Merge 5d69475 into 5642703

.github/workflows/ci.yml

@@ -0,0 +1,32 @@
+name: CI
+on: [push]
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        ruby:
+          - '2.5.x'
+          - '2.6.x'
+          - '2.7.x'
+    steps:
+      steps:
+        - uses: actions/checkout@v2
+      - name: Install tesseract-ocr
+        run: |
+          sudo add-apt-repository ppa:alex-p/tesseract-ocr -y
+          sudo apt-get update -q
+          sudo apt-get install tesseract-ocr tesseract-ocr-eng ghostscript -y
+      - name: Setup Ruby
+        uses: actions/setup-ruby@v1
+        with:
+          ruby-version: ${{ matrix.ruby }}
+      - name: Bundle
+        env:
+          MTSR_RAILS_VERSION: ${{ matrix.rails }}
+        run: |
+          gem uninstall -aIx bundler
+          gem install bundler
+          bundle install --jobs 4 --retry 3
+      - name: Test
+        run: bundle exec rake

spec/rtesseract/box_spec.rb

@@ -3,12 +3,14 @@
 RSpec.describe RTesseract::Box do
   let(:path) { Pathname.new(File.dirname(__FILE__)).join('..') }
   let(:words_image) { path.join('resources', 'test_words.png').to_s }
+  let(:words) { ['If', 'you', 'are', 'a', 'friend,', 'you', 'speak', 'the', 'password,', 'and', 'the', 'doors', 'will', 'open.'] }
+  let(:instance) { RTesseract.new(words_image) }
 
-  it 'bounding box' do
-    expect(RTesseract.new(words_image).to_s).to eql("If you are a friend,\nyou speak the password,\nand the doors will open.\n\f")
-    expect(RTesseract.new(words_image).to_box).to include(word: 'you', x_start: 69, y_start: 17, x_end: 100, y_end: 31)
+  it 'returns the list of words' do
+    expect(instance.words).to eql(words)
+  end
 
-    words = ['If', 'you', 'are', 'a', 'friend,', 'you', 'speak', 'the', 'password,', 'and', 'the', 'doors', 'will', 'open.']
-    expect(RTesseract.new(words_image).words).to eql(words)
+  it 'bounding box' do
+    expect(instance.to_box).to include(word: 'you', x_start: 69, y_start: 17, x_end: 100, y_end: 31)
   end
 end

spec/rtesseract/configuration_spec.rb

@@ -3,18 +3,39 @@
 RSpec.describe RTesseract do
   let(:path) { Pathname.new(File.dirname(__FILE__)).join('..') }
 
-  it ' support default config' do
-    RTesseract.configure { |config| config.psm = 7 }
+  context 'with global spm' do
+    before { described_class.configure { |config| config.psm = 7 } }
 
-    expect(RTesseract.config.psm).to eql(7)
-    expect(RTesseract.new(path, psm: 2).config.psm).to eql(2)
+    it 'gets the global psm value' do
+      expect(described_class.config.psm).to be(7)
+    end
 
-    expect(RTesseract.config.command).to eql('tesseract')
-    expect(RTesseract.new(path, command: '/usr/bin/tesseract4').config.command).to eql('/usr/bin/tesseract4')
+    it 'gets instance psm value' do
+      expect(described_class.new(path, psm: 2).config.psm).to be(2)
+    end
+  end
+
+  context 'with default command' do
+    it 'gets the global psm value' do
+      expect(described_class.config.command).to eql('tesseract')
+    end
+
+    it 'gets instance command value' do
+      expect(described_class.new(path, command: '/usr/bin/tesseract4').config.command).to eql('/usr/bin/tesseract4')
+    end
+  end
+
+  context 'with other options' do
+    it 'allows to setup oem' do
+      expect(described_class.new(path, oem: 1).config.oem).to be(1)
+    end
+
+    it 'allows to setup lang' do
+      expect(described_class.new(path, lang: 'eng').config.lang).to eql('eng')
+    end
 
