Merge branch 'dedup4'

sanitize.go

@@ -0,0 +1,84 @@
+package dns
+
+// Dedup removes identical RRs from rrs. It preserves the original ordering.
+// The lowest TTL of any duplicates is used in the remaining one. Dedup modifies
+// rrs.
+// m is used to store the RRs temporay. If it is nil a new map will be allocated.
+func Dedup(rrs []RR, m map[string]RR) []RR {
+	if m == nil {
+		m = make(map[string]RR)
+	}
+	// Save the keys, so we don't have to call normalizedString twice.
+	keys := make([]*string, 0, len(rrs))
+
+	for _, r := range rrs {
+		key := normalizedString(r)
+		keys = append(keys, &key)
+		if _, ok := m[key]; ok {
+			// Shortest TTL wins.
+			if m[key].Header().Ttl > r.Header().Ttl {
+				m[key].Header().Ttl = r.Header().Ttl
+			}
+			continue
+		}
+
+		m[key] = r
+	}
+	// If the length of the result map equals the amount of RRs we got,
+	// it means they were all different. We can then just return the original rrset.
+	if len(m) == len(rrs) {
+		return rrs
+	}
+
+	j := 0
+	for i, r := range rrs {
+		// If keys[i] lives in the map, we should copy and remove it.
+		if _, ok := m[*keys[i]]; ok {
+			delete(m, *keys[i])
+			rrs[j] = r
+			j++
+		}
+
+		if len(m) == 0 {
+			break
+		}
+	}
+
+	return rrs[:j]
+}
+
+// normalizedString returns a normalized string from r. The TTL
+// is removed and the domain name is lowercased. We go from this:
+// DomainName<TAB>TTL<TAB>CLASS<TAB>TYPE<TAB>RDATA to:
+// lowercasename<TAB>CLASS<TAB>TYPE...
+func normalizedString(r RR) string {
+	// A string Go DNS makes has: domainname<TAB>TTL<TAB>...
+	b := []byte(r.String())
+
+	// find the first non-escaped tab, then another, so we capture where the TTL lives.
+	esc := false
+	ttlStart, ttlEnd := 0, 0
+	for i := 0; i < len(b) && ttlEnd == 0; i++ {
+		switch {
+		case b[i] == '\\':
+			esc = !esc
+		case b[i] == '\t' && !esc:
+			if ttlStart == 0 {
+				ttlStart = i
+				continue
+			}
+			if ttlEnd == 0 {
+				ttlEnd = i
+			}
+		case b[i] >= 'A' && b[i] <= 'Z' && !esc:
+			b[i] += 32
+		default:
+			esc = false
+		}
+	}
+
+	// remove TTL.
+	copy(b[ttlStart:], b[ttlEnd:])
+	cut := ttlEnd - ttlStart
+	return string(b[:len(b)-cut])
+}

sanitize_test.go

@@ -0,0 +1,85 @@
+package dns
+
+import "testing"
+
+func TestDedup(t *testing.T) {
+	// make it []string
+	testcases := map[[3]RR][]string{
+		[...]RR{
+			newRR(t, "mIek.nl. IN A 127.0.0.1"),
+			newRR(t, "mieK.nl. IN A 127.0.0.1"),
+			newRR(t, "miek.Nl. IN A 127.0.0.1"),
+		}: []string{"mIek.nl.\t3600\tIN\tA\t127.0.0.1"},
+		[...]RR{
+			newRR(t, "miEk.nl. 2000 IN A 127.0.0.1"),
+			newRR(t, "mieK.Nl. 1000 IN A 127.0.0.1"),
+			newRR(t, "Miek.nL. 500 IN A 127.0.0.1"),
+		}: []string{"miEk.nl.\t500\tIN\tA\t127.0.0.1"},
+		[...]RR{
+			newRR(t, "miek.nl. IN A 127.0.0.1"),
+			newRR(t, "miek.nl. CH A 127.0.0.1"),
+			newRR(t, "miek.nl. IN A 127.0.0.1"),
+		}: []string{"miek.nl.\t3600\tIN\tA\t127.0.0.1",
+			"miek.nl.\t3600\tCH\tA\t127.0.0.1",
+		},
+		[...]RR{
+			newRR(t, "miek.nl. CH A 127.0.0.1"),
+			newRR(t, "miek.nl. IN A 127.0.0.1"),
+			newRR(t, "miek.de. IN A 127.0.0.1"),
+		}: []string{"miek.nl.\t3600\tCH\tA\t127.0.0.1",
+			"miek.nl.\t3600\tIN\tA\t127.0.0.1",
+			"miek.de.\t3600\tIN\tA\t127.0.0.1",
+		},
+		[...]RR{
+			newRR(t, "miek.de. IN A 127.0.0.1"),
+			newRR(t, "miek.nl. 200 IN A 127.0.0.1"),
+			newRR(t, "miek.nl. 300 IN A 127.0.0.1"),
+		}: []string{"miek.de.\t3600\tIN\tA\t127.0.0.1",
+			"miek.nl.\t200\tIN\tA\t127.0.0.1",
+		},
+	}
+
+	for rr, expected := range testcases {
+		out := Dedup([]RR{rr[0], rr[1], rr[2]}, nil)
+		for i, o := range out {
+			if o.String() != expected[i] {
+				t.Fatalf("expected %v, got %v", expected[i], o.String())
+			}
+		}
+	}
+}
+
+func BenchmarkDedup(b *testing.B) {
+	rrs := []RR{
+		newRR(nil, "miEk.nl. 2000 IN A 127.0.0.1"),
+		newRR(nil, "mieK.Nl. 1000 IN A 127.0.0.1"),
+		newRR(nil, "Miek.nL. 500 IN A 127.0.0.1"),
+	}
+	m := make(map[string]RR)
+	for i := 0; i < b.N; i++ {
+		Dedup(rrs,m )
+	}
+}
+
+func TestNormalizedString(t *testing.T) {
+	tests := map[RR]string{
+		newRR(t, "mIEk.Nl. 3600 IN A 127.0.0.1"):     "miek.nl.\tIN\tA\t127.0.0.1",
+		newRR(t, "m\\ iek.nL. 3600 IN A 127.0.0.1"):  "m\\ iek.nl.\tIN\tA\t127.0.0.1",
+		newRR(t, "m\\\tIeK.nl. 3600 in A 127.0.0.1"): "m\\tiek.nl.\tIN\tA\t127.0.0.1",
+	}
+	for tc, expected := range tests {
+		n := normalizedString(tc)
+		if n != expected {
+			t.Logf("expected %s, got %s", expected, n)
+			t.Fail()
+		}
+	}
+}
+
+func newRR(t *testing.T, s string) RR {
+	r, e := NewRR(s)
+	if e != nil {
+		t.Logf("newRR: %s", e)
+	}
+	return r
+}