ibge 0.2.0 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 80232f32e1ca3fe301c1abfe7a5a87981df49c71
-  data.tar.gz: bc5e203e2309aa62d263418a9b5d4fec2807b60e
+  metadata.gz: 6ed56fa3c4b240472a91542f75c8a216f78ce28c
+  data.tar.gz: 9e49eb09876bbd6c12d1ae6778303b227c7ca68d
 SHA512:
-  metadata.gz: d51d20081b0f09a6096f3f881d77bb6468528fbf604a85c9ff042390abc875ca3d4921ea046e26ac4e0d6d5d73986ef6b9a93e87377211c1b7744ce9898231f8
-  data.tar.gz: 9c5c550044bd2d2cff27b079b97ec40d1ef4056ef48419b8e707e62f9cc2057ec857e4d9cdd3df1531d048f8e8049fe24b2312a62a56b8dcc1734157e46787a2
+  metadata.gz: 95c9383f0d6f2cdc4c8a79a7847159408ab6f1acad4e92027f8bcf41837d5a23efc6cb878c619c7a1ed6ec3160372a81d24f326e3372bde43b290293e4a0119c
+  data.tar.gz: 23cb3405ae6b2be14e7cd5c0afad076947b445fe62341b6c778322ae9bd05ba77cd23a5007fc361c59162c833ae9fd5e27359be1299addf37296162e7523767a

data/lib/ibge.rb

@@ -6,13 +6,14 @@ module Ibge
   DATA_PATH = File.expand_path(File.join(__dir__, '..', 'data'))
 
   module Reader
-    autoload :Base,         'ibge/reader/base'
-    autoload :State,        'ibge/reader/state'
-    autoload :MesoRegion,   'ibge/reader/meso_region'
-    autoload :MicroRegion,  'ibge/reader/micro_region'
-    autoload :Municipality, 'ibge/reader/municipality'
-    autoload :District,     'ibge/reader/district'
-    autoload :SubDistrict,  'ibge/reader/sub_district'
-    autoload :Neighborhood, 'ibge/reader/neighborhood'
+    autoload :Helpers, "ibge/reader/helpers"
+    autoload :Base, "ibge/reader/base"
+    autoload :State, "ibge/reader/state"
+    autoload :MesoRegion, "ibge/reader/meso_region"
+    autoload :MicroRegion, "ibge/reader/micro_region"
+    autoload :Municipality, "ibge/reader/municipality"
+    autoload :District, "ibge/reader/district"
+    autoload :SubDistrict, "ibge/reader/sub_district"
+    autoload :Neighborhood, "ibge/reader/neighborhood"
   end
 end

data/lib/ibge/reader/base.rb

@@ -3,30 +3,7 @@ require 'spreadsheet'
 module Ibge
   module Reader
     class Base
-      class << self
-        SOURCES = {
-          dtb: 'DTB_2014_subdistrito.xls',
-          sidra: 'sidra.xls'
-        }
-
-        attr_reader :defined_columns, :selected_source
-
-        def columns(columns = {})
-          @defined_columns = columns
-        end
-
-        def source(selected_source = :dtb)
-          unless SOURCES.keys.include?(selected_source)
-            raise ArgumentError, "Source must be one of those '#{SOURCES.keys.flatten.join(', ')}'"
-          end
-
-          @selected_source = selected_source
-        end
-
-        def sources
-          SOURCES
-        end
-      end
+      extend Ibge::Reader::Helpers
 
       attr_reader :filename
 
@@ -35,37 +12,102 @@ module Ibge
       end
 
       def initialize(options = {})
-        @filename = options.fetch(:filename, default_filename)
+        @spreadsheet = {}
+        @rows = {}
+        @selected_sheet = {}
+
+        self.selected_source = self.class.selected_source
+
+        unless options.fetch(:filename, nil).nil?
+          filename = options.fetch(:filename)
+
+          if filename.is_a?(Hash)
+            sources_filename = options.fetch(:filename).keys
+
+            sources_filename.each do |source_filename|
+              self.class.valid_informed_source(source_filename)
+            end
+          else
+            raise ArgumentError,
+                  "Filename must be a hash with the source as key"
+          end
+        end
+
+        default_filename = default_filename_to(selected_source)
+
+        @filename = options.fetch(:filename, "#{DATA_PATH}/#{default_filename}")
       end
 
