Fix/lazy load encode improvements

lib/estratto/content.rb

@@ -3,9 +3,9 @@
 module Estratto
   class Content
     def self.for(file_path)
-      content = File.read(file_path)
+      content = File.open(file_path)
       encoded_content = Encoder.new(content).encode
-      encoded_content.split("\n")
+      encoded_content
     end
   end
 end

lib/estratto/encoder.rb

@@ -9,13 +9,16 @@ def initialize(content)
     end
 
     def encode
-      CharlockHolmes::Converter.convert(content, encoding, 'UTF-8')
+      content.lazy.map do |line|
+        charset = detect(line)
+        CharlockHolmes::Converter.convert(line, charset[:encoding], 'UTF-8').chomp
+      end
     end
 
     private
 
-    def encoding
-      CharlockHolmes::EncodingDetector.detect(content)[:encoding]
+    def detect(line)
+      CharlockHolmes::EncodingDetector.detect(line)
     end
   end
 end

lib/estratto/helpers/register_enumerator.rb

@@ -0,0 +1,22 @@
+require_relative '../register'
+
+module Estratto
+  module Helpers
+    class RegisterEnumerator < Enumerator::Lazy
+      def initialize(raw_content, layout)
+        super(raw_content.each_with_index) do |yielder, (line, index)|
+          register_layout = layout.register_fields_for(line[layout.prefix_range])
+          next if register_layout.nil?
+
+          register = Register.new(line, index, register_layout).refine
+
+          if block_given?
+            yield register
+          else
+            yielder << register
+          end
+        end
+      end
+    end
+  end
+end

lib/estratto/parser.rb

@@ -1,5 +1,5 @@
-require_relative 'register'
 require_relative 'content'
+require_relative 'helpers/register_enumerator'
 
 module Estratto
   class Parser
@@ -11,11 +11,7 @@ def initialize(file_path, layout)
     end
 
     def perform
-      @data ||= raw_content.map.with_index do |line, index|
-        register_layout = layout.register_fields_for(line[layout.prefix_range])
-        next if register_layout.nil?
-        Register.new(line, index, register_layout).refine
-      end.compact
+      @data ||= Helpers::RegisterEnumerator.new(raw_content, layout)
     end
 
     def raw_content

spec/lib/estratto/content_spec.rb

@@ -10,8 +10,13 @@
         "025000020181225JUSTDESCRIPTION"
       ]
     end
-    it do
-      expect(described_class.for(file_path)).to eq(expected_content)
+
+    it "returns an Enumerator::Lazy object" do
+      expect(described_class.for(file_path)).to be_a Enumerator::Lazy
+    end
+
+    it "returns the right content" do
+      expect(described_class.for(file_path).to_a).to match_array expected_content
     end
   end
 end

spec/lib/estratto/document_spec.rb

@@ -25,8 +25,13 @@
         }
       ]
     end
-    it do
-      expect(subject.process(file: file_path, layout: layout_path)).to eq(data_refined)
+
+    it "returns an Enumerator::Lazy object" do
+      expect(subject.process(file: file_path, layout: layout_path)).to be_a Enumerator::Lazy
+    end
+
+    it "returns the right content" do
+      expect(subject.process(file: file_path, layout: layout_path).to_a).to match_array data_refined
     end
   end
 end

spec/lib/estratto/encoder_spec.rb

@@ -1,20 +1,25 @@
 RSpec.describe Estratto::Encoder do
-  subject { described_class.new(original_content) }
+  let(:instance) { described_class.new(original_content) }
 
-  context 'ISO-8859-1 original encoding' do
-    let(:original_content) { "026000020181227DESCRI\xC7\xC3O\n" }
-    let(:encoded_content) { "026000020181227DESCRIÇÃO\n" }
+  describe "#encode" do
+    subject { instance.encode }
 
-    it 'converts the original content to UTF-8' do
-      expect(subject.encode).to eq(encoded_content)
+    context 'ISO-8859-1 original encoding' do
+      let(:original_content) { ["026000020181227DESCRI\xC7\xC3O\n"] }
+      let(:encoded_content) { ["026000020181227DESCRIÇÃO"] }
+
+      it 'converts the original content to UTF-8' do
+        expect(subject.to_a).to match_array encoded_content
+      end
     end
-  end
 
-  context 'UTF-8 original encoding' do
-    let(:original_content) { "025000020181225JUSTDESCRIPTION\n" }
+    context 'UTF-8 original encoding' do
+      let(:original_content) { ["025000020181225JUSTDESCRIPTION\n"] }
+      let(:encoded_content) { ["025000020181225JUSTDESCRIPTION"] }
 
-    it 'keeps the original content' do
-      expect(subject.encode).to eq(original_content)
+      it 'keeps the original content' do
+        expect(subject.to_a).to match_array encoded_content
+      end
     end
   end
 end

spec/lib/estratto/helpers/register_enumerator_spec.rb

@@ -0,0 +1,51 @@
+RSpec.describe Estratto::Helpers::RegisterEnumerator do
+  subject { instance }
+
+  let(:instance) { described_class.new(content, layout) }
+  let(:content) { Estratto::Content.for(file_path) }
+  let(:layout) { Estratto::Layout::Factory.fabricate(layout_path) }
+  let(:file_path) { 'spec/fixtures/files/data_to_parse.txt' }
+  let(:layout_path) { 'spec/fixtures/files/complete_layout.yml' }
+
+  let(:john_doe) do
+    {
+      name: 'JOHN DOE',
+      gifted_at: Date.new(2018, 12, 25),
+      value: 375.95,
+      buyed_times: 123
+    }
+  end
+  let(:mary_jones) do
+    {
+      name: 'MARY JONES',
+      gifted_at: Date.new(2018, 12, 25),
+      value: 500.0,
+      buyed_times: 12345
+    }
+  end
+  let(:description) do
+    {
+      value: 500.0,
+      generated_at: Date.new(2018, 12, 25),
+      description: 'JUSTDESCRIPTION'
+    }
+  end
+
+  it { is_expected.to be_a Enumerator::Lazy }
+
+  describe '#to_a' do
+    subject { instance.to_a }
+
+    it 'return only rows which prefix is mapped by the layout' do
+      is_expected.to eq [john_doe, mary_jones, description]
+    end
+  end
+
+  describe '#each' do
+    it "yields only Registers for rows which prefix is mapped by the layout" do
+      expect { |block| instance.each(&block) }.to(
+        yield_successive_args(john_doe, mary_jones, description)
+      )
+    end
+  end
+end

spec/lib/estratto/parser_spec.rb

@@ -28,8 +28,13 @@
           }
         ]
       end
-      it do
-        expect(subject.perform).to eq(data_refined)
+
+      it "returns an Enumerator::Lazy object" do
+        expect(subject.perform).to be_a Estratto::Helpers::RegisterEnumerator
+      end
+
+      it "returns the right content" do
+        expect(subject.perform.to_a).to match_array data_refined
       end
     end
 
@@ -44,8 +49,13 @@
           }
         ]
       end
-      it do
-        expect(subject.perform).to eq(data_refined)
+
+      it "returns an Enumerator::Lazy object" do
+        expect(subject.perform).to be_a Enumerator::Lazy
+      end
+
+      it "returns the right content" do
+        expect(subject.perform.to_a).to match_array data_refined
       end
     end
   end