_name_required?
+    true
+  end
+
+  def serialized_name(name)
+    name + 'method'
+  end
+end
+
+describe AttrHelper::BaseAttr do
+  it "should initialize" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::BaseAttr.new(:name)
+
+    attribute.name.should == :name
+    attribute.key.should == 'name'
+    attribute.serialize.should == true
+  end
+
+  describe "Serialization" do
+    it "should serialize by default" do
+      valueObj = ValueObj.new(:name => 'value')
+      attribute = AttrHelper::BaseAttr.new(:name)
+
+      attribute.serializable?.should == true
+      attribute.serialized(valueObj).should == 'value'
+    end
+
+    it "should serialize with a proc" do
+      valueObj = ValueObj.new(:name => 'value')
+      attribute = AttrHelper::BaseAttr.new(:name, :serialize => Proc.new{|name| name + 'proc'})
+
+      attribute.serializable?.should == true
+      attribute.serialized(valueObj).should == 'valueproc'
+    end
+
+    it "should serialize with a symbol" do
+      valueObj = ValueObj.new(:name => 'value')
+      attribute = AttrHelper::BaseAttr.new(:name, :serialize => :serialized_name)
+
+      attribute.serializable?.should == true
+      attribute.serialized(valueObj).should == 'valuemethod'
+    end
+
+    it "should disable serialization" do
+      valueObj = ValueObj.new(:name => 'value')
+      attribute = AttrHelper::BaseAttr.new(:name, :serialize => false)
+
+      attribute.serializable?.should == false
+      attribute.serialized(valueObj).should == nil
+    end
+  end
+end

data/spec/attr_helpers/required_attr_spec.rb

@@ -0,0 +1,49 @@
+require 'spec_helper'
+
+class ValueObj
+  attr_accessor :name
+
+  def initialize(attrs = {})
+    @name = attrs[:name]
+  end
+
+  def is_name_required?
+    true
+  end
+end
+
+describe AttrHelper::RequiredAttr do
+  it "should initialize" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::RequiredAttr.new(:name)
+
+    attribute.name.should == :name
+    attribute.key.should == 'name'
+    attribute.serialize.should == true
+    attribute.required?(valueObj).should == true
+  end
+
+  it "should check for if requirement with a proc" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::RequiredAttr.new(:name, :if => Proc.new{|parent| true})
+    attribute.required?(valueObj).should == true
+  end
+
+  it "should check for if requirement with a symbol" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::RequiredAttr.new(:name, :if => :is_name_required?)
+    attribute.required?(valueObj).should == true
+  end
+
+  it "should check for unless requirement with a proc" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::RequiredAttr.new(:name, :unless => Proc.new{|parent| true})
+    attribute.required?(valueObj).should == false
+  end
+
+  it "should check for unless requirement with a symbol" do
+    valueObj = ValueObj.new(:name => 'value')
+    attribute = AttrHelper::RequiredAttr.new(:name, :unless => :is_name_required?)
+    attribute.required?(valueObj).should == false
+  end
+end

data/spec/attr_helpers/serialization_spec.rb

@@ -0,0 +1,9 @@
+require 'spec_helper'
+require 'klass_helper'
+
+describe AttrHelper do
+  it "should serialize" do
+    obj = KlassHelper::BaseKlass.new(:name => 'myvalue')
+    obj.to_json.should == Yajl::Encoder.encode({:name => 'myvalue'}, :pretty => true)
+  end
+end

data/spec/data_package/base_spec.rb

@@ -0,0 +1,10 @@
+require 'spec_helper'
+
+describe DataPackage::Base do
+  it "should list required and optional attributes" do
+    base = DataPackage::Base.new
+
+    base.required_attributes.should == []
+    base.optional_attributes.should == []
+  end
+end

data/spec/data_package/dialect_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe DataPackage::Dialect do
+  let(:json) {
+    {
+      "delimiter" => "\t",
+      "doublequote" => false,
+      "lineterminator" => "\n",
+      "quotechar" => "\"",
+      "skipinitialspace" => true
+    }
+  }
+
+  it "should initialize with defaults" do
+    dialect = DataPackage::Dialect.new
+
+    dialect.delimiter.should == ','
+    dialect.double_quote.should == false
+    dialect.line_terminator.should == "\n"
+    dialect.quote_char.should == "\""
+    dialect.skip_initial_space.should == false
+  end
+
+  it "should initialize and serialize" do
+    dialect = DataPackage::Dialect.new(json)
+
+    dialect.delimiter.should == "\t"
+    dialect.double_quote.should == false
+    dialect.line_terminator.should == "\n"
+    dialect.quote_char.should == "\""
+    dialect.skip_initial_space.should == true
+
+    dialect.to_json.should == Yajl::Encoder.encode(json, :pretty => true)
+  end
+end