       def read
-        transformed_rows = rows[1..-1].map do |row|
-          transform(row.compact.values_at(*column_range), column_range)
+        transformed_rows = []
+        previous_column_range = nil
+
+        if multiple_sources?
+          loaded_sources = defined_columns.keys.sort
+
+          loaded_sources.each do |loaded_source|
+            self.selected_source = loaded_source
+            @filename = default_filename_to(loaded_source)
+
+            if transformed_rows == []
+              transformed_rows = rows[1..-1].map do |row|
+                transform(row.compact.values_at(*column_range), column_range)
+              end
+
+              transformed_rows = transformed_rows.uniq
+
+              previous_column_range = column_range
+            else
+              transformed_rows.each do |transformed_row|
+                selected_rows = rows[1..-1].select do |r|
+                  r[previous_column_range.first] == transformed_row.values.first
+                end
+
+                new_row = selected_rows.compact.uniq { |r| r[0] }.map do |row|
+                  transform(row.compact.values_at(*column_range), column_range)
+                end
+
+                if new_row.empty?
+                  columns = self.class.defined_columns[selected_source]
+
+                  columns.each do |column|
+                    new_row << { column[1] => "" }
+                  end
+                end
+
+                transformed_row.merge!(*new_row)
+              end
+            end
+          end
+        else
+          transformed_rows = rows[1..-1].map do |row|
+            transform(row.compact.values_at(*column_range), column_range)
+          end
         end
 
-        transformed_rows.uniq.sort_by { |hash| data_sort(hash) }
+        transformed_rows.uniq.sort_by { |h| data_sort(h) }
       end
 
       protected
 
-      def default_filename
-        File.join DATA_PATH, sources[selected_source]
+      attr_accessor :selected_source
+
+      def default_filename_to(source)
+        sources[source.to_sym]
       end
 
       def spreadsheet
-        @spreadsheet ||= Spreadsheet.open filename
+        filename = File.join(DATA_PATH, default_filename_to(selected_source))
+
+        @spreadsheet[selected_source] ||= Spreadsheet.open filename
       end
 
       def sheet
-        @selected_sheet ||= spreadsheet.worksheets.first
+        @selected_sheet[selected_source] ||= spreadsheet.worksheets.first
       end
 
       def rows
-        @rows ||= sheet.rows
-      end
-
-      def selected_source
-        self.class.selected_source || :dtb
+        @rows[selected_source] ||= sheet.rows
       end
 
       def sources
@@ -77,7 +119,7 @@ module Ibge
       end
 
       def column_keys
-        defined_columns.keys.map(&:to_s).map(&:to_i)
+        defined_columns[selected_source].keys.map(&:to_s).map(&:to_i)
       end
 
       def column_range
@@ -86,9 +128,13 @@ module Ibge
 
       def transform(row, column_range)
         row.each_with_index.inject({}) do |item, (raw_data, index)|
-          column_index = column_range[index]
+          column_index = column_range[index].to_s.to_sym
+
+          hash = {
+            defined_columns[selected_source][column_index] => raw_data
+          }
 
-          item.merge!(self.class.defined_columns[column_index.to_s.to_sym] => raw_data)
+          item.merge!(hash)
         end
       end
 
@@ -102,6 +148,10 @@ module Ibge
 
         order_values
       end
+
+      def multiple_sources?
+        defined_columns.keys.size > 1
+      end
     end
   end
 end

data/lib/ibge/reader/helpers.rb

@@ -0,0 +1,78 @@
+module Ibge
+  module Reader
+    module Helpers
+      SOURCES = {
+        dtb: "DTB_2014_subdistrito.xls",
+        sidra: "sidra.xls"
+      }
+
+      attr_reader :defined_columns, :selected_source
+
+      def columns(columns)
+        if columns.empty?
+          raise ArgumentError,
+                "Columns must be provided as hash which contains " /
+                  "the index and field name"
+        end
+
+        @defined_columns ||= {}
+        @selected_source ||= :dtb
+
+        distinct_sources = []
+
+        columns.each do |_, v|
+          if v.is_a?(Hash)
+            hash_column_definition = v.values.first
+
+            unless hash_column_definition.key?(:from_source)
+              raise ArgumentError, "#{v.key} must have the key from_source"
+            end
+
+            valid_informed_source(hash_column_definition[:from_source])
+
+            distinct_sources << hash_column_definition[:from_source]
+          end
+        end
+
+        distinct_sources = distinct_sources.compact.uniq.sort
+
+        distinct_sources.each do |source|
+          extracted_columns = columns.select do |_, v|
+            v.is_a?(Hash) && v.values.first[:from_source] == source
+          end
+
+          @defined_columns[source] = extracted_columns.inject({}) do |h, (k, v)|
+            h.merge!(k.to_s.to_sym => v.is_a?(Hash) ? v.keys.first : v)
+
+            h
+          end
+        end
+
+        if @selected_source
+          @defined_columns[@selected_source] = columns.select do |_, v|
+            v.is_a?(Symbol)
+          end
+        end
+      end
+
+      def source(selected_source = :dtb)
+        valid_informed_source(selected_source)
+
+        @selected_source = selected_source
+      end
+
+      def sources
+        SOURCES
+      end
+
+      def valid_informed_source(informed_source)
+        source_keys = sources.keys
+
+        unless source_keys.include?(informed_source)
+          raise ArgumentError,
+                "Source must be one of those '#{source_keys.join(', ')}'"
+        end
+      end
+    end
+  end
+end