-    expect(RTesseract.new(path, psm: 2).config.psm).to eql(2)
-    expect(RTesseract.new(path, oem: 1).config.oem).to eql(1)
-    expect(RTesseract.new(path, lang: 'eng').config.lang).to eql('eng')
-    expect(RTesseract.new(path, lang: 'eng+por').config.lang).to eql('eng+por')
+    it 'allows to setup multiple langs' do
+      expect(described_class.new(path, lang: 'eng+por').config.lang).to eql('eng+por')
+    end
   end
 end

spec/rtesseract/pdf_spec.rb

@@ -2,17 +2,19 @@
 
 RSpec.describe RTesseract::Pdf do
   let(:path) { Pathname.new(File.dirname(__FILE__)).join('..') }
+  let(:words_image) { path.join('resources', 'test-pdf.png').to_s }
+  let(:file) { RTesseract.new(words_image).to_pdf }
 
-  let(:image_pdf_path) { path.join('resources', 'test-pdf.png').to_s }
-
-  it ' support pdf output mode' do
-    pdf_ocr = RTesseract.new(image_pdf_path).to_pdf
-
-    expect(File.extname(pdf_ocr.path)).to eql('.pdf')
-    expect(File.exist?(pdf_ocr.path)).to be_truthy
+  after do
+    file.close
+    File.unlink(file)
+  end
 
-    pdf_ocr.close
+  it 'returns a file with extension .pdf' do
+    expect(File.extname(file.path)).to eql('.pdf')
+  end
 
-    File.unlink(pdf_ocr)
+  it 'checks if file pdf exisits' do
+    expect(File).to exist(file.path)
   end
 end

spec/rtesseract/text_spec.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+RSpec.describe RTesseract::Text do
+  let(:path) { Pathname.new(File.dirname(__FILE__)).join('..') }
+  let(:image_path) { path.join('resources', 'test.tif').to_s }
+  let(:pdf_path) { path.join('resources', 'test.tif').to_s }
+
+  let(:words_image) { path.join('resources', 'test_words.png').to_s }
+
+  it 'translate image to text' do
+    expect(RTesseract.new(image_path).to_s_without_spaces).to eql('43XF')
+  end
+
+  it 'translate tif image to text' do
+    expect(RTesseract.new(path.join('resources', 'test1.tif').to_s).to_s_without_spaces).to eql('V2V4')
+  end
+
+  it 'translate tif image with spaces to text' do
+    expect(RTesseract.new(path.join('resources', 'test with spaces.tif').to_s).to_s_without_spaces).to eql('V2V4')
+  end
+
+  it 'translate png image with spaces to text' do
+    expect(RTesseract.new(path.join('resources', 'test.png').to_s).to_s_without_spaces).to eql('HW9W')
+  end
+
+  it 'translate jpg image with spaces to text' do
+    expect(RTesseract.new(path.join('resources', 'test.jpg').to_s).to_s_without_spaces).to eql('3R8F')
+  end
+
+  it 'translate image to text with options' do
+    expect(RTesseract.new(image_path, psm: 7, oem: 1).to_s_without_spaces).to eql('43XF')
+  end
+
+  it 'tests output text' do
+    expect(RTesseract.new(words_image).to_s).to eql("If you are a friend,\nyou speak the password,\nand the doors will open.\n\f")
+  end
+end

spec/rtesseract/tsv_spec.rb

@@ -1,18 +1,20 @@
 # frozen_string_literal: true
 
-require 'csv'
-
 RSpec.describe RTesseract::Tsv do
   let(:path) { Pathname.new(File.dirname(__FILE__)).join('..') }
   let(:words_image) { path.join('resources', 'test_words.png').to_s }
+  let(:file) { RTesseract.new(words_image).to_tsv }
 
-  it ' support tsv output mode' do
-    tsv_ocr = RTesseract.new(words_image).to_tsv
+  after do
+    file.close
+    File.unlink(file)
+  end
 
-    expect(File.extname(tsv_ocr.path)).to eql('.tsv')
-    expect(tsv_ocr.read).to include('level	page_num	block_num	par_num	line_num	word_num	left	top	width	height	conf	text')
+  it 'returns a file with extension .tsv' do
+    expect(File.extname(file.path)).to eql('.tsv')
+  end
 
-    tsv_ocr.close
-    File.unlink(tsv_ocr)
+  it ' support tsv output mode' do
+    expect(file.read).to include('level	page_num	block_num	par_num	line_num	word_num	left	top	width	height	conf	text')
   end
 end

spec/rtesseract_spec.rb

@@ -5,60 +5,57 @@
   let(:image_path) { path.join('resources', 'test.tif').to_s }
   let(:pdf_path) { path.join('resources', 'test.tif').to_s }
 