data/spec/data_package/field_spec.rb

@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+describe DataPackage::Field do
+  let(:json) {
+    {
+      'name' => 'income',
+      'type' => 'number',
+      'title' => 'Personal Income',
+      'description' => 'My Desc'
+    }
+  }
+
+  it "should initialize and serialize" do
+    field = DataPackage::Field.new(json)
+
+    field.type.should == :number
+    field.name.should == json['name']
+    field.title.should == json['title']
+    field.description.should == json['description']
+
+    field.to_hash.should == json
+    field.to_json.should == Yajl::Encoder.encode(field.to_hash, :pretty => true)
+  end
+
+  it "should require a name" do
+    field = DataPackage::Field.new
+    field.missing_attributes.collect(&:name).should == [:name]
+  end
+end

data/spec/data_package/license_spec.rb

@@ -0,0 +1,18 @@
+require 'spec_helper'
+
+describe DataPackage::License do
+  let(:json) {
+    {
+      'id' => 'PDDL',
+      'url' => 'http://opendatacommons.org/licenses/pddl/'
+    }
+  }
+
+  it "should initialize and serialize" do
+    license = DataPackage::License.new(json)
+
+    license.id.should == 'PDDL'
+    license.url.should == 'http://opendatacommons.org/licenses/pddl/'
+    license.to_json.should == Yajl::Encoder.encode(json, :pretty => true)
+  end
+end

data/spec/data_package/package_spec.rb

@@ -0,0 +1,59 @@
+require 'spec_helper'
+
+describe DataPackage::Package do
+  let(:tmpdir) {
+    Dir.mktmpdir
+  }
+
+  let(:package_path) {
+    File.join(base_path, 'package')
+  }
+
+  it "should init" do
+    package = DataPackage::Package.init(tmpdir, 'mypackage')
+
+    package.name.should == 'mypackage'
+    package.version.should == '0.0.1'
+    package.resources.should == nil
+    package.base_path.should == tmpdir
+  end
+
+  it "should raise on open if the package file doesn't exist" do
+    expect { package = DataPackage::Package.open(tmpdir) }.to raise_error
+  end
+
+  it "should check if a package exists" do
+    DataPackage::Package.exist?(package_path).should == true
+  end
+
+  it "should initialize and serialize" do
+    package = DataPackage::Package.open(package_path)
+
+    package.name.should == 'standard'
+    package.title.should == 'Standard Data Package'
+    package.version.should == '0.0.1'
+
+    package.resources.length.should == 1
+
+    package.resources.first.schema.fields.length.should == 10
+    package.resources.first.schema.primary_key.should == ['id']
+
+    package.sources.length.should == 1
+    package.licenses.length.should == 1
+    package.maintainers.length.should == 1
+    package.contributors.length.should == 1
+  end
+
+  it "should save the package" do
+    package = DataPackage::Package.open(package_path)
+
+    package.base_path = tmpdir
+    package.name = "tested_package"
+
+    package.save
+
+    reopened = DataPackage::Package.open(tmpdir)
+
+    reopened.name.should == "tested_package"
+  end
+end

data/spec/data_package/person_spec.rb

@@ -0,0 +1,21 @@
+require 'spec_helper'
+
+describe DataPackage::Person do
+  let(:json) {
+    {
+      'name' => 'Josh F',
+      'email' => 'josh@modeanalytics.com',
+      'web' => 'http://www.modeanalytics.com/josh'
+    }
+  }
+
+  it "should initialize and serialize" do
+    person = DataPackage::Person.new(json)
+
+    person.name.should == 'Josh F'
+    person.email.should == 'josh@modeanalytics.com'
+    person.web.should == 'http://www.modeanalytics.com/josh'
+
+    person.to_json.should == Yajl::Encoder.encode(json, :pretty => true)
+  end
+end

