Reduce default registry lookup allocations

defaults.go

@@ -11,8 +11,12 @@ func IsDefaultRegistry(purlType, registryURL string) bool {
 		return true
 	}
 
-	cfg := TypeInfo(purlType)
-	if cfg == nil || cfg.DefaultRegistry == nil {
+	data, err := loadTypes()
+	if err != nil {
+		return false
+	}
+	cfg, ok := data.Types[purlType]
+	if !ok || cfg.DefaultRegistry == nil {
 		return false
 	}
 
@@ -21,8 +25,11 @@ func IsDefaultRegistry(purlType, registryURL string) bool {
 		return false
 	}
 
-	// Compare hosts
-	defaultHost := extractHost(defaultURL)
+	defaultHost, ok := data.defaultHosts[defaultURL]
+	if !ok {
+		// TypeInfo exposes the default URL through a shared pointer.
+		defaultHost = extractHost(defaultURL)
+	}
 	givenHost := extractHost(registryURL)
 
 	if defaultHost == "" || givenHost == "" {

defaults_cache_test.go

@@ -0,0 +1,96 @@
+package purl
+
+import (
+	"net/url"
+	"strings"
+	"testing"
+)
+
+func TestBuildPURLRegistryHostSemantics(t *testing.T) {
+	for _, tc := range []struct {
+		registry  string
+		isDefault bool
+	}{
+		{"https://registry.npmjs.org/package/-/package.tgz", true},
+		{"http://registry.npmjs.org:8080/package", true},
+		{"//registry.npmjs.org/package", true},
+		{"https://cdn.registry.npmjs.org/package", true},
+		{"https://user:pass@registry.npmjs.org/package", true},
+		{"https://evil.example@registry.npmjs.org/package", true},
+		{"https://registry.npmjs.org@evil.example/package", false},
+		{"https://registry.npmjs.org.evil.example/package", false},
+		{"https://notregistry.npmjs.org/package", false},
+		{"https://REGISTRY.NPMJS.ORG/package", false},
+		{"https://registry.npmjs.org./package", false},
+		{"https://registry.npmjs.org:bad/package", false},
+		{"https://registry.npmjs.org/%zz", false},
+		{"https://[::1]/package", false},
+		{"registry.npmjs.org/package", false},
+	} {
+		t.Run(tc.registry, func(t *testing.T) {
+			got := BuildPURLString("npm", "package", "1.2.3", tc.registry)
+			if tc.isDefault {
+				if got != "pkg:npm/package@1.2.3" {
+					t.Fatalf("default registry retained: %q", got)
+				}
+			} else if !strings.HasPrefix(got, "pkg:npm/package@1.2.3?repository_url=") {
+				t.Fatalf("registry lost: %q", got)
+			}
+		})
+	}
+}
+
+func TestDefaultRegistryPointerChanges(t *testing.T) {
+	cfg := TypeInfo("npm")
+	original := *cfg.DefaultRegistry
+	t.Cleanup(func() { *cfg.DefaultRegistry = original })
+	for _, registry := range []string{"https://custom.example/registry", "https://crates.io", "", "https://[invalid", original} {
+		*cfg.DefaultRegistry = registry
+		for _, input := range []string{original, "https://custom.example/package.tgz", "https://crates.io", ""} {
+			want := uncachedDefaultRegistry("npm", input)
+			if got := IsDefaultRegistry("npm", input); got != want {
+				t.Fatalf("default %q, input %q: got %t, want %t", registry, input, got, want)
+			}
+			got := BuildPURLString("npm", "package", "1.2.3", input)
+			if strings.Contains(got, "?repository_url=") == want {
+				t.Fatalf("default %q, input %q: unexpected PURL %q", registry, input, got)
+			}
+		}
+	}
+}
+
+func uncachedDefaultRegistry(purlType, registry string) bool {
+	if registry == "" {
+		return true
+	}
+	cfg := TypeInfo(purlType)
+	if cfg == nil || cfg.DefaultRegistry == nil {
+		return false
+	}
+	standard, err := url.Parse(*cfg.DefaultRegistry)
+	if err != nil || standard.Hostname() == "" {
+		return false
+	}
+	given, err := url.Parse(registry)
+	if err != nil || given.Hostname() == "" {
+		return false
+	}
+	return given.Hostname() == standard.Hostname() || strings.HasSuffix(given.Hostname(), "."+standard.Hostname())
+}
+
+func FuzzDefaultRegistryCompatibility(f *testing.F) {
+	for _, kind := range append(KnownTypes(), "unknown") {
+		f.Add(kind, DefaultRegistry(kind))
+		f.Add(kind, "https://user@registry.npmjs.org:443/package.tgz")
+		f.Add(kind, "https://registry.npmjs.org@evil.example/%zz")
+	}
+	f.Fuzz(func(t *testing.T, kind, registry string) {
+		want := uncachedDefaultRegistry(kind, registry)
+		if got := IsDefaultRegistry(kind, registry); got != want {
+			t.Fatalf("IsDefaultRegistry(%q, %q) = %t, want %t", kind, registry, got, want)
+		}
+		if got := IsNonDefaultRegistry(kind, registry); got == want {
+			t.Fatalf("IsNonDefaultRegistry(%q, %q) = %t, want %t", kind, registry, got, !want)
+		}
+	})
+}

