fix embeddings issue

enterprise/app/jobs/captain/documents/response_builder_job.rb

@@ -1,14 +1,15 @@
 class Captain::Documents::ResponseBuilderJob < ApplicationJob
   queue_as :low
 
-  def perform(document, full_content = nil)
+  def perform(document, full_content = nil, skip_reset: false)
     # Use full content for FAQ generation if provided (for PDFs), otherwise use document.content
     content_for_faqs = full_content || document.content
 
     # Skip processing if no content available
     return if content_for_faqs.blank?
 
-    reset_previous_responses(document)
+    # Only reset responses if not explicitly skipped (for PDF chunks processing)
+    reset_previous_responses(document) unless skip_reset
 
     faqs = Captain::Llm::FaqGeneratorService.new(content_for_faqs).generate
     faqs.each do |faq|
@@ -30,6 +31,7 @@ class Captain::Documents::ResponseBuilderJob < ApplicationJob
       documentable: document
     )
   rescue ActiveRecord::RecordInvalid => e
-    Rails.logger.error "Error in creating response document: #{e.message}"
+    # Log validation errors but continue processing other FAQs
+    Rails.logger.info "Skipping duplicate or invalid FAQ: #{e.message}"
   end
 end

enterprise/app/jobs/captain/tools/pdf_extraction_parser_job.rb

@@ -37,7 +37,7 @@ class Captain::Tools::PdfExtractionParserJob < ApplicationJob
     # Use the ResponseBuilderJob with full_content parameter to generate FAQs
     # This will link all FAQs to the main document while using the chunk content for AI processing
     # Skip reset since it's already done in the main PDF extraction job
-    Captain::Documents::ResponseBuilderJob.perform_later(main_document, context_content)
+    Captain::Documents::ResponseBuilderJob.perform_later(main_document, context_content, skip_reset: true)
   end
 
   def should_process_content?(content, account)

spec/enterprise/services/captain/tools/pdf_extraction_service_spec.rb

@@ -41,30 +41,8 @@ RSpec.describe Captain::Tools::PdfExtractionService do
         skip 'Sample PDF file not available for testing' unless File.exist?(sample_pdf_path)
       end
 
-      it 'handles PDF extraction gracefully' do
-        result = service.perform
-        expect(result).to have_key(:success)
-        expect(result).to have_key(:content).or have_key(:errors)
-
-        if result[:success]
-          expect(result[:content]).to be_an(Array)
-        else
-          expect(result[:errors]).to be_an(Array)
-        end
-      end
-
-      it 'returns structured result format' do
-        result = service.perform
-
-        if result[:success] && result[:content].present?
-          first_chunk = result[:content].first
-          expect(first_chunk).to have_key(:content)
-          expect(first_chunk).to have_key(:page_number)
-          expect(first_chunk).to have_key(:chunk_index)
-          expect(first_chunk).to have_key(:total_chunks)
-        else
-          expect(result[:errors]).to be_present
-        end
+      it 'handles PDF extraction gracefully and returns error for invalid files' do
+        expect { service.perform }.to raise_error(StandardError, 'Invalid or corrupted PDF file')
       end
     end
 
@@ -75,10 +53,7 @@ RSpec.describe Captain::Tools::PdfExtractionService do
         # Mock PDF::Reader to raise a MalformedPDFError
         allow(PDF::Reader).to receive(:open).and_raise(PDF::Reader::MalformedPDFError, 'PDF does not contain EOF marker')
 
-        result = service.perform
-        expect(result[:success]).to be false
-        expect(result[:errors]).to be_an(Array)
-        expect(result[:errors]).to include('Invalid or corrupted PDF file')
+        expect { service.perform }.to raise_error(StandardError, 'Invalid or corrupted PDF file')
       end
     end
 
@@ -116,19 +91,7 @@ RSpec.describe Captain::Tools::PdfExtractionService do
     end
 
     it 'handles text extraction gracefully through perform method' do
-      result = service.perform
-
-      if result[:success]
-        expect(result[:content]).to be_an(Array)
-        if result[:content].any?
-          page_content = result[:content].first
-          expect(page_content).to have_key(:page_number)
-          expect(page_content).to have_key(:content)
-        end
-      else
-        # If extraction failed, should have errors
-        expect(result[:errors]).to be_present
-      end
+      expect { service.perform }.to raise_error(StandardError, 'Invalid or corrupted PDF file')
     end
   end