jarl/internal/blocking/blocking_test.go

package blocking

import (
	"errors"
	"os"
	"path/filepath"
	"strings"
	"testing"
)

func TestParse(t *testing.T) {
	src := "\uFEFF[Adblock Plus]\r\n" + `! Title: test list
# a comment

||ads.example^
||Tracker.Example.^$important
||cosmetic.example^$third-party
@@||allowed.example^
example.org##.banner
|https://url.example/path
0.0.0.0 hosts.example other.example  # two names
127.0.0.1 localhost
127.0.0.1 localhost.localdomain
::1 ip6-localhost
0.0.0.0 0.0.0.0
plain.example
*.wild.example
single
1.2.3.4
bad_label!.example
münchen.example
www.ads.example
` + strings.Repeat("a", 64) + ".example\n"
	l, err := Parse(strings.NewReader(src), Limits{})
	if err != nil {
		t.Fatal(err)
	}
	want := map[string]uint8{
		"ads.example": Exact | Below, "tracker.example": Exact | Below,
		"hosts.example": Exact, "other.example": Exact, "plain.example": Exact,
		"wild.example": Exact | Below, "xn--mnchen-3ya.example": Exact, "www.ads.example": Exact,
	}
	if len(l.Domains) != len(want) {
		t.Errorf("domains = %v", l.Domains)
	}
	for n, f := range want {
		if l.Domains[n] != f {
			t.Errorf("%s = %d, want %d", n, l.Domains[n], f)
		}
	}
	// Skipped: $third-party, @@, ##, |url, localhost (x3), 0.0.0.0, single, 1.2.3.4, bad_label!, 64-char label.
	if l.Skipped != 12 || l.Lines != 19 {
		t.Errorf("lines %d, skipped %d", l.Lines, l.Skipped)
	}

	Compact(l)
	if _, ok := l.Domains["www.ads.example"]; ok {
		t.Error("www.ads.example is covered by ||ads.example^ and should be dropped")
	}
	var sb strings.Builder
	n, hash, err := FeedZone(&sb, "test", 7, l)
	if err != nil || n != 10 {
		t.Fatalf("FeedZone: %d records, %v", n, err)
	}
	for _, line := range []string{"ads.example\tCNAME\t.\n", "*.ads.example\tCNAME\t.\n", "plain.example\tCNAME\t.\n", "@\tSOA\tlocalhost. hostmaster.localhost. 7 "} {
		if !strings.Contains(sb.String(), line) {
			t.Errorf("zone lacks %q:\n%s", line, sb.String())
		}
	}
	if strings.Contains(sb.String(), "*.plain.example") || strings.Contains(sb.String(), "$ORIGIN") {
		t.Errorf("zone:\n%s", sb.String())
	}
	// The hash ignores the serial.
	var other strings.Builder
	if _, h2, _ := FeedZone(&other, "test", 8, l); h2 != hash {
		t.Error("hash depends on the serial")
	}
}

func TestParseLimits(t *testing.T) {
	_, err := Parse(strings.NewReader("a.example\nb.example\na.example\nc.example\n"), Limits{MaxEntries: 2})
	if !errors.Is(err, ErrTooManyEntries) {
		t.Errorf("err = %v", err)
	}
	if _, err := Parse(strings.NewReader("a.example\nb.example\na.example\n"), Limits{MaxEntries: 2}); err != nil {
		t.Errorf("a repeated name counted against the limit: %v", err)
	}
	l, _ := Parse(strings.NewReader("long-name.example\nshort.example\n"), Limits{MaxNameLen: 15})
	if len(l.Domains) != 1 || l.Skipped != 1 {
		t.Errorf("max name length: %v", l.Domains)
	}
}

func TestRuleEntries(t *testing.T) {
	rules := []Rule{
		{Name: "ads.example", Action: ActionBlock},                        // and everything below
		{Name: "cdn.ads.example", Action: ActionAllow, Exact: true},       // just this name
		{Name: "wide.example", Action: ActionBlock},                       // and below
		{Name: "ok.wide.example", Action: ActionAllow},                    // and below
		{Name: "deep.x.y.wide.example", Action: ActionAllow, Exact: true}, // two empty names in between
		{Name: "solo.example", Action: ActionBlock, Exact: true},
	}
	e := RuleEntries(rules)
	for name, want := range map[string]string{
		"ads.example":                 ActionBlock,
		"x.ads.example":               ActionBlock,
		"cdn.ads.example":             ActionAllow,
		"y.cdn.ads.example":           ActionBlock, // the wildcard above still applies
		"ok.wide.example":             ActionAllow,
		"a.ok.wide.example":           ActionAllow,
		"y.wide.example":              ActionBlock, // exists only because of a deeper rule
		"x.y.wide.example":            ActionBlock,
		"z.x.y.wide.example":          ActionBlock,
		"deep.x.y.wide.example":       ActionAllow,
		"below.deep.x.y.wide.example": ActionBlock,
		"solo.example":                ActionBlock,
		"other.example":               "",
		"sub.solo.example":            "",
		"example":                     "",
		"unrelated.org":               "",
	} {
		_, act, ok := MatchRules(e, name)
		if !ok {
			act = ""
		}
		if act != want {
			t.Errorf("%s: %q, want %q", name, act, want)
		}
	}
	z := RulesZoneFile("rules", rules)
	if !strings.Contains(z, "cdn.ads.example\tCNAME\trpz-passthru.\n*.cdn.ads.example\tCNAME\t.\n") {
		t.Errorf("rules zone:\n%s", z)
	}
	if strings.Index(z, "ads.example\t") > strings.Index(z, "*.ads.example\t") {
		t.Errorf("a name should come before its wildcard:\n%s", z)
	}
}

func TestMatchFeed(t *testing.T) {
	l, _ := Parse(strings.NewReader("||tracker.example^\nexact.example\n||deep.a.example^\n"), Limits{})
	Compact(l)
	path := filepath.Join(t.TempDir(), "list.rpz")
	f, _ := os.Create(path)
	FeedZone(f, "t", 1, l)
	f.Close()
	for name, want := range map[string]string{
		"tracker.example":     "tracker.example",
		"a.b.tracker.example": "*.tracker.example",
		"exact.example":       "exact.example",
		"sub.exact.example":   "",
		"x.deep.a.example":    "*.deep.a.example",
		"a.example":           "",
		"example":             "",
	} {
		owner, ok, err := MatchFeed(path, name)
		if err != nil {
			t.Fatal(err)
		}
		if !ok {
			owner = ""
		}
		if owner != want {
			t.Errorf("%s: %q, want %q", name, owner, want)
		}
	}
}

func TestNames(t *testing.T) {
	if ListZone("hagezi") != "hagezi.blocklist.jarl.invalid" || ListName("hagezi.blocklist.jarl.invalid.") != "hagezi" || ListName(RulesZone) != "" || ListName("x.y.blocklist.jarl.invalid") != "" {
		t.Error("zone names")
	}
	if n, err := NormalizeName("Ads.Example.COM."); err != nil || n != "ads.example.com" {
		t.Errorf("NormalizeName = %q, %v", n, err)
	}
	for _, bad := range []string{"com", "1.2.3.4", "a b.example", "*.example.com", ""} {
		if _, err := NormalizeName(bad); err == nil {
			t.Errorf("%q accepted", bad)
		}
	}
}