makepurl_bench_test.go

@@ -0,0 +1,89 @@
+package purl
+
+import (
+	"fmt"
+	"testing"
+)
+
+var benchmarkRegistryDefault bool
+
+var resolvedURLCases = []struct {
+	name     string
+	registry string
+	want     string
+}{
+	{"none", "", "pkg:npm/%40scope/package@1.2.3"},
+	{"default", "https://registry.npmjs.org/@scope/package/-/package-1.2.3.tgz", "pkg:npm/%40scope/package@1.2.3"},
+	{"subdomain", "https://cdn.registry.npmjs.org/@scope/package/-/package-1.2.3.tgz", "pkg:npm/%40scope/package@1.2.3"},
+	{"private", "https://npm.example.invalid/@scope/package/-/package-1.2.3.tgz", "pkg:npm/%40scope/package@1.2.3?repository_url=https:%2F%2Fnpm.example.invalid%2F%40scope%2Fpackage%2F-%2Fpackage-1.2.3.tgz"},
+}
+
+func TestBuildPURLResolvedURLs(t *testing.T) {
+	for _, tc := range resolvedURLCases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := BuildPURLString("npm", "@scope/package", "1.2.3", tc.registry); got != tc.want {
+				t.Fatalf("got %q, want %q", got, tc.want)
+			}
+		})
+	}
+}
+
+func BenchmarkBuildPURLResolvedURL(b *testing.B) {
+	for _, tc := range resolvedURLCases {
+		b.Run(tc.name, func(b *testing.B) {
+			b.ReportAllocs()
+			for b.Loop() {
+				benchmarkVersionPURLString = BuildPURLString("npm", "@scope/package", "1.2.3", tc.registry)
+			}
+			if benchmarkVersionPURLString != tc.want {
+				b.Fatalf("got %q, want %q", benchmarkVersionPURLString, tc.want)
+			}
+		})
+	}
+}
+
+func BenchmarkBuildPURLResolvedURLMixed(b *testing.B) {
+	const count = 1000
+	type input struct{ name, url, want string }
+	inputs := make([]input, count)
+	for i := range inputs {
+		name := fmt.Sprintf("@scope/package-%d", i)
+		inputs[i] = input{
+			name: name,
+			url:  "https://registry.npmjs.org/" + name + "/-/package-1.2.3.tgz",
+			want: fmt.Sprintf("pkg:npm/%%40scope/package-%d@1.2.3", i),
+		}
+		if got := BuildPURLString("npm", name, "1.2.3", inputs[i].url); got != inputs[i].want {
+			b.Fatalf("got %q, want %q", got, inputs[i].want)
+		}
+	}
+	i := 0
+	b.ReportAllocs()
+	for b.Loop() {
+		in := inputs[i%count]
+		benchmarkVersionPURLString = BuildPURLString("npm", in.name, "1.2.3", in.url)
+		i++
+	}
+}
+
+func BenchmarkDefaultRegistryResolvedURL(b *testing.B) {
+	for _, tc := range resolvedURLCases {
+		b.Run(tc.name, func(b *testing.B) {
+			b.ReportAllocs()
+			for b.Loop() {
+				benchmarkRegistryDefault = IsDefaultRegistry("npm", tc.registry)
+			}
+		})
+	}
+}
+
+func BenchmarkCleanVersionNPM(b *testing.B) {
+	for _, version := range []string{"1.2.3", "1.2.3-beta.1", "^1.2.3"} {
+		b.Run(version, func(b *testing.B) {
+			b.ReportAllocs()
+			for b.Loop() {
+				benchmarkVersionPURLString = CleanVersion(version, "npm")
+			}
+		})
+	}
+}

types.go

@@ -55,11 +55,12 @@ func (t *TypeConfig) NamespaceProhibited() bool {
 }
 
 type typesData struct {
-	Version     string                `json:"version"`
-	Description string                `json:"description"`
-	Source      string                `json:"source"`
-	LastUpdated string                `json:"last_updated"`
-	Types       map[string]TypeConfig `json:"types"`
+	defaultHosts map[string]string
+	Version      string                `json:"version"`
+	Description  string                `json:"description"`
+	Source       string                `json:"source"`
+	LastUpdated  string                `json:"last_updated"`
+	Types        map[string]TypeConfig `json:"types"`
 }
 
 var (
@@ -72,6 +73,13 @@ func loadTypes() (*typesData, error) {
 	loadOnce.Do(func() {
 		loadedData = &typesData{}
 		loadErr = json.Unmarshal(typesJSON, loadedData)
+		loadedData.defaultHosts = make(map[string]string)
+		for _, cfg := range loadedData.Types {
+			if cfg.DefaultRegistry != nil {
+				registry := *cfg.DefaultRegistry
+				loadedData.defaultHosts[registry] = extractHost(registry)
+			}
+		}
 	})
 	return loadedData, loadErr
 }