Fix JSON::GeneratorError masking errors for malformed UTF-8 request bodies

lib/rollbar/request_data_extractor.rb

@@ -4,6 +4,7 @@
 require 'rollbar/scrubbers'
 require 'rollbar/scrubbers/url'
 require 'rollbar/scrubbers/params'
+require 'rollbar/util'
 require 'rollbar/util/ip_obfuscator'
 require 'rollbar/util/ip_anonymizer'
 require 'rollbar/json'
@@ -91,14 +92,23 @@ def scrub_params(params, sensitive_params)
     def mergeable_raw_body_params(rack_req)
       raw_body_params = rollbar_raw_body_params(rack_req)
 
-      case raw_body_params
-      when Hash
-        raw_body_params
-      when Array
-        { 'body.multi' => raw_body_params }
-      else
-        { 'body.value' => raw_body_params }
-      end
+      params = case raw_body_params
+               when Hash
+                 raw_body_params
+               when Array
+                 { 'body.multi' => raw_body_params }
+               else
+                 { 'body.value' => raw_body_params }
+               end
+
+      # The request body is serialized to JSON right here in the extractor,
+      # before the payload-wide UTF-8 pass in Rollbar::Item runs. A parsed
+      # body carrying invalid UTF-8 (e.g. a malformed request) would make that
+      # serialization raise JSON::GeneratorError and mask the original error
+      # being reported, so normalize the encoding up front.
+      Rollbar::Util.enforce_valid_utf8(params)
+
+      params
     end
 
     def rollbar_request_method(env)

spec/rollbar/request_data_extractor_spec.rb

@@ -322,6 +322,27 @@ class ExtractorDummy
       end
     end
 
+    context 'with JSON POST body containing invalid UTF-8' do
+      let(:body) { "{\"key\":\"\xAE\"}".b }
+      let(:env) do
+        Rack::MockRequest.env_for('/?foo=bar',
+                                  'CONTENT_TYPE' => 'application/json',
+                                  :input => body,
+                                  :method => 'POST')
+      end
+
+      it 'does not raise and serializes a valid body' do
+        result = nil
+
+        expect { result = subject.extract_request_data_from_rack(env) }
+          .not_to raise_error
+
+        expect(result[:body]).to be_kind_of(String)
+        expect(result[:body].encoding).to be_eql(Encoding::UTF_8)
+        expect(result[:body]).to be_valid_encoding
+      end
+    end
+
     context 'with JSON DELETE body' do
       let(:params) { { 'key' => 'value' } }
       let(:body) { params.to_json }