🐛 fix extras failing at document level if missing from prediction (#138)

Author: sebastianMindee

lib/mindee/parsing/common/document.rb

@@ -61,8 +61,7 @@ def to_s
         private
 
         def inject_full_text_ocr(raw_prediction)
-          return unless raw_prediction.dig('inference', 'pages') &&
-                        raw_prediction['inference']['pages'][0]['extras']['full_text_ocr']
+          return unless raw_prediction.dig('inference', 'pages', 0, 'extras', 'full_text_ocr')
 
           full_text_ocr = String.new
           raw_prediction.dig('inference', 'pages').each do |page|

spec/extras/full_text_ocr_spec.rb

@@ -25,6 +25,16 @@
       prediction_data
     ).document
   end
+  let(:load_invalid_document) do
+    prediction_data = JSON.parse(
+      File.read(File.join(DIR_PRODUCTS, 'bank_statement_fr', 'response_v1', 'complete.json'))
+    )
+    Mindee::Parsing::Common::ApiResponse.new(
+      Mindee::Product::FR::BankStatement::BankStatementV1,
+      prediction_data,
+      prediction_data
+    ).document
+  end
 end
 
 describe 'FullTextOCR' do
@@ -40,4 +50,98 @@
     expect(full_text_ocr.to_s.strip).to eq(expected_text.strip)
     # expect(page0_ocr).to eq(expected_text.split("\n").join("\n"))
   end
+
+  it "doesn't get full text when the payload is empty" do
+    full_text_ocr = load_invalid_document
+    expect(full_text_ocr.extras).to be_nil
+  end
+
+  it "doesn't get full text when loading dummy data" do
+    synthetic_response = {
+      'id' => 'mock_id',
+      'name' => 'mock_name',
+      'inference' => {
+        'extras' => {},
+        'finished_at' => '1900-01-00T00:00:00.000000',
+        'is_rotation_applied' => nil,
+        'product' => {
+          'name' => 'mock_name',
+          'type' => 'mock_type',
+          'version' => 'mock_version',
+        },
+        'pages' => [
+          {},
+          {
+            'extras' => {},
+            'id' => 0,
+            'orientation' => {
+              'value' => nil,
+            },
+            'prediction' => {},
+          },
+        ],
+        'prediction' => {
+          'account_number' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'bank_address' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'bank_name' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'client_address' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'client_name' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'closing_balance' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'opening_balance' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'statement_date' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'statement_end_date' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'statement_start_date' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'total_credits' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'total_debits' => {
+            'value' => nil,
+            'confidence' => 0,
+          },
+          'transactions' => [
+            'value' => nil,
+            'confidence' => 0,
+          ],
+        },
+      },
+      'extras' => {},
+      'n_pages' => 0,
+    }
+    built_doc = Mindee::Parsing::Common::Document.new(
+      Mindee::Product::FR::BankStatement::BankStatementV1,
+      synthetic_response
+    )
+    expect(built_doc.extras).to be_nil
+  end
 end