✨ Add support for extraction in crop, split, and classify products

lib/mindee/v2/product/classification/classification_classifier.rb

@@ -8,10 +8,16 @@ module Classification
         class ClassificationClassifier
           # @return [String] The document type, as identified on given classification values.
           attr_reader :document_type
+          attr_reader :extraction_response
 
           # @param server_response [Hash] Hash representation of the JSON returned by the service.
           def initialize(server_response)
             @document_type = server_response['document_type']
+            # rubocop:disable Style/GuardClause
+            unless server_response['extraction_response'].nil?
+              @extraction_response = V2::Product::Extraction::ExtractionResponse.new(server_response['extraction_response'])
+            end
+            # rubocop:enable Style/GuardClause
           end
 
           # @return [String] String representation.

lib/mindee/v2/product/crop/crop_item.rb

@@ -10,11 +10,17 @@ class CropItem
           attr_reader :object_type
           # @return [V2::Parsing::Field::FieldLocation] Coordinates of the detected object on the document.
           attr_reader :location
+          attr_reader :extraction_response
 
           # @param server_response [Hash] Hash representation of the JSON returned by the service.
           def initialize(server_response)
             @object_type = server_response['object_type']
             @location = Mindee::V2::Parsing::Field::FieldLocation.new(server_response['location'])
+            # rubocop:disable Style/GuardClause
+            unless server_response['extraction_response'].nil?
+              @extraction_response = V2::Product::Extraction::ExtractionResponse.new(server_response['extraction_response'])
+            end
+            # rubocop:enable Style/GuardClause
           end
 
           # String representation.

lib/mindee/v2/product/split/split_range.rb

@@ -11,11 +11,17 @@ class SplitRange
           attr_reader :page_range
           # @return [String] The document type, as identified on given classification values.
           attr_reader :document_type
+          attr_reader :extraction_response
 
           # @param server_response [Hash] Hash representation of the JSON returned by the service.
           def initialize(server_response)
             @page_range = server_response['page_range']
             @document_type = server_response['document_type']
+            # rubocop:disable Style/GuardClause
+            unless server_response['extraction_response'].nil?
+              @extraction_response = V2::Product::Extraction::ExtractionResponse.new(server_response['extraction_response'])
+            end
+            # rubocop:enable Style/GuardClause
           end
 
           # String representation.

sig/mindee/v2/product/classification/classification_classifier.rbs

@@ -6,6 +6,7 @@ module Mindee
       module Classification
         class ClassificationClassifier
           attr_reader document_type: String
+          attr_reader extraction_response: Extraction::ExtractionResponse
 
           def initialize: (Hash[String | Symbol, untyped]) -> void
         end

sig/mindee/v2/product/crop/crop_item.rbs

@@ -3,8 +3,9 @@ module Mindee
     module Product
       module Crop
         class CropItem
+          attr_reader extraction_response: Extraction::ExtractionResponse
           attr_reader object_type: String
-          attr_reader location: Mindee::V2::Parsing::Field::FieldLocation
+          attr_reader location: Parsing::Field::FieldLocation
 
           def initialize: (Hash[String | Symbol, untyped]) -> void
 

sig/mindee/v2/product/split/split_range.rbs

@@ -3,6 +3,7 @@ module Mindee
     module Product
       module Split
         class SplitRange
+          attr_reader extraction_response: Extraction::ExtractionResponse
           attr_reader page_range: Array[Integer]
           attr_reader document_type: String
 

spec/v2/file_operations/crop_operation_integration.rb

@@ -65,7 +65,7 @@ def check_findoc_return(findoc_response)
 
     extracted_images.save_all_to_disk(OUTPUT_DIR)
 
