Update parquet-go and drop string workaround

With parquet-go v0.26.0 it now properly handles optional []byte data so
we no longer need to store it as a string.

```
go get github.com/parquet-go/parquet-go
go mod tidy
```

Using a []byte also means we can make the test code check for nil rather
than a zero length string which is what we actually want.

Author: eest

go.mod

@@ -12,7 +12,7 @@ require (
 	github.com/hashicorp/golang-lru/v2 v2.0.7
 	github.com/lestrrat-go/jwx/v2 v2.1.6
 	github.com/miekg/dns v1.1.68
-	github.com/parquet-go/parquet-go v0.25.1
+	github.com/parquet-go/parquet-go v0.26.0
 	github.com/prometheus/client_golang v1.23.2
 	github.com/segmentio/go-hll v1.0.1
 	github.com/smhanov/dawg v0.0.0-20220118194912-66057bdbf2e3
@@ -65,6 +65,8 @@ require (
 	github.com/lestrrat-go/option v1.0.1 // indirect
 	github.com/lestrrat-go/option/v2 v2.0.0 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
+	github.com/parquet-go/bitpack v0.2.0 // indirect
+	github.com/parquet-go/jsonlite v0.8.1 // indirect
 	github.com/pelletier/go-toml/v2 v2.2.4 // indirect
 	github.com/pierrec/lz4/v4 v4.1.22 // indirect
 	github.com/pkg/errors v0.9.1 // indirect

go.sum

@@ -127,8 +127,12 @@ github.com/miekg/dns v1.1.68 h1:jsSRkNozw7G/mnmXULynzMNIsgY2dHC8LO6U6Ij2JEA=
 github.com/miekg/dns v1.1.68/go.mod h1:fujopn7TB3Pu3JM69XaawiU0wqjpL9/8xGop5UrTPps=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
-github.com/parquet-go/parquet-go v0.25.1 h1:l7jJwNM0xrk0cnIIptWMtnSnuxRkwq53S+Po3KG8Xgo=
-github.com/parquet-go/parquet-go v0.25.1/go.mod h1:AXBuotO1XiBtcqJb/FKFyjBG4aqa3aQAAWF3ZPzCanY=
+github.com/parquet-go/bitpack v0.2.0 h1:1qA39QcA+HeExChZOATm78XMs5W2NY/Y2l17M5kDUuE=
+github.com/parquet-go/bitpack v0.2.0/go.mod h1:XnVk9TH+O40eOOmvpAVZ7K2ocQFrQwysLMnc6M/8lgs=
+github.com/parquet-go/jsonlite v0.8.1 h1:TdvfyPaVLTlz/Zsl+amWO4h0tpEwXwRkd7xa4iPhL5E=
+github.com/parquet-go/jsonlite v0.8.1/go.mod h1:nDjpkpL4EOtqs6NQugUsi0Rleq9sW/OtC1NnZEnxzF0=
+github.com/parquet-go/parquet-go v0.26.0 h1:5rWuYYCKouRlo1kLihNAcw2+mb/OLJhIZjjpFu1lX9k=
+github.com/parquet-go/parquet-go v0.26.0/go.mod h1:7K8PVhWjeOLCtcV0cT3DFMfegbcM9uwvVNc2F+Cmsw4=
 github.com/pelletier/go-toml/v2 v2.2.4 h1:mye9XuhQ6gvn5h28+VilKrrPoQVanw5PMw/TB0t5Ec4=
 github.com/pelletier/go-toml/v2 v2.2.4/go.mod h1:2gIqNv+qfxSVS7cM2xJQKtLSTLUE9V8t9Stt+h56mCY=
 github.com/pierrec/lz4/v4 v4.1.22 h1:cKFw6uJDK+/gfw5BcDL0JL5aBsAFdsIT18eRtLj7VIU=

pkg/runner/runner.go

@@ -167,15 +167,10 @@ type histogramData struct {
 	V4ClientCount uint64 `parquet:"v4client_count"`
 	V6ClientCount uint64 `parquet:"v6client_count"`
 
