filters: apply replace rules only after obfuscation

cmd/trace-agent/agent.go

@@ -19,9 +19,7 @@ import (
 	"github.com/DataDog/datadog-trace-agent/writer"
 )
 
-const (
-	processStatsInterval = time.Minute
-)
+const processStatsInterval = time.Minute
 
 type processedTrace struct {
 	Trace         model.Trace
@@ -50,7 +48,8 @@ func (pt *processedTrace) getSamplingPriority() (int, bool) {
 type Agent struct {
 	Receiver           *HTTPReceiver
 	Concentrator       *Concentrator
-	Filters            []filters.Filter
+	Blacklister        *filters.Blacklister
+	Replacer           *filters.Replacer
 	ScoreSampler       *Sampler
 	ErrorsScoreSampler *Sampler
 	PrioritySampler    *Sampler
@@ -94,7 +93,6 @@ func NewAgent(ctx context.Context, conf *config.AgentConfig) *Agent {
 		conf.BucketInterval.Nanoseconds(),
 		statsChan,
 	)
-	f := filters.Setup(conf)
 
 	obf := obfuscate.NewObfuscator(conf.Obfuscation)
 	ss := NewScoreSampler(conf)
@@ -110,7 +108,8 @@ func NewAgent(ctx context.Context, conf *config.AgentConfig) *Agent {
 	return &Agent{
 		Receiver:           r,
 		Concentrator:       c,
-		Filters:            f,
+		Blacklister:        filters.NewBlacklister(conf.Ignore["resource"]),
+		Replacer:           filters.NewReplacer(conf.ReplaceTags),
 		ScoreSampler:       ss,
 		ErrorsScoreSampler: ess,
 		PrioritySampler:    ps,
@@ -179,8 +178,6 @@ func (a *Agent) Run() {
 // passes it downstream.
 func (a *Agent) Process(t model.Trace) {
 	if len(t) == 0 {
-		// XXX Should never happen since we reject empty traces during
-		// normalization.
 		log.Debugf("skipping received empty trace")
 		return
 	}
@@ -230,15 +227,10 @@ func (a *Agent) Process(t model.Trace) {
 		return
 	}
 
-	for _, f := range a.Filters {
-		if f.Keep(root, &t) {
-			continue
-		}
-
-		log.Debugf("rejecting trace by filter: %T  %v", f, *root)
+	if !a.Blacklister.Allows(root) {
+		log.Debugf("trace rejected by blacklister. root: %v", root)
 		atomic.AddInt64(&ts.TracesFiltered, 1)
 		atomic.AddInt64(&ts.SpansFiltered, int64(len(t)))
-
 		return
 	}
 
@@ -264,6 +256,8 @@ func (a *Agent) Process(t model.Trace) {
 		span.Truncate()
 	}
 
+	a.Replacer.Replace(&t)
+
 	pt := processedTrace{
 		Trace:         t,
 		WeightedTrace: wt,

filters/blacklister.go

@@ -0,0 +1,45 @@
+package filters
+
+import (
+	"regexp"
+
+	"github.com/DataDog/datadog-trace-agent/model"
+
+	log "github.com/cihub/seelog"
+)
+
+// Blacklister holds a list of regular expressions which will match resources
+// on spans that should be dropped.
+type Blacklister struct {
+	list []*regexp.Regexp
+}
+
+// Allows returns true if the Blacklister permits this span.
+func (f *Blacklister) Allows(span *model.Span) bool {
+	for _, entry := range f.list {
+		if entry.MatchString(span.Resource) {
+			return false
+		}
+	}
+	return true
+}
+
+// NewBlacklister creates a new Blacklister based on the given list of
+// regular expressions.
+func NewBlacklister(exprs []string) *Blacklister {
+	return &Blacklister{list: compileRules(exprs)}
+}
+
+// compileRules compiles as many rules as possible from the list of expressions.
+func compileRules(exprs []string) []*regexp.Regexp {
+	list := make([]*regexp.Regexp, 0, len(exprs))
+	for _, entry := range exprs {
+		rule, err := regexp.Compile(entry)
+		if err != nil {
+			log.Errorf("invalid resource filter: %q", entry)
+			continue
+		}
+		list = append(list, rule)
+	}
+	return list
+}