data/lib/ibge/reader/state.rb

@@ -1,11 +1,9 @@
 module Ibge
   module Reader
     class State < Base
-      source :sidra
-
       columns "0": :code,
               "1": :state_name,
-              "2": :state_federative_unit
+              "2": { state_federative_unit: { from_source: :sidra } }
     end
   end
 end

data/lib/ibge/version.rb

@@ -1,3 +1,3 @@
 module Ibge
-  VERSION = "0.2.0"
+  VERSION = "0.2.1"
 end

data/spec/ibge/reader/base_spec.rb

@@ -1,47 +1,97 @@
-require 'spec_helper'
+require "spec_helper"
 
 describe Ibge::Reader::Base do
-  class Reader < Ibge::Reader::Base
-    source :sidra
+  context "source on top of the reader" do
+    class ReaderWithSource < Ibge::Reader::Base
+      source :sidra
 
-    columns "0": :code
-  end
+      columns "0": :code
+    end
 
-  subject { Reader }
+    subject { ReaderWithSource }
 
-  describe '.columns' do
-    it { should respond_to :columns }
+    describe ".columns" do
+      it { should respond_to :columns }
 
-    it 'maps the column index to a recognized column name' do
-      expect(subject.defined_columns).to eq "0": :code
+      it "maps the column index to a recognized column name" do
+        expect(subject.defined_columns).to eq sidra: { "0": :code }
+      end
     end
-  end
 
-  describe '.read' do
-    context 'without pass a :filename as option' do
-      it 'reads the file inside data folder' do
-        expect(subject.read).to include code: "11"
+    describe ".read" do
+      context "without pass a :filename as option" do
+        it "reads the file inside data folder" do
+          expect(subject.read).to include code: "11"
+        end
+      end
+
+      context "passing a :filename" do
+        it "reads the file provided" do
+          filename = "spec/fixtures/DTB_example_subdistrito.xls"
+
+          content = subject.read(filename: { dtb: filename })
+
+          expect(content).to include code: "11"
+        end
       end
     end
 
-    context 'passing a :filename' do
-      it 'reads the file provided' do
-        filename = "spec/fixtures/DTB_example_subdistrito.xls"
+    describe ".selected_source" do
+      it "show the default file source to read" do
+        expect(subject.selected_source).to eq :sidra
+      end
+    end
 
-        expect(subject.read(filename: filename)).to include code: "11"
+    describe "#filename" do
+      it "sets the default to file inside data folder" do
+        expect(subject.new.filename).to include "data/sidra.xls"
       end
     end
   end
 
-  describe '.selected_source' do
-    it 'show the default file source to read' do
-      expect(subject.selected_source).to eq :sidra
+  context "without source defined" do
+    class ReaderWithoutSource < Ibge::Reader::Base
+      columns "0": :code
     end
-  end
 
-  describe '#filename' do
-    it 'sets the default to file inside data folder' do
-      expect(subject.new.filename).to include 'data/sidra.xls'
+    subject { ReaderWithoutSource }
+
+    describe ".columns" do
+      it { should respond_to :columns }
+
+      it "maps the column index to a recognized column name" do
+        expect(subject.defined_columns).to eq dtb: { "0": :code }
+      end
+    end
+
+    describe ".read" do
+      context "without pass a :filename as option" do
+        it "reads the file inside data folder" do
+          expect(subject.read).to include code: "11"
+        end
+      end
+
+      context "passing a :filename" do
+        it "reads the file provided" do
+          filename = "spec/fixtures/DTB_example_subdistrito.xls"
+
+          content = subject.read(filename: { dtb: filename })
+
+          expect(content).to include code: "11"
+        end
+      end
+    end
+
+    describe ".selected_source" do
+      it "show the default file source to read" do
+        expect(subject.selected_source).to eq :dtb
+      end
+    end
+
+    describe "#filename" do
+      it "sets the default to file inside data folder" do
+        expect(subject.new.filename).to include "data/DTB_2014_subdistrito.xls"
+      end
     end
   end
 end

data/spec/ibge/reader/district_spec.rb

@@ -3,13 +3,21 @@ require 'spec_helper'
 describe Ibge::Reader::District do
   subject { described_class }
 