-	// Would probably be cleaner to use a []byte instead of string with
-	// struct tag "bytes" here, but it seems the parquet-go library does
-	// not handle "optional" []byte fields correctly right now, see:
-	// https://github.com/parquet-go/parquet-go/issues/303
-	//
 	// These fields are NULL when HLL uses explicit storage, otherwise
 	// contain the probabilistic HLL bytes
-	V4ClientCountHLLBytes string `parquet:"v4client_count_hll,bytes,optional"`
-	V6ClientCountHLLBytes string `parquet:"v6client_count_hll,bytes,optional"`
+	V4ClientCountHLLBytes []byte `parquet:"v4client_count_hll,optional"`
+	V6ClientCountHLLBytes []byte `parquet:"v6client_count_hll,optional"`
 }
 
 // We need to create the session data schema by hand instead of basing it of
@@ -2720,10 +2715,10 @@ func (edm *dnstapMinimiser) writeHistogramParquet(output io.Writer, startTime ti
 
 		// Include bytes from our hll data structures if they are stored with a probabilistic storage type
 		if v4HLLType == hllSparse || v4HLLType == hllDense {
-			hGramData.V4ClientCountHLLBytes = string(v4HLLBytes)
+			hGramData.V4ClientCountHLLBytes = v4HLLBytes
 		}
 		if v6HLLType == hllSparse || v6HLLType == hllDense {
-			hGramData.V6ClientCountHLLBytes = string(v6HLLBytes)
+			hGramData.V6ClientCountHLLBytes = v6HLLBytes
 		}
 
 		_, err = parquetWriter.Write([]histogramData{*hGramData})

pkg/runner/runner_test.go

@@ -1879,8 +1879,8 @@ func TestHistogramWriter(t *testing.T) {
 		FailCount:             19,
 		OtherRcodeCount:       20,
 		EDMStatusBits:         21,
-		V4ClientCountHLLBytes: string(v4hll.ToBytes()),
-		V6ClientCountHLLBytes: string(v6hll.ToBytes()),
+		V4ClientCountHLLBytes: v4hll.ToBytes(),
+		V6ClientCountHLLBytes: v6hll.ToBytes(),
 	}
 
 	_, err = parquetWriter.Write([]histogramData{hd})
@@ -1957,8 +1957,8 @@ func BenchmarkHistogramWriter(b *testing.B) {
 		FailCount:             19,
 		OtherRcodeCount:       20,
 		EDMStatusBits:         21,
-		V4ClientCountHLLBytes: string(v4hll.ToBytes()),
-		V6ClientCountHLLBytes: string(v6hll.ToBytes()),
+		V4ClientCountHLLBytes: v4hll.ToBytes(),
+		V6ClientCountHLLBytes: v6hll.ToBytes(),
 	}
 
 	for b.Loop() {
@@ -2130,14 +2130,14 @@ func TestWriteHistogramParquetExplicitThreshold(t *testing.T) {
 		}
 
 		for _, row := range rows {
-			if test.ipv4HllIsNull && len(row.V4ClientCountHLLBytes) != 0 {
-				t.Fatalf("IPv4 HLL data should be length 0 but is %d", len(row.V4ClientCountHLLBytes))
+			if test.ipv4HllIsNull && row.V4ClientCountHLLBytes != nil {
+				t.Fatalf("IPv4 HLL data should be nil but is %#v", row.V4ClientCountHLLBytes)
 			}
 			if !test.ipv4HllIsNull && len(row.V4ClientCountHLLBytes) == 0 {
 				t.Fatal("IPv4 HLL data is 0 when it should have content")
 			}
-			if test.ipv6HllIsNull && len(row.V6ClientCountHLLBytes) != 0 {
-				t.Fatalf("IPv6 HLL data should be length 0 but is %d", len(row.V6ClientCountHLLBytes))
+			if test.ipv6HllIsNull && row.V6ClientCountHLLBytes != nil {
+				t.Fatalf("IPv6 HLL data should be nil but is %#v", row.V6ClientCountHLLBytes)
 			}
 			if !test.ipv6HllIsNull && len(row.V6ClientCountHLLBytes) == 0 {
 				t.Fatal("IPv6 HLL data is 0 when it should have content")