Adding tests to ensure geoip emits UTF-8 encoded fields.

Makefile

@@ -17,6 +17,8 @@ COLLECTD_VERSION=5.4.0
 TYPESDB_URL=https://collectd.org/files/collectd-$(COLLECTD_VERSION).tar.gz
 GEOIP=vendor/geoip/GeoLiteCity.dat
 GEOIP_URL=http://logstash.objects.dreamhost.com/maxmind/GeoLiteCity-2013-01-18.dat.gz
+GEOIP_ASN=vendor/geoip/GeoIPASNum.dat
+GEOIP_ASN_URL=http://logstash.objects.dreamhost.com/maxmind/GeoIPASNum-2014-02-12.dat.gz
 KIBANA_URL=https://download.elasticsearch.org/kibana/kibana/kibana-3.0.0milestone5.tar.gz
 PLUGIN_FILES=$(shell git ls-files | egrep '^lib/logstash/(inputs|outputs|filters|codecs)/[^/]+$$' | egrep -v '/(base|threadable).rb$$|/inputs/ganglia/')
 QUIET=@
@@ -165,6 +167,13 @@ $(GEOIP): | vendor/geoip
 	$(QUIET)rm "$@.tmp.gz"
 	$(QUIET)mv $@.tmp $@
 
+$(GEOIP_ASN): | vendor/geoip
+	$(QUIET)$(DOWNLOAD_COMMAND) $@.tmp.gz $(GEOIP_ASN_URL)
+	$(QUIET)gzip -dc $@.tmp.gz > $@.tmp
+	$(QUIET)rm "$@.tmp.gz"
+	$(QUIET)mv $@.tmp $@
+
+vendor/collectd: | vendor
 vendor/collectd: | vendor
 	$(QUIET)mkdir $@
 

spec/filters/geoip.rb

@@ -59,4 +59,54 @@
       reject { subject }.include?("src_ip")
     end
   end
+
+  describe "correct encodings with default db" do
+    config <<-CONFIG
+      filter {
+        geoip {
+          source => "ip"
+        }
+      }
+    CONFIG
+    expected_fields = %w(ip country_code2 country_code3 country_name
+                           continent_code region_name city_name postal_code
+                           dma_code area_code timezone)
+
+    sample("ip" => "1.1.1.1") do
+      expected_fields.each do |f|
+        insist { subject["geoip"][f].encoding } == Encoding::UTF_8
+      end
+    end
+    sample("ip" => "189.2.0.0") do
+      expected_fields.each do |f|
+        insist { subject["geoip"][f].encoding } == Encoding::UTF_8
+      end
+    end
+
+  end
+
+  describe "correct encodings with ASN db" do
+    config <<-CONFIG
+      filter {
+        geoip {
+          source => "ip"
+          database => "vendor/geoip/GeoIPASNum.dat"
+        }
+      }
+    CONFIG
+
+
+    sample("ip" => "1.1.1.1") do
+      insist { subject["geoip"]["asn"].encoding } == Encoding::UTF_8
+    end
+    sample("ip" => "187.2.0.0") do
+      insist { subject["geoip"]["asn"].encoding } == Encoding::UTF_8
+    end
+    sample("ip" => "189.2.0.0") do
+      insist { subject["geoip"]["asn"].encoding } == Encoding::UTF_8
+    end
+    sample("ip" => "161.24.0.0") do
+      insist { subject["geoip"]["asn"].encoding } == Encoding::UTF_8
+    end
+  end
 end