data/spec/data_package/resource_spec.rb

@@ -0,0 +1,57 @@
+require 'spec_helper'
+require 'resource_helper'
+
+describe DataPackage::Resource do
+  it "should initialize and serialize" do    
+    resource = DataPackage::Resource.new(base_path, standard_resource)
+
+    resource.format.should == 'csv'
+    resource.name.should == 'standard'
+    resource.path.should == 'standard.csv'
+
+    resource.schema.fields.length.should == 10
+    resource.schema.primary_key.should == ['id']
+
+    resource.to_hash.should == standard_resource.merge('format' => 'csv', 'dialect' => DataPackage::Dialect.new.to_hash)
+  end
+
+  describe "#each_row" do
+    it "should enumerate over inline data" do
+      resource = DataPackage::Resource.new(base_path, inline_resource)
+
+      row_count = 0
+      resource.each_row do |row|
+        row_count += 1
+      end
+      row_count.should == 3
+    end
+
+    it "should enumerate over local file data" do
+      resource = DataPackage::Resource.new(base_path, standard_resource)
+
+      row_count = 0
+      resource.each_row do |row|
+        row_count += 1
+      end
+      row_count.should == 10
+    end
+
+    it "should not enumerate URL data" do
+      resource = DataPackage::Resource.new(base_path, http_resource)
+      expect{ resource.each_row{|row| nil} }.to raise_error
+    end
+
+    it "should raise if the data format is unexpected" do
+      resource = DataPackage::Resource.new(base_path, standard_resource.merge('format' => 'unknown'))
+      expect{ resource.each_row{|row| nil} }.to raise_error
+    end
+
+    it "should raise if data, path or url aren't present" do
+      invalid = standard_resource.dup
+      invalid.delete('path')
+
+      resource = DataPackage::Resource.new(base_path, invalid)
+      expect{ resource.each_row{|row| nil} }.to raise_error
+    end
+  end
+end

data/spec/data_package/schema_spec.rb

@@ -0,0 +1,25 @@
+require 'spec_helper'
+
+describe DataPackage::Schema do
+  let(:json) {
+    {
+      'fields' => [
+        {
+          'name' => 'income',
+          'type' => 'number'
+        }
+      ],
+
+      'primaryKey' => ['income']
+    }
+  }
+
+  it "should initialize" do
+    schema = DataPackage::Schema.new(json)
+
+    schema.fields.length.should == 1
+    schema.fields.first.name.should == 'income'
+    schema.primary_key.should == ['income']
+    schema.to_json.should == Yajl::Encoder.encode(json, :pretty => true)
+  end
+end

data/spec/data_package/source_spec.rb

@@ -0,0 +1,22 @@
+require 'spec_helper'
+
+describe DataPackage::Source do
+  let(:json) {
+    {
+      'name' => 'Josh F',
+      'email' => 'josh@modeanalytics.com',
+      'web' => 'http://www.modeanalytics.com/josh'
+    }
+  }
+
+  it "should initialize and serialize" do
+    person = DataPackage::Source.new(json)
+
+    person.name.should == 'Josh F'
+    person.email.should == 'josh@modeanalytics.com'
+    person.web.should == 'http://www.modeanalytics.com/josh'
+
+    person.to_hash.should == json
+    person.to_json.should == Yajl::Encoder.encode(person.to_hash, :pretty => true)
+  end
+end

data/spec/fixtures/package/datapackage.json

@@ -0,0 +1,94 @@
+{
+  "name": "standard",
+  "version": "0.0.1",
+  "title": "Standard Data Package",
+  
+  "sources": [
+    {
+      "name": "Josh F",
+      "email": "josh@modeanalytics.com",
+      "web": "http://www.modeanalytics.com/josh"
+    }
+  ],
+
+  "licenses": [
+    {
+      "id": "PDDL",
+      "url": "http://opendatacommons.org/licenses/pddl/"
+    }
+  ],
+
+  "maintainers": [
+    {
+      "name": "Josh F",
+      "email": "josh@modeanalytics.com",
+      "web": "http://www.modeanalytics.com/josh"
+    }
+  ],
+
+  "contributors": [
+    {
+      "name": "Josh F",
+      "email": "josh@modeanalytics.com",
+      "web": "http://www.modeanalytics.com/josh"
+    }
+  ], 
+
+  "resources": [
+    {
+      "name": "standard",
+      "path": "standard.csv",
+
+      "dialect": {
+        "delimiter": ","
+      },
+
+      "schema": {
+        "fields": [
+          {
+            "name": "id",
+            "type": "integer"
+          },
+          {
+            "name": "first_name",
+            "type": "string"
+          },
+          {
+            "name": "last_name",
+            "type": "string"
+          },
+          {
+            "name": "email",
+            "type": "string"
+          },
+          {
+            "name": "country",
+            "type": "string"
+          },
+          {
+            "name": "ip_address",
+            "type": "string"
+          },
+          {
+            "name": "amount",
+            "type": "number"
+          },
+          {
+            "name": "active",
+            "type": "boolean"
+          },
+          {
+            "name": "activated_at",
+            "type": "datetime"
+          },
+          {
+            "name": "address",
+            "type": "string"
+          }
+        ],
+
+        "primaryKey": ["id"]
+      }
+    }
+  ]
+}