-    expect(File.size(File.join(OUTPUT_DIR, 'crop_001.jpg'))).to be_between(600_000, 672_913)
-    expect(File.size(File.join(OUTPUT_DIR, 'crop_002.jpg'))).to be_between(600_000, 675_728)
+    expect(File.size(File.join(OUTPUT_DIR, 'crop_001.jpg'))).to be_between(560_000, 675_000)
+    expect(File.size(File.join(OUTPUT_DIR, 'crop_002.jpg'))).to be_between(580_000, 680_000)
   end
 end

spec/v2/product/classification/classification_integration.rb

@@ -14,7 +14,7 @@
 
   it 'processes classification default sample correctly' do
     input_source = Mindee::Input::Source::PathInputSource.new(
-      File.join(V2_PRODUCT_DATA_DIR, 'classification', 'default_invoice.jpg')
+      File.join(V2_PRODUCT_DATA_DIR, 'classification', 'default_sample.jpg')
     )
 
     params = { model_id: classification_model_id }
@@ -26,7 +26,7 @@
     )
 
     expect(response.inference).not_to be_nil
-    expect(response.inference.file.name).to eq('default_invoice.jpg')
+    expect(response.inference.file.name).to eq('default_sample.jpg')
     expect(response.inference.result.classification).not_to be_nil
     expect(response.inference.result.classification.document_type).to eq('invoice')
   end

spec/v2/product/classification/classification_spec.rb

@@ -6,10 +6,9 @@
 describe Mindee::V2::Product::Classification::Classification, :v2 do
   let(:classification_data_dir) { File.join(V2_PRODUCT_DATA_DIR, 'classification') }
 
-  it 'parses a single classification properly' do
-    json_path = File.join(classification_data_dir, 'classification_single.json')
+  it 'should load a single result' do
+    json_path = File.join(classification_data_dir, 'default_sample.json')
     json_sample = JSON.parse(File.read(json_path))
-
     response = Mindee::V2::Product::Classification::ClassificationResponse.new(json_sample)
 
     expect(response.inference).to be_a(Mindee::V2::Product::Classification::ClassificationInference)
@@ -20,4 +19,18 @@
 
     expect(response.inference.result.classification.document_type).to eq('invoice')
   end
+
+  it 'should load a single result with extraction' do
+    json_path = File.join(classification_data_dir, 'default_sample_extraction.json')
+    json_sample = JSON.parse(File.read(json_path))
+    response = Mindee::V2::Product::Classification::ClassificationResponse.new(json_sample)
+
+    classification = response.inference.result.classification
+    expect(classification.document_type).to eq('invoice')
+
+    expect(classification.extraction_response).to be_a(Mindee::V2::Product::Extraction::ExtractionResponse)
+    expect(
+      classification.extraction_response.inference.result.fields.get_simple_field('customer_name').value
+    ).to eq('Jiro Doi')
+  end
 end

spec/v2/product/crop/crop_spec.rb

@@ -8,7 +8,7 @@
 describe Mindee::V2::Product::Crop::Crop do
   let(:crop_data_dir) { File.join(V2_PRODUCT_DATA_DIR, 'crop') }
 
-  it 'parses a single crop properly' do
+  it 'should load a single result' do
     json_path = File.join(crop_data_dir, 'crop_single.json')
     rst_path  = File.join(crop_data_dir, 'crop_single.rst')
 
@@ -37,7 +37,7 @@
     expect(response.to_s).to eq(rst_sample)
   end
 
-  it 'parses multiple crops properly' do
+  it 'should load multiple results' do
     json_path = File.join(crop_data_dir, 'crop_multiple.json')
     rst_path  = File.join(crop_data_dir, 'crop_multiple.rst')
 
@@ -52,35 +52,65 @@
     expect(response.inference.result.crops.size).to eq(2)
 
