ocrsdk 0.1.0

data/spec/ocrsdk/image_spec.rb

@@ -0,0 +1,93 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Image do
+  describe ".as_text" do
+    subject { OCRSDK::Image.new 'image.jpg' }
+    before { mock_ocrsdk }
+    
+    it "should call api and return Promise" do
+      subject.as_text([:russian]).should be_kind_of(OCRSDK::Promise)
+    end
+  end
+
+  describe ".as_text_sync" do
+    subject { OCRSDK::Image.new 'image.jpg' }
+    before { mock_ocrsdk }
+
+    it "should wait till Promise is done and return result" do
+      subject.as_text_sync([:russian], 0).should == 'meow'
+    end
+  end
+
+  describe ".as_pdf" do
+    subject { OCRSDK::Image.new 'image.jpg' }
+    before { mock_ocrsdk }
+    
+    it "should call api and return Promise" do
+      subject.as_pdf([:russian]).should be_kind_of(OCRSDK::Promise)
+    end
+  end
+
+  describe ".as_pdf_sync" do
+    subject { OCRSDK::Image.new 'image.jpg' }
+    before { mock_ocrsdk }
+
+    it "should wait till Promise is done and return result if output file isn't specified" do
+      subject.as_pdf_sync([:russian], nil, 0).should == 'meow'
+    end
+
+    it "should wait till Promise is done and write result in file" do
+      outpath = File.join(File.dirname(File.expand_path(__FILE__)), 'output.pdf')
+      subject.as_pdf_sync([:russian], outpath, 0)
+      File.exists?(outpath).should be_true
+      File.delete outpath
+    end
+  end
+
+  describe ".api_process_image" do
+    subject { OCRSDK::Image.new 'image.jpg', 'app_id', 'pass' }
+
+    it "should raise UnsupportedLanguage on unsupported language" do
+      expect {
+        subject.instance_eval { api_process_image 'image.jpg', [:meow] }
+      }.to raise_error(OCRSDK::UnsupportedLanguage)
+    end
+
+    it "should raise UnsupportedInputFormat on unsupported input format" do
+      expect {
+        subject.instance_eval { api_process_image 'image.meow', [:russian] }
+      }.to raise_error(OCRSDK::UnsupportedInputFormat)
+    end
+
+    it "should raise UnsupportedOutputFormat on unsupported output format" do
+      expect {
+        subject.instance_eval { api_process_image 'image.jpg', [:russian], :meow }
+      }.to raise_error(OCRSDK::UnsupportedOutputFormat)
+    end
+
+    it "should raise UnsupportedProfile on unsupported profile" do
+      expect {
+        subject.instance_eval { api_process_image 'image.jpg', [:russian], :txt, :meow }
+      }.to raise_error(OCRSDK::UnsupportedProfile)
+    end
+
+    it "should raise NetworkError on problems with REST request" do
+      RestClient.stub(:post) {|url, params| raise RestClient::ExceptionWithResponse }
+      RestClient.should_receive(:post).once
+
+      expect {
+        subject.instance_eval { api_process_image TestFiles.russian_jpg_path, [:russian] }
+      }.to raise_error(OCRSDK::NetworkError)
+    end
+
+    it "should do a post request with correct url and file attached" do
+      RestClient.stub(:post) do |uri, params|
+        uri.to_s.should == "http://app_id:pass@#{OCRSDK::SERVICE_URL}/processImage?language=Russian%2CEnglish&exportFormat=txt&profile=documentConversion"
+        params[:upload][:file].should be_kind_of(File)
+      end
+      RestClient.should_receive(:post).once
+      subject.instance_eval { api_process_image TestFiles.russian_jpg_path, [:russian, :english], :txt, :document_conversion }
+    end
+  end
+end

data/spec/ocrsdk/pdf_spec.rb

@@ -0,0 +1,26 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::PDF do
+  describe ".recognizeable?" do
+    it "should return false for a document with text only" do
+      OCRSDK::PDF.new(TestFiles.lorem_pdf).recognizeable?.should be_false
+    end
+
+    it "should return false for document with a lot of text and some images" do
+      OCRSDK::PDF.new(TestFiles.lorem_complex_pdf).recognizeable?.should be_false
+    end
+
+    it "should return true for document with images only" do
+      OCRSDK::PDF.new(TestFiles.recognizeable_pdf).recognizeable?.should be_true
+    end
+
+    it "should return false for malformed document" do
+      OCRSDK::PDF.new(TestFiles.malformed_pdf).recognizeable?.should be_false
+    end
+
+    it "should return true for 'searchable' document with malformed text underneath the pic" do
+      OCRSDK::PDF.new(TestFiles.searchable_malformed_pdf).recognizeable?.should be_true
+    end
+  end
+end