data/spec/fixtures/package/standard.csv

@@ -0,0 +1,11 @@
+id,first_name,last_name,email,country,ip_address,amount,active,activated_at,address
+1,Todd,Hamilton,thamilton@plambee.edu,Norfolk Island,66.157.128.241,$7.72,true,10/3/1955,0582 Dwight Street
+2,Melissa,Kelly,mkelly@twinte.name,Singapore,204.221.167.233,$4.33,false,7/9/2013,06 Cardinal Crossing
+3,Donald,Wheeler,dwheeler@edgeify.mil,Madagascar,34.201.104.193,$2.92,true,12/12/1993,4 Del Sol Hill
+4,Ruby,Hall,rhall@cogilith.com,USSR,237.243.109.67,$8.27,false,12/15/1975,7 Ramsey Avenue
+5,Jessica,Cole,jcole@shuffletag.info,Cyprus,25.40.138.137,$8.16,false,6/2/1939,8142 Novick Hill
+6,Doris,Nelson,dnelson@zoombox.biz,Svalbard and Jan Mayen Islands,233.43.155.229,$6.26,false,5/23/1948,596 Veith Road
+7,Robert,Hansen,rhansen@miboo.edu,Ghana,41.194.33.211,$4.90,true,9/28/1999,529 Oak Pass
+8,Matthew,Freeman,mfreeman@midel.name,Sudan,53.186.162.65,$6.63,false,5/24/1996,70682 Declaration Center
+9,Julia,Nelson,jnelson@skajo.net,Vatican City State (Holy See),249.49.124.178,$9.80,true,1/26/1940,96 Hermina Lane
+10,Wanda,Palmer,wpalmer@ntags.biz,Indonesia,0.200.163.200,$5.89,false,1/1/1959,26837 Donald Trail

data/spec/fixtures/standard.csv

@@ -0,0 +1,11 @@
+id,first_name,last_name,email,country,ip_address,amount,active,activated_at,address
+1,Todd,Hamilton,thamilton@plambee.edu,Norfolk Island,66.157.128.241,$7.72,true,10/3/1955,0582 Dwight Street
+2,Melissa,Kelly,mkelly@twinte.name,Singapore,204.221.167.233,$4.33,false,7/9/2013,06 Cardinal Crossing
+3,Donald,Wheeler,dwheeler@edgeify.mil,Madagascar,34.201.104.193,$2.92,true,12/12/1993,4 Del Sol Hill
+4,Ruby,Hall,rhall@cogilith.com,USSR,237.243.109.67,$8.27,false,12/15/1975,7 Ramsey Avenue
+5,Jessica,Cole,jcole@shuffletag.info,Cyprus,25.40.138.137,$8.16,false,6/2/1939,8142 Novick Hill
+6,Doris,Nelson,dnelson@zoombox.biz,Svalbard and Jan Mayen Islands,233.43.155.229,$6.26,false,5/23/1948,596 Veith Road
+7,Robert,Hansen,rhansen@miboo.edu,Ghana,41.194.33.211,$4.90,true,9/28/1999,529 Oak Pass
+8,Matthew,Freeman,mfreeman@midel.name,Sudan,53.186.162.65,$6.63,false,5/24/1996,70682 Declaration Center
+9,Julia,Nelson,jnelson@skajo.net,Vatican City State (Holy See),249.49.124.178,$9.80,true,1/26/1940,96 Hermina Lane
+10,Wanda,Palmer,wpalmer@ntags.biz,Indonesia,0.200.163.200,$5.89,false,1/1/1959,26837 Donald Trail

data/spec/klass_helper.rb

@@ -0,0 +1,41 @@
+module KlassHelper
+  class BaseKlass
+    include AttrHelper
+    include AttrHelper::Serialization
+    
+    attr_required :name
+    attr_optional :title
+    attr_optional :format
+
+    attr_required :data, :if => Proc.new{ |o| o.path.nil? && o.url.nil?  }
+    attr_required :path, :if => Proc.new{ |o| o.data.nil? && o.url.nil?  }
+    attr_required :url,  :if => Proc.new{ |o| o.data.nil? && o.path.nil? }
+
+    attr_required :dialect, :if => :format_requires_dialect?
+
+    attr_required :email, :unless => :empty_name?
+    attr_required :web, :unless => Proc.new{ |o| o.name.nil? }
+
+    def initialize(attrs = {})
+      write_attributes(attrs)
+    end
+
+    def title=(value)
+      write_attribute(:title, value + 'test')
+    end
+
+    def format_requires_dialect?
+      format == 'csv'
+    end
+
+    def empty_name?
+      name.nil?
+    end
+  end
+
+  class ChildKlass < BaseKlass
+    def title=(value)
+      write_attribute(:title, value + 'child')
+    end
+  end
+end