Redact InvalidURIError error messages

Merge branch 'security-37261-clean-passwords-from-invalid-uri-error-14-9' into '14-9-stable-ee'

See merge request gitlab-org/security/gitlab!2295

Changelog: security

lib/gitlab/error_tracking.rb

@@ -19,7 +19,8 @@ module Gitlab
     PROCESSORS = [
       ::Gitlab::ErrorTracking::Processor::SidekiqProcessor,
       ::Gitlab::ErrorTracking::Processor::GrpcErrorProcessor,
-      ::Gitlab::ErrorTracking::Processor::ContextPayloadProcessor
+      ::Gitlab::ErrorTracking::Processor::ContextPayloadProcessor,
+      ::Gitlab::ErrorTracking::Processor::SanitizeErrorMessageProcessor
     ].freeze
     class << self

lib/gitlab/error_tracking/processor/concerns/processes_exceptions.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module ErrorTracking
+    module Processor
+      module Concerns
+        module ProcessesExceptions
+          private
+
+          def extract_exceptions_from(event)
+            exceptions = if event.is_a?(Raven::Event)
+                           event.instance_variable_get(:@interfaces)[:exception]&.values
+                         else
+                           event&.exception&.instance_variable_get(:@values)
+                         end
+
+            Array.wrap(exceptions)
+          end
+
+          def set_exception_message(exception, message)
+            if exception.respond_to?(:value=)
+              exception.value = message
+            else
+              exception.instance_variable_set(:@value, message)
+            end
+          end
+
+          def valid_exception?(exception)
+            case exception
+            when Raven::SingleExceptionInterface, Sentry::SingleExceptionInterface
+              exception&.value.present?
+            else
+              false
+            end
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/error_tracking/processor/grpc_error_processor.rb

@@ -4,6 +4,8 @@ module Gitlab
   module ErrorTracking
     module Processor
       module GrpcErrorProcessor
+        extend Gitlab::ErrorTracking::Processor::Concerns::ProcessesExceptions
+
         DEBUG_ERROR_STRING_REGEX = RE2('(.*) debug_error_string:(.*)')
         class << self
@@ -19,9 +21,6 @@ module Gitlab
           def process_first_exception_value(event)
             # Better in new version, will be event.exception.values
             exceptions = extract_exceptions_from(event)
-
-            return unless exceptions.is_a?(Array)
-
             exception = exceptions.first
             return unless valid_exception?(exception)
@@ -39,11 +38,7 @@ module Gitlab
               exceptions.each do |exception|
                 next unless valid_exception?(exception)
-                if exception.respond_to?(:value=)
-                  exception.value = message
-                else
-                  exception.instance_variable_set(:@value, message)
-                end
+                set_exception_message(exception, message)
               end
             end
@@ -59,16 +54,6 @@ module Gitlab
             fingerprint[1] = message if message
           end
-          private
-
-          def extract_exceptions_from(event)
-            if event.is_a?(Raven::Event)
-              event.instance_variable_get(:@interfaces)[:exception]&.values
-            else
-              event.exception&.instance_variable_get(:@values)
-            end
-          end
-
           def custom_grpc_fingerprint?(fingerprint)
             fingerprint.is_a?(Array) && fingerprint.length == 2 && fingerprint[0].start_with?('GRPC::')
           end
@@ -82,15 +67,6 @@ module Gitlab
             [match[1], match[2]]
           end
-
-          def valid_exception?(exception)
-            case exception
-            when Raven::SingleExceptionInterface, Sentry::SingleExceptionInterface
-              exception&.value
-            else
-              false
-            end
-          end
         end
       end
     end

lib/gitlab/error_tracking/processor/sanitize_error_message_processor.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module ErrorTracking
+    module Processor
+      module SanitizeErrorMessageProcessor
+        extend Gitlab::ErrorTracking::Processor::Concerns::ProcessesExceptions
+
+        class << self
+          def call(event)
+            exceptions = extract_exceptions_from(event)
+
+            exceptions.each do |exception|
+              next unless valid_exception?(exception)
+
+              message = Gitlab::Sanitizers::ExceptionMessage.clean(exception.type, exception.value)
+
+              set_exception_message(exception, message)
+            end
+
+            event
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/exception_log_formatter.rb

