Index all PDF imports into vector store with type metadata

app/jobs/process_pdf_job.rb

@@ -11,6 +11,7 @@ class ProcessPdfJob < ApplicationJob
 
     begin
       pdf_import.process_with_ai
+      upload_to_vector_store(pdf_import)
 
       # For bank statements, extract transactions and generate import rows
       if pdf_import.bank_statement?
@@ -58,4 +59,19 @@ class ProcessPdfJob < ApplicationJob
                error: error.class.name.demodulize)
       end
     end
+
+    def upload_to_vector_store(pdf_import)
+      filename = pdf_import.pdf_file.filename.to_s
+      file_content = pdf_import.pdf_file_content
+
+      family_document = pdf_import.family.upload_document(
+        file_content: file_content,
+        filename: filename,
+        metadata: { "type" => pdf_import.document_type }
+      )
+
+      return if family_document
+
+      Rails.logger.warn("ProcessPdfJob: Vector store upload failed for import #{pdf_import.id}")
+    end
 end

app/models/family/vector_searchable.rb

@@ -37,7 +37,7 @@ module Family::VectorSearchable
     response.success? ? response.data : []
   end
 
-  def upload_document(file_content:, filename:)
+  def upload_document(file_content:, filename:, metadata: {})
     adapter = vector_store_adapter
     return nil unless adapter
 
@@ -57,7 +57,8 @@ module Family::VectorSearchable
       content_type: Marcel::MimeType.for(name: filename),
       file_size: file_content.bytesize,
       provider_file_id: response.data[:file_id],
-      status: "ready"
+      status: "ready",
+      metadata: metadata || {}
     )
   end
 

test/jobs/process_pdf_job_test.rb

@@ -32,4 +32,62 @@ class ProcessPdfJobTest < ActiveJob::TestCase
     # Should not change status since already complete
     assert_equal "complete", processed_import.reload.status
   end
+
+  test "uploads non-bank PDF to vector store with classified type metadata" do
+    pdf_content = attach_pdf!(@import)
+
+    @import.stubs(:process_with_ai) do
+      @import.update!(ai_summary: "A tax return", document_type: "financial_document")
+    end
+    @import.stubs(:send_next_steps_email)
+    @import.expects(:extract_transactions).never
+
+    @family.expects(:upload_document).with do |file_content:, filename:, metadata:|
+      assert_equal pdf_content, file_content
+      assert_equal "sample_bank_statement.pdf", filename
+      assert_equal({ "type" => "financial_document" }, metadata)
+      true
+    end.returns(family_documents(:tax_return))
+
+    ProcessPdfJob.perform_now(@import)
+
+    assert_equal "complete", @import.reload.status
+  end
+
+  test "uploads bank statement PDF to vector store with classified type metadata" do
+    pdf_content = attach_pdf!(@import)
+
+    @import.stubs(:process_with_ai) do
+      @import.update!(ai_summary: "A bank statement", document_type: "bank_statement")
+    end
+    @import.expects(:extract_transactions).once
+    @import.expects(:generate_rows_from_extracted_data).once do
+      @import.update_column(:rows_count, 1)
+    end
+    @import.expects(:sync_mappings).once
+    @import.stubs(:send_next_steps_email)
+
+    @family.expects(:upload_document).with do |file_content:, filename:, metadata:|
+      assert_equal pdf_content, file_content
+      assert_equal "sample_bank_statement.pdf", filename
+      assert_equal({ "type" => "bank_statement" }, metadata)
+      true
+    end.returns(family_documents(:tax_return))
+
+    ProcessPdfJob.perform_now(@import)
+
+    assert_equal "pending", @import.reload.status
+  end
+
+  private
+
+    def attach_pdf!(import)
+      pdf_content = file_fixture("imports/sample_bank_statement.pdf").binread
+      import.pdf_file.attach(
+        io: StringIO.new(pdf_content),
+        filename: "sample_bank_statement.pdf",
+        content_type: "application/pdf"
+      )
+      pdf_content
+    end
 end

test/models/family_test.rb

@@ -43,4 +43,33 @@ class FamilyTest < ActiveSupport::TestCase
 
     assert_includes family.available_merchants, new_merchant
   end
+
+  test "upload_document stores provided metadata on family document" do
+    family = families(:dylan_family)
+    family.update!(vector_store_id: nil)
+
+    adapter = mock("vector_store_adapter")
+    adapter.expects(:create_store).with(name: "Family #{family.id} Documents").returns(
+      VectorStore::Response.new(success?: true, data: { id: "vs_test123" }, error: nil)
+    )
+    adapter.expects(:upload_file).with(
+      store_id: "vs_test123",
+      file_content: "hello",
+      filename: "notes.txt"
+    ).returns(
+      VectorStore::Response.new(success?: true, data: { file_id: "file-xyz" }, error: nil)
+    )
+
+    VectorStore::Registry.stubs(:adapter).returns(adapter)
+
+    document = family.upload_document(
+      file_content: "hello",
+      filename: "notes.txt",
+      metadata: { "type" => "financial_document" }
+    )
+
+    assert_not_nil document
+    assert_equal({ "type" => "financial_document" }, document.metadata)
+    assert_equal "vs_test123", family.reload.vector_store_id
+  end
 end