data/spec/ocrsdk/promise_spec.rb

@@ -0,0 +1,165 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Promise do
+
+  describe ".parse_response" do
+    context "correct response" do
+      subject { OCRSDK::Promise.new(nil).parse_response process_image_response }
+
+      its(:task_id)    { should == '22345200-abe8-4f60-90c8-0d43c5f6c0f6' }
+      its(:status)     { should == :submitted }
+      its(:result_url) { should == 'http://domain/blob ID' }
+      its(:estimate_processing_time) { should == 3600 }
+      its(:estimate_completion) { should == DateTime.parse("2001-01-01T13:18:22Z") + 3600.seconds }      
+    end
+
+    context "incorrect response" do
+      subject { OCRSDK::Promise.new nil }
+
+      it "should raise OCRSDKError" do
+        expect {
+          subject.parse_response ''
+        }.to raise_error(OCRSDK::OCRSDKError)
+      end
+    end
+
+    context "insufficient credits" do
+      subject { OCRSDK::Promise.new nil }
+
+      it "should raise an OCRSDK::NotEnoughCredits error" do
+        expect {
+          subject.parse_response process_image_response_credits
+        }.to raise_error(OCRSDK::NotEnoughCredits)
+      end
+    end
+  end 
+
+  describe "self.from_response" do
+    subject { OCRSDK::Promise.from_response process_image_response }
+
+    its(:task_id)    { should == '22345200-abe8-4f60-90c8-0d43c5f6c0f6' }
+    its(:status)     { should == :submitted }
+    its(:result_url) { should == 'http://domain/blob ID' }
+    its(:estimate_processing_time) { should == 3600 }
+    its(:estimate_completion) { should == DateTime.parse("2001-01-01T13:18:22Z") + 3600.seconds }
+  end
+
+  describe ".update" do
+    subject { OCRSDK::Promise.new 'update-task-id' }
+    before do
+      subject.stub(:api_update_status) { process_image_updated_response }
+      subject.update
+    end
+
+    its(:task_id) { should == 'update-task-id' }
+    its(:status)  { should == :in_progress }
+  end
+
+  describe ".api_update_status" do
+    subject { OCRSDK::Promise.new 'test', 'app_id', 'pass' }
+
+    it "should make an api call with correct url" do
+      RestClient.stub(:get) do |url| 
+        url.to_s.should == "http://app_id:pass@#{OCRSDK::SERVICE_URL}/getTaskStatus?taskId=test"
+      end
+      RestClient.should_receive(:get).once
+      subject.instance_eval { api_update_status }
+    end
+
+    it "should raise a NetworkError in case REST request fails" do
+      RestClient.stub(:get) {|url| raise RestClient::ExceptionWithResponse }
+      RestClient.should_receive(:get).once
+
+      expect {
+        subject.instance_eval { api_update_status }
+      }.to raise_error(OCRSDK::NetworkError)
+    end
+  end
+
+  describe ".result" do
+    context "processing completed without errors" do
+      subject { OCRSDK::Promise.from_response process_image_completed_response }
+
+      it "should call api method" do
+        subject.should_receive(:api_get_result).once
+        subject.result
+      end
+
+      it "should get file with coorect url and return its contents" do
+        RestClient.stub(:get) do |url| 
+          url.to_s.should == "http://domain/blob ID"
+          "meow"
+        end
+        RestClient.should_receive(:get).once
+        subject.result.should == 'meow'
+      end
+
+      it "should raise NetworkError in case getting file fails" do
+        RestClient.stub(:get) {|url| raise RestClient::ExceptionWithResponse }
+        RestClient.should_receive(:get).once
+
+        expect {
+          subject.result
+        }.to raise_error(OCRSDK::NetworkError)        
+      end
+    end
+
+    context "processing failed" do
+      subject { OCRSDK::Promise.from_response process_image_failed_response }
+
+      it "should raise an ProcessingFailed" do
+        expect {
+          subject.result
+        }.to raise_error(OCRSDK::ProcessingFailed)
+      end
+    end
+  end
+
+  describe ".completed? and .failed?" do
+    context "processed job" do
+      subject { OCRSDK::Promise.from_response process_image_updated_response }
+
+      its(:processing?) { should be_true }
+      its(:completed?)  { should be_false }
+      its(:failed?)     { should be_false }
+    end
+
+    context "completed job" do
+      subject { OCRSDK::Promise.from_response process_image_completed_response }
+
+      its(:processing?) { should be_false }
+      its(:completed?)  { should be_true }
+      its(:failed?)     { should be_false }
+    end
+
+    context "failed job" do
+      subject { OCRSDK::Promise.from_response process_image_failed_response }
+
+      its(:processing?) { should be_false }
+      its(:completed?)  { should be_false }
+      its(:failed?)     { should be_true }
+    end
+  end
+
+  describe ".wait" do
+    subject { OCRSDK::Promise.from_response process_image_response }
+
+    it "should check the status as many times as needed waiting while ocr is completed" do
+      called_once = false
+      subject.stub(:update) do
+        if called_once
+          subject.parse_response process_image_completed_response
+        else
+          called_once = true
+        end
+      end
+      subject.should_receive(:update).twice
+
+      start = Time.now
+      subject.wait 0.1
+      (Time.now - start).should >= 0.2
+    end
+  end
+
+end