-  it 'has a version number' do
-    expect(RTesseract::VERSION).not_to be nil
-
-    expect(RTesseract.tesseract_version).to be > 3.05
+  it 'returns the tesseract version' do
+    expect(described_class.tesseract_version).to be > 3.05
   end
 
-  it 'be instantiable' do
-    expect(RTesseract.new.class).to eql(RTesseract)
-    expect(RTesseract.new('').class).to eql(RTesseract)
-    expect(RTesseract.new(image_path).class).to eql(RTesseract)
+  it 'be instantiable without path' do
+    expect(described_class.new.class).to eql(described_class)
   end
 
-  it 'translate image to text' do
-    expect(RTesseract.new(image_path).to_s_without_spaces).to eql('43XF')
-    {
-      'test1.tif' => 'V2V4',
-      'test with spaces.tif' => 'V2V4',
-      'test.png' => 'HW9W',
-      'test.jpg' => '3R8F'
-    }.each do |file, value|
-      expect(RTesseract.new(path.join('resources', file).to_s).to_s_without_spaces).to eql(value)
-    end
+  it 'be instantiable with blank string' do
+    expect(described_class.new('').class).to eql(described_class)
   end
 
-  it 'translate image to text with options' do
-    expect(RTesseract.new(image_path, psm: 7, oem: 1).to_s_without_spaces).to eql('43XF')
+  it 'be instantiable with a path' do
+    expect(described_class.new(image_path).class).to eql(described_class)
   end
 
-  it 'get tesseract version' do
-    expect(RTesseract.tesseract_version).to be > 0
+  context 'when tesseract not installed' do
+    before do
+      described_class.configure { |config| config.command = 'tesseract_not_installed' }
+    end
 
-    RTesseract.configure { |config| config.command = 'tesseract_not_installed' }
+    it 'returns zero on #tesseract_version' do
+      expect(described_class.tesseract_version).to be(0)
+    end
 
-    expect(RTesseract.tesseract_version).to eql(0)
+    it 'raise a error if tesseract version < 3.05' do
+      expect { described_class.check_version! }.to raise_error(RTesseract::Error)
+    end
   end
 
-  it 'raise a error if tesseract version < 3.05' do
-    RTesseract.configure { |config| config.command = 'tesseract_not_installed' }
+  context 'without source' do
+    let(:instance) { described_class.new }
 
-    expect { RTesseract.check_version! }.to raise_error(RTesseract::Error)
-  end
-
-  it 'raise a error when tesseract raise a error' do
-    expect { RTesseract.new.to_s }.to raise_error(RTesseract::Error)
+    it 'raise an exception' do
+      aggregate_failures 'raise an exception' do
+        expect { instance.to_s }.to raise_error(RTesseract::Error)
+        expect(instance.errors.first).to include('Error during processing')
+      end
+    end
   end
 
-  it 'store the error on a variable to debug' do
-    instance = RTesseract.new
-    expect { instance.to_s }.to raise_error(RTesseract::Error)
-    expect(instance.errors.first).to include('Error during processing')
-
-    error_intance = RTesseract.new(path.join('resources', 'image_with_error.png').to_s)
+  context 'with errors on image' do
+    let(:error_intance) do
+      described_class.new(path.join('resources', 'image_with_error.png').to_s)
+    end
 
-    expect(error_intance.to_s_without_spaces).to eql('RTX-0003-03-02-01PRE')
-    expect(error_intance.errors).to eql(["Error in boxClipToRectangle: box outside rectangle\nError in pixScanForForeground: invalid box\n"])
+    it 'stores the error on a variable to debug' do
+      aggregate_failures 'stores the error on a variable to debug' do
+        expect(error_intance.to_s_without_spaces).to eql('RTX-0003-03-02-01PRE')
+        expect(error_intance.errors).to eql(["Error in boxClipToRectangle: box outside rectangle\nError in pixScanForForeground: invalid box\n"])
+      end
+    end
   end
 end

spec/spec_helper.rb

@@ -20,7 +20,7 @@
     c.syntax = :expect
   end
 
-  config.before(:each) do
+  config.before do
     RTesseract.reset_config!
   end
 end