     # First Crop assertions
-    crop_zero = response.inference.result.crops[0]
-    expect(crop_zero.location.polygon.size).to eq(4)
-    expect(crop_zero.location.polygon[0][0]).to eq(0.214)
-    expect(crop_zero.location.polygon[0][1]).to eq(0.079)
-    expect(crop_zero.location.polygon[1][0]).to eq(0.476)
-    expect(crop_zero.location.polygon[1][1]).to eq(0.079)
-    expect(crop_zero.location.polygon[2][0]).to eq(0.476)
-    expect(crop_zero.location.polygon[2][1]).to eq(0.979)
-    expect(crop_zero.location.polygon[3][0]).to eq(0.214)
-    expect(crop_zero.location.polygon[3][1]).to eq(0.979)
-
-    expect(crop_zero.location.page).to eq(0)
-    expect(crop_zero.object_type).to eq('invoice')
+    crop0 = response.inference.result.crops[0]
+    expect(crop0.location.polygon.size).to eq(4)
+    expect(crop0.location.polygon[0][0]).to eq(0.214)
+    expect(crop0.location.polygon[0][1]).to eq(0.079)
+    expect(crop0.location.polygon[1][0]).to eq(0.476)
+    expect(crop0.location.polygon[1][1]).to eq(0.079)
+    expect(crop0.location.polygon[2][0]).to eq(0.476)
+    expect(crop0.location.polygon[2][1]).to eq(0.979)
+    expect(crop0.location.polygon[3][0]).to eq(0.214)
+    expect(crop0.location.polygon[3][1]).to eq(0.979)
+
+    expect(crop0.location.page).to eq(0)
+    expect(crop0.object_type).to eq('invoice')
 
     # Second Crop assertions
-    crop_one = response.inference.result.crops[1]
-    expect(crop_one.location.polygon.size).to eq(4)
-    expect(crop_one.location.polygon[0][0]).to eq(0.547)
-    expect(crop_one.location.polygon[0][1]).to eq(0.15)
-    expect(crop_one.location.polygon[1][0]).to eq(0.862)
-    expect(crop_one.location.polygon[1][1]).to eq(0.15)
-    expect(crop_one.location.polygon[2][0]).to eq(0.862)
-    expect(crop_one.location.polygon[2][1]).to eq(0.97)
-    expect(crop_one.location.polygon[3][0]).to eq(0.547)
-    expect(crop_one.location.polygon[3][1]).to eq(0.97)
-
-    expect(crop_one.location.page).to eq(0)
-    expect(crop_one.object_type).to eq('invoice')
+    crop1 = response.inference.result.crops[1]
+    expect(crop1.location.polygon.size).to eq(4)
+    expect(crop1.location.polygon[0][0]).to eq(0.547)
+    expect(crop1.location.polygon[0][1]).to eq(0.15)
+    expect(crop1.location.polygon[1][0]).to eq(0.862)
+    expect(crop1.location.polygon[1][1]).to eq(0.15)
+    expect(crop1.location.polygon[2][0]).to eq(0.862)
+    expect(crop1.location.polygon[2][1]).to eq(0.97)
+    expect(crop1.location.polygon[3][0]).to eq(0.547)
+    expect(crop1.location.polygon[3][1]).to eq(0.97)
+
+    expect(crop1.location.page).to eq(0)
+    expect(crop1.object_type).to eq('receipt')
 
     expect(response.to_s).to eq(rst_sample)
   end
+
+  it 'should load extraction properties' do
+    json_path = File.join(crop_data_dir, 'default_sample_extraction.json')
+    json_sample = JSON.parse(File.read(json_path))
+    response = Mindee::V2::Product::Crop::CropResponse.new(json_sample)
+
+    expect(response.inference).to be_a(Mindee::V2::Product::Crop::CropInference)
+    crops = response.inference.result.crops
+    expect(response.inference.result.crops.size).to eq(2)
+
+    crop0 = crops[0]
+    expect(crop0.object_type).to eq('receipt')
+    expect(crop0.location.polygon).not_to be_nil
+    expect(crop0.location.page).to eq(0)
+    extraction_response0 = crop0.extraction_response
+    expect(extraction_response0).not_to be_nil
+    expect(
+      extraction_response0.inference.result.fields.get_simple_field('supplier_name').value
+    ).to eq('CHEZ ALAIN MIAM MIAM')
+
+    crop1 = crops[1]
+    expect(crop1.object_type).to eq('receipt')
+    expect(crop1.location.polygon).not_to be_nil
+    expect(crop1.location.page).to eq(0)
+    extraction_response1 = crop1.extraction_response
+    expect(extraction_response1).not_to be_nil
+    expect(
+      extraction_response1.inference.result.fields.get_simple_field('supplier_name').value
+    ).to eq('La cerise sur la pizza')
+  end
 end