data/spec/ocrsdk/verifiers/format_spec.rb

@@ -0,0 +1,51 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Verifiers::Format do
+  let (:class_with_module) {
+    Class.new do
+      include OCRSDK::Verifiers::Format
+    end
+  }
+  subject { class_with_module.new }
+
+  it "should have list of possible input formats" do
+    OCRSDK::Verifiers::Format::INPUT_FORMATS.length.should > 0
+  end
+
+  it "should have list of possible output formats" do
+    OCRSDK::Verifiers::Format::OUTPUT_FORMATS.length.should > 0
+  end
+
+  it "should convert format to string" do
+    subject.format_to_s(:meow_meow).should == 'meowMeow'
+  end
+
+  describe ".supported_input_format?" do
+    it "should return false for incorrect input format" do 
+      subject.supported_input_format?(:meow_meow).should be_false
+    end
+
+    it "should return true for correct input format as symbol" do
+      subject.supported_input_format?(:pdf).should be_true
+    end
+
+    it "should return true for correct input format as string" do
+      subject.supported_input_format?("pdf").should be_true
+    end
+  end
+
+  describe ".supported_output_format?" do
+    it "should return false for incorrect output format" do 
+      subject.supported_output_format?(:meow_meow).should be_false
+    end
+
+    it "should return true for correct output format as symbol" do
+      subject.supported_output_format?(:txt).should be_true
+    end
+
+    it "should return true for correct output format as string" do
+      subject.supported_output_format?("pdfSearchable").should be_true
+    end
+  end  
+end

data/spec/ocrsdk/verifiers/language_spec.rb

@@ -0,0 +1,55 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Verifiers::Language do
+  let (:class_with_module) {
+    Class.new do
+      include OCRSDK::Verifiers::Language
+    end
+  }
+  subject { class_with_module.new }
+
+  it "should have list of languages" do
+    OCRSDK::Verifiers::Language::LANGUAGES.length.should > 0
+  end
+
+  it "should convert language to string" do
+    subject.language_to_s(:meow_meow).should == 'MeowMeow'
+  end
+
+  describe ".language_to_sym" do
+    it "should convert language to symbol" do
+      subject.language_to_sym("MeowMeow").should == :meow_meow
+    end
+
+    it "should produce reversible results" do
+      subject.language_to_sym(subject.language_to_s(:meow_meow)).should == :meow_meow
+    end
+  end
+
+  describe ".languages_to_s" do
+    it "should convert list of languages to strings" do
+      subject.languages_to_s([:hebrew, :serbian_cyrillic]).should == ['Hebrew', 'SerbianCyrillic']
+    end
+
+    it "should raise an exception if language doesn't exist" do
+      expect {
+          subject.languages_to_s([:meow_meow, :pew_pew])
+        }.to raise_error(OCRSDK::UnsupportedLanguage)
+    end
+  end
+
+  describe ".supported_language?" do
+    it "should return false for incorrect language" do 
+      subject.supported_language?(:meow_meow).should be_false
+    end
+
+    it "should return true for correct language as symbol" do
+      subject.supported_language?(:russian).should be_true
+    end
+
+    it "should return true for correct language as string" do
+      subject.supported_language?("Russian").should be_true
+    end
+  end
+end