@@ -10,7 +10,7 @@ module Gitlab
         # Use periods to flatten the fields.
         payload.merge!(
           'exception.class' => exception.class.name,
-          'exception.message' => exception.message
+          'exception.message' => sanitize_message(exception)
         )
         if exception.backtrace
@@ -38,6 +38,10 @@ module Gitlab
       rescue PgQuery::ParseError
         sql
       end
+
+      def sanitize_message(exception)
+        Gitlab::Sanitizers::ExceptionMessage.clean(exception.class.name, exception.message)
+      end
     end
   end
 end

lib/gitlab/sanitizers/exception_message.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Sanitizers
+    module ExceptionMessage
+      FILTERED_STRING = '[FILTERED]'
+      EXCEPTION_NAMES = %w(URI::InvalidURIError Addressable::URI::InvalidURIError).freeze
+      MESSAGE_REGEX = %r{(\A[^:]+:\s).*\Z}.freeze
+
+      class << self
+        def clean(exception_name, message)
+          return message unless exception_name.in?(EXCEPTION_NAMES)
+
+          message.sub(MESSAGE_REGEX, '\1' + FILTERED_STRING)
+        end
+      end
+    end
+  end
+end

spec/lib/gitlab/error_tracking/processor/sanitize_error_message_processor_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::ErrorTracking::Processor::SanitizeErrorMessageProcessor, :sentry do
+  describe '.call' do
+    let(:exception) { StandardError.new('raw error') }
+    let(:result_hash) { described_class.call(event).to_hash }
+
+    shared_examples 'processes the exception' do
+      it 'cleans the exception message' do
+        expect(Gitlab::Sanitizers::ExceptionMessage).to receive(:clean).with('StandardError', 'raw error').and_return('cleaned')
+
+        expect(result_hash[:exception][:values].first).to include(
+          type: 'StandardError',
+          value: 'cleaned'
+        )
+      end
+    end
+
+    context 'with Raven event' do
+      let(:raven_required_options) do
+        {
+          configuration: Raven.configuration,
+          context: Raven.context,
+          breadcrumbs: Raven.breadcrumbs
+        }
+      end
+
+      let(:event) { Raven::Event.from_exception(exception, raven_required_options) }
+
+      it_behaves_like 'processes the exception'
+    end
+
+    context 'with Sentry event' do
+      let(:event) { Sentry.get_current_client.event_from_exception(exception) }
+
+      it_behaves_like 'processes the exception'
+    end
+
+    context 'with invalid event' do
+      let(:event) { instance_double('Sentry::Event', to_hash: { invalid: true }) }
+
+      it 'does nothing' do
+        extracted_exception = instance_double('Sentry::SingleExceptionInterface', value: nil)
+        allow(described_class).to receive(:extract_exceptions_from).and_return([extracted_exception])
+
+        expect(Gitlab::Sanitizers::ExceptionMessage).not_to receive(:clean)
+        expect(result_hash).to eq(invalid: true)
+      end
+    end
+  end
+end