spec/v2/product/split/split_spec.rb

@@ -24,30 +24,58 @@
   it 'parses multiple splits properly' do
     json_path = File.join(split_data_dir, 'split_multiple.json')
     json_sample = JSON.parse(File.read(json_path))
-
     response = Mindee::V2::Product::Split::SplitResponse.new(json_sample)
 
     expect(response.inference).to be_a(Mindee::V2::Product::Split::SplitInference)
     expect(response.inference.result).to be_a(Mindee::V2::Product::Split::SplitResult)
     expect(response.inference.result.splits[0]).to be_a(Mindee::V2::Product::Split::SplitRange)
     expect(response.inference.result.splits.size).to eq(3)
 
-    split_zero = response.inference.result.splits[0]
-    expect(split_zero.page_range.size).to eq(2)
-    expect(split_zero.page_range[0]).to eq(0)
-    expect(split_zero.page_range[1]).to eq(0)
-    expect(split_zero.document_type).to eq('invoice')
-
-    split_one = response.inference.result.splits[1]
-    expect(split_one.page_range.size).to eq(2)
-    expect(split_one.page_range[0]).to eq(1)
-    expect(split_one.page_range[1]).to eq(3)
-    expect(split_one.document_type).to eq('invoice')
-
-    split_two = response.inference.result.splits[2]
-    expect(split_two.page_range.size).to eq(2)
-    expect(split_two.page_range[0]).to eq(4)
-    expect(split_two.page_range[1]).to eq(4)
-    expect(split_two.document_type).to eq('invoice')
+    split0 = response.inference.result.splits[0]
+    expect(split0.page_range.size).to eq(2)
+    expect(split0.page_range[0]).to eq(0)
+    expect(split0.page_range[1]).to eq(0)
+    expect(split0.document_type).to eq('passport')
+
+    split1 = response.inference.result.splits[1]
+    expect(split1.page_range.size).to eq(2)
+    expect(split1.page_range[0]).to eq(1)
+    expect(split1.page_range[1]).to eq(3)
+    expect(split1.document_type).to eq('invoice')
+
+    split2 = response.inference.result.splits[2]
+    expect(split2.page_range.size).to eq(2)
+    expect(split2.page_range[0]).to eq(4)
+    expect(split2.page_range[1]).to eq(4)
+    expect(split2.document_type).to eq('receipt')
+  end
+
+  it 'should load extraction properties' do
+    json_path = File.join(split_data_dir, 'default_sample_extraction.json')
+    json_sample = JSON.parse(File.read(json_path))
+    response = Mindee::V2::Product::Split::SplitResponse.new(json_sample)
+
+    expect(response.inference).to be_a(Mindee::V2::Product::Split::SplitInference)
+
+    splits = response.inference.result.splits
+    expect(splits.size).to eq(2)
+
+    split0 = splits[0]
+    expect(split0.document_type).to eq('invoice')
+    expect(split0.page_range[0]).to eq(0)
+    extraction_response0 = split0.extraction_response
+    expect(extraction_response0).not_to be_nil
+    expect(
+      extraction_response0.inference.result.fields.get_simple_field('supplier_phone_number').value
+    ).to eq('05 05 44 44 90')
+
+    split1 = splits[1]
+    expect(split1.document_type).to eq('invoice')
+    expect(split1.page_range[0]).to eq(1)
+    extraction_response1 = split1.extraction_response
+    expect(extraction_response1).not_to be_nil
+    expect(
+      extraction_response1.inference.result.fields.get_simple_field('supplier_phone_number').value
+    ).to eq('416-555-1212')
   end
 end