filters/blacklister_test.go

@@ -0,0 +1,47 @@
+package filters
+
+import (
+	"testing"
+
+	"github.com/DataDog/datadog-trace-agent/fixtures"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestBlacklister(t *testing.T) {
+	tests := []struct {
+		filter      []string
+		resource    string
+		expectation bool
+	}{
+		{[]string{"/foo/bar"}, "/foo/bar", false},
+		{[]string{"/foo/b.r"}, "/foo/bar", false},
+		{[]string{"[0-9]+"}, "/abcde", true},
+		{[]string{"[0-9]+"}, "/abcde123", false},
+		{[]string{"\\(foobar\\)"}, "(foobar)", false},
+		{[]string{"\\(foobar\\)"}, "(bar)", true},
+		{[]string{"(GET|POST) /healthcheck"}, "GET /foobar", true},
+		{[]string{"(GET|POST) /healthcheck"}, "GET /healthcheck", false},
+		{[]string{"(GET|POST) /healthcheck"}, "POST /healthcheck", false},
+		{[]string{"SELECT COUNT\\(\\*\\) FROM BAR"}, "SELECT COUNT(*) FROM BAR", false},
+		{[]string{"[123"}, "[123", true},
+		{[]string{"\\[123"}, "[123", false},
+		{[]string{"ABC+", "W+"}, "ABCCCC", false},
+		{[]string{"ABC+", "W+"}, "WWW", false},
+	}
+
+	for _, test := range tests {
+		span := fixtures.RandomSpan()
+		span.Resource = test.resource
+		filter := NewBlacklister(test.filter)
+
+		assert.Equal(t, test.expectation, filter.Allows(span))
+	}
+}
+
+func TestCompileRules(t *testing.T) {
+	filter := NewBlacklister([]string{"[123", "]123", "{6}"})
+	for i := 0; i < 100; i++ {
+		span := fixtures.RandomSpan()
+		assert.True(t, filter.Allows(span))
+	}
+}

filters/tag.go → filters/replacer.go

@@ -5,22 +5,20 @@ import (
 	"github.com/DataDog/datadog-trace-agent/model"
 )
 
-var _ Filter = (*tagReplacer)(nil)
-
-// tagReplacer is a filter which replaces tag values based on its
+// Replacer is a filter which replaces tag values based on its
 // settings. It keeps all spans.
-type tagReplacer struct {
-	// replace maps tag keys to one or more sets of replacements
-	replace []*config.ReplaceRule
+type Replacer struct {
+	rules []*config.ReplaceRule
 }
 
-func newTagReplacer(c *config.AgentConfig) *tagReplacer {
-	return &tagReplacer{replace: c.ReplaceTags}
+// NewReplacer returns a new Replacer which will use the given set of rules.
+func NewReplacer(rules []*config.ReplaceRule) *Replacer {
+	return &Replacer{rules: rules}
 }
 
-// Keep implements Filter.
-func (f tagReplacer) Keep(root *model.Span, trace *model.Trace) bool {
-	for _, rule := range f.replace {
+// Replace replaces all tags matching the Replacer's rules.
+func (f Replacer) Replace(trace *model.Trace) {
+	for _, rule := range f.rules {
 		key, str, re := rule.Name, rule.Repl, rule.Re
 		for _, s := range *trace {
 			switch key {
@@ -42,6 +40,4 @@ func (f tagReplacer) Keep(root *model.Span, trace *model.Trace) bool {
 			}
 		}
 	}
-	// always return true as the goal of this filter is only to mutate data
-	return true
 }

filters/tag_test.go → filters/replacer_test.go

@@ -9,7 +9,7 @@ import (
 	"github.com/stretchr/testify/assert"
 )
 
-func TestTagReplacer(t *testing.T) {
+func TestReplacer(t *testing.T) {
 	assert := assert.New(t)
 	for _, tt := range []struct {
 		rules     [][3]string
@@ -54,12 +54,11 @@ func TestTagReplacer(t *testing.T) {
 		},
 	} {
 		rules := parseRulesFromString(tt.rules)
-		tr := &tagReplacer{replace: rules}
+		tr := NewReplacer(rules)
 		root := replaceFilterTestSpan(tt.got)
 		childSpan := replaceFilterTestSpan(tt.got)
 		trace := model.Trace{root, childSpan}
-		ok := tr.Keep(root, &trace)
-		assert.True(ok)
+		tr.Replace(&trace)
 		for k, v := range tt.want {
 			switch k {
 			case "resource.name":