+  let(:columns) do
+    {
+      dtb: {
+        "0": :state_code,
+        "2": :meso_region_code,
+        "4": :micro_region_code,
+        "6": :municipality_code,
+        "9": :code,
+        "10": :full_code,
+        "11": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :state_code,
-                                               "2": :meso_region_code,
-                                               "4": :micro_region_code,
-                                               "6": :municipality_code,
-                                               "9": :code,
-                                               "10": :full_code,
-                                               "11": :name
+    expect(subject.defined_columns).to eq columns
   end
 end

data/spec/ibge/reader/meso_region_spec.rb

@@ -3,9 +3,17 @@ require 'spec_helper'
 describe Ibge::Reader::MesoRegion do
   subject { described_class }
 
+  let(:columns) do
+    {
+      dtb: {
+        "0": :state_code,
+        "2": :code,
+        "3": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :state_code,
-                                               "2": :code,
-                                               "3": :name
+    expect(subject.defined_columns).to eq columns
   end
 end

data/spec/ibge/reader/micro_region_spec.rb

@@ -3,10 +3,18 @@ require 'spec_helper'
 describe Ibge::Reader::MicroRegion do
   subject { described_class }
 
+  let(:columns) do
+    {
+      dtb: {
+        "0": :state_code,
+        "2": :meso_region_code,
+        "4": :code,
+        "5": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :state_code,
-                                               "2": :meso_region_code,
-                                               "4": :code,
-                                               "5": :name
+    expect(subject.defined_columns).to eq columns
   end
 end

data/spec/ibge/reader/municipality_spec.rb

@@ -3,12 +3,20 @@ require 'spec_helper'
 describe Ibge::Reader::Municipality do
   subject { described_class }
 
+  let(:columns) do
+    {
+      dtb: {
+        "0": :state_code,
+        "2": :meso_region_code,
+        "4": :micro_region_code,
+        "6": :code,
+        "7": :full_code,
+        "8": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :state_code,
-                                               "2": :meso_region_code,
-                                               "4": :micro_region_code,
-                                               "6": :code,
-                                               "7": :full_code,
-                                               "8": :name
+    expect(subject.defined_columns).to eq columns
   end
 end

data/spec/ibge/reader/neighborhood_spec.rb

@@ -3,7 +3,17 @@ require 'spec_helper'
 describe Ibge::Reader::Neighborhood do
   subject { described_class }
 
+  let(:columns) do
+    {
+      sidra: {
+        "0": :state_code,
+        "3": :municipality_full_code,
+        "5": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :state_code, "3": :municipality_full_code, "5": :name
+    expect(subject.defined_columns).to eq columns
   end
 end

data/spec/ibge/reader/state_spec.rb

@@ -1,9 +1,28 @@
-require 'spec_helper'
+require "spec_helper"
 
 describe Ibge::Reader::State do
   subject { described_class }
 
-  it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0": :code, "1": :state_name
+  let(:columns) do
+    {
+      dtb: {
+        "0": :code,
+        "1": :state_name
+      },
+
+      sidra: {
+        "2": :state_federative_unit
+      }
+    }
+  end
+
+  it "should have defined columns" do
+    expect(subject.defined_columns).to eq columns
+  end
+
+  it "should read from multiple sources" do
+    expect(subject.read).to include code: "11",
+                                    state_name: "Rondônia",
+                                    state_federative_unit: "RO"
   end
 end

data/spec/ibge/reader/sub_district_spec.rb

@@ -3,14 +3,22 @@ require 'spec_helper'
 describe Ibge::Reader::SubDistrict do
   subject { described_class }
 
+  let(:columns) do
+    {
+      dtb: {
+        "0": :state_code,
+        "2": :meso_region_code,
+        "4": :micro_region_code,
+        "6": :municipality_code,
+        "9": :district_code,
+        "12": :code,
+        "13": :full_code,
+        "14": :name
+      }
+    }
+  end
+
   it 'should have defined columns' do
-    expect(subject.defined_columns).to include "0":  :state_code,
-                                               "2":  :meso_region_code,
-                                               "4":  :micro_region_code,
-                                               "6":  :municipality_code,
-                                               "9":  :district_code,
-                                               "12": :code,
-                                               "13": :full_code,
-                                               "14": :name
+    expect(subject.defined_columns).to include columns
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ibge
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.2.1
 platform: ruby
 authors:
 - Bruno Arueira
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-07-13 00:00:00.000000000 Z
+date: 2015-08-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: spreadsheet
@@ -73,6 +73,7 @@ files:
 - lib/ibge.rb
 - lib/ibge/reader/base.rb
 - lib/ibge/reader/district.rb
+- lib/ibge/reader/helpers.rb
 - lib/ibge/reader/meso_region.rb
 - lib/ibge/reader/micro_region.rb
 - lib/ibge/reader/municipality.rb