data/spec/ocrsdk/verifiers/profile_spec.rb

@@ -0,0 +1,33 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Verifiers::Profile do
+  let (:class_with_module) {
+    Class.new do
+      include OCRSDK::Verifiers::Profile
+    end
+  }
+  subject { class_with_module.new }
+
+  it "should have list of possible profiles" do
+    OCRSDK::Verifiers::Profile::PROFILES.length.should > 0
+  end
+
+  it "should convert profile to string" do
+    subject.profile_to_s(:meow_meow).should == 'meowMeow'
+  end
+
+  describe ".supported_profile?" do
+    it "should return false for incorrect profile" do 
+      subject.supported_profile?(:meow_meow).should be_false
+    end
+
+    it "should return true for correct profile as symbol" do
+      subject.supported_profile?(:document_conversion).should be_true
+    end
+
+    it "should return true for correct profile as string" do
+      subject.supported_profile?("documentConversion").should be_true
+    end
+  end
+end

data/spec/ocrsdk/verifiers/status_spec.rb

@@ -0,0 +1,43 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK::Verifiers::Status do
+  let (:class_with_module) {
+    Class.new do
+      include OCRSDK::Verifiers::Status
+    end
+  }
+  subject { class_with_module.new }
+
+  it "should have list of statuses" do
+    OCRSDK::Verifiers::Status::STATUSES.length.should > 0
+  end
+
+  it "should convert status to string" do
+    subject.status_to_s(:meow_meow).should == 'MeowMeow'
+  end
+
+  describe ".status_to_sym" do
+    it "should convert status to symbol" do
+      subject.status_to_sym("MeowMeow").should == :meow_meow
+    end
+
+    it "should produce reversible results" do
+      subject.status_to_sym(subject.status_to_s(:meow_meow)).should == :meow_meow
+    end
+  end
+
+  describe ".supported_status?" do
+    it "should return false for incorrect language" do 
+      subject.supported_status?(:meow_meow).should be_false
+    end
+
+    it "should return true for correct status as symbol" do
+      subject.supported_status?(:submitted).should be_true
+    end
+
+    it "should return true for correct status as string" do
+      subject.supported_status?("Submitted").should be_true
+    end
+  end
+end

data/spec/ocrsdk_spec.rb

@@ -0,0 +1,8 @@
+# -*- encoding : utf-8 -*-
+require 'spec_helper'
+
+describe OCRSDK do
+  it "should have service url" do 
+    OCRSDK::SERVICE_URL.length.should > 0
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,8 @@
+if RUBY_VERSION >= '1.9'
+  require 'cover_me'
+end
+
+require 'rspec'
+require 'ocrsdk'
+require 'helpers/ocrsdk_helpers'
+require 'support/test_files'

data/spec/support/test_files.rb

@@ -0,0 +1,36 @@
+# As a separate module, because we might want to
+# mock some files in the future or use generator
+# for the same reason it returns opened instance of file
+#
+# Each method check if corresponding file exists in
+# spec/support/files/file.name.kitten.ext
+# if method has `_path` in the end Pathname instance would be returned
+# otherwise it will be File instance
+module TestFiles
+  def self.respond_to?(method)
+    File.exists? self.filename(method)
+  end
+
+  def self.method_missing(method, *args, &block)
+    fname = self.filename(method)
+    if File.exists? fname
+      if only_path? method
+        fname
+      else
+        File.open(fname)
+      end
+    else
+      super
+    end
+  end
+
+  def self.filename(method)
+    method = method[0..-6]  if only_path? method
+    File.join(File.dirname(File.expand_path(__FILE__)), '..', 'fixtures', 'files', method.to_s.gsub('_', '.'))
+  end
+
+  def self.only_path?(method)
+    method[-5..-1] == '_path'
+  end
+
+end

data/travis.yml

@@ -0,0 +1,4 @@
+rvm:
+  - 1.8.7
+  - 1.9.2
+  - 1.9.3