spec/lib/gitlab/error_tracking_spec.rb

@@ -368,5 +368,61 @@ RSpec.describe Gitlab::ErrorTracking do
         end
       end
     end
+
+    context 'when processing invalid URI exceptions' do
+      let(:invalid_uri) { 'http://foo:bar' }
+      let(:raven_exception_values) { raven_event['exception']['values'] }
+      let(:sentry_exception_values) { sentry_event.exception.to_hash[:values] }
+
+      context 'when the error is a URI::InvalidURIError' do
+        let(:exception) do
+          URI.parse(invalid_uri)
+        rescue URI::InvalidURIError => error
+          error
+        end
+
+        it 'filters the URI from the error message' do
+          track_exception
+
+          expect(raven_exception_values).to include(
+            hash_including(
+              'type' => 'URI::InvalidURIError',
+              'value' => 'bad URI(is not URI?): [FILTERED]'
+            )
+          )
+          expect(sentry_exception_values).to include(
+            hash_including(
+              type: 'URI::InvalidURIError',
+              value: 'bad URI(is not URI?): [FILTERED]'
+            )
+          )
+        end
+      end
+
+      context 'when the error is a Addressable::URI::InvalidURIError' do
+        let(:exception) do
+          Addressable::URI.parse(invalid_uri)
+        rescue Addressable::URI::InvalidURIError => error
+          error
+        end
+
+        it 'filters the URI from the error message' do
+          track_exception
+
+          expect(raven_exception_values).to include(
+            hash_including(
+              'type' => 'Addressable::URI::InvalidURIError',
+              'value' => 'Invalid port number: [FILTERED]'
+            )
+          )
+          expect(sentry_exception_values).to include(
+            hash_including(
+              type: 'Addressable::URI::InvalidURIError',
+              value: 'Invalid port number: [FILTERED]'
+            )
+          )
+        end
+      end
+    end
   end
 end

spec/lib/gitlab/exception_log_formatter_spec.rb

@@ -22,6 +22,14 @@ RSpec.describe Gitlab::ExceptionLogFormatter do
       expect(payload['exception.sql']).to be_nil
     end
+    it 'cleans the exception message' do
+      expect(Gitlab::Sanitizers::ExceptionMessage).to receive(:clean).with('RuntimeError', 'bad request').and_return('cleaned')
+
+      described_class.format!(exception, payload)
+
+      expect(payload['exception.message']).to eq('cleaned')
+    end
+
     context 'when exception is ActiveRecord::StatementInvalid' do
       let(:exception) { ActiveRecord::StatementInvalid.new(sql: 'SELECT "users".* FROM "users" WHERE "users"."id" = 1 AND "users"."foo" = $1') }

spec/lib/gitlab/sanitizers/exception_message_spec.rb

+# frozen_string_literal: true
+
+require 'fast_spec_helper'
+require 'rspec-parameterized'
+
+RSpec.describe Gitlab::Sanitizers::ExceptionMessage do
+  describe '.clean' do
+    let(:exception_name) { exception.class.name }
+    let(:exception_message) { exception.message }
+
+    subject { described_class.clean(exception_name, exception_message) }
+
+    context 'when error is a URI::InvalidURIError' do
+      let(:exception) do
+        URI.parse('http://foo:bar')
+      rescue URI::InvalidURIError => error
+        error
+      end
+
+      it { is_expected.to eq('bad URI(is not URI?): [FILTERED]') }
+    end
+
+    context 'when error is an Addressable::URI::InvalidURIError' do
+      using RSpec::Parameterized::TableSyntax
+
+      let(:exception) do
+        Addressable::URI.parse(uri)
+      rescue Addressable::URI::InvalidURIError => error
+        error
+      end
+
+      where(:uri, :result) do
+        'http://foo:bar' | 'Invalid port number: [FILTERED]'
+        'http://foo:%eb' | 'Invalid encoding in port'
+        'ht%0atp://foo'  | 'Invalid scheme format: [FILTERED]'
+        'http:'          | 'Absolute URI missing hierarchical segment: [FILTERED]'
+        '::http'         | 'Cannot assemble URI string with ambiguous path: [FILTERED]'
+        'http://foo bar' | 'Invalid character in host: [FILTERED]'
+      end
+
+      with_them do
+        it { is_expected.to eq(result) }
+      end
+    end
+
+    context 'with any other exception' do
+      let(:exception) { StandardError.new('Error message: http://foo@bar:baz@ex:ample.com') }
+
+      it 'is not invoked and does nothing' do
+        is_expected.to eq('Error message: http://foo@bar:baz@ex:ample.com')
+      end
+    end
+  end
+end