Improve gibberish detection (#822)

Author: grcevski

pkg/internal/transform/route/clusterurl/cluster.go

@@ -162,8 +162,26 @@ func (csf *ClusterURLClassifier) okWord(w string) bool {
 	if ok {
 		return ok
 	}
-	if gibberish.IsGibberish(w, csf.classifier) {
-		return false
+
+	start := 0
+	for i, c := range w {
+		if c == '-' || c == '_' || c == '.' {
+			if i == start {
+				return false
+			}
+
+			if gibberish.IsGibberish(w[start:i], csf.classifier) {
+				return false
+			}
+
+			start = i + 1
+		}
+	}
+
+	if start < len(w) {
+		if gibberish.IsGibberish(w[start:], csf.classifier) {
+			return false
+		}
 	}
 
 	csf.cache.Add(w, true)

pkg/internal/transform/route/clusterurl/cluster_test.go

@@ -13,7 +13,9 @@ import (
 func TestClusterURL(t *testing.T) {
 	csf, err := NewClusterURLClassifier(DefaultConfig())
 	assert.NoError(t, err)
+	assert.Equal(t, "*", csf.ClusterURL("registry-apjkmyp"))
 	assert.Empty(t, csf.ClusterURL(""))
+	assert.Equal(t, "*", csf.ClusterURL("apjkmyp"))
 	assert.Equal(t, "/users/*/j4elk/*/job/*", csf.ClusterURL("/users/fdklsd/j4elk/23993/job/2"))
 	assert.Equal(t, "*", csf.ClusterURL("123"))
 	assert.Equal(t, "/*", csf.ClusterURL("/123"))
@@ -51,6 +53,10 @@ func TestClusterURL(t *testing.T) {
 	assert.Equal(t, "HTTP GET", csf.ClusterURL("HTTP GET"))
 	assert.Equal(t, "GET /api/cart", csf.ClusterURL("GET /api/cart?sessionId=55f4e5ea-5d6d-482a-80c4-799e3c72dfb0&currencyCode=USD"))
 	assert.Equal(t, "/getquote", csf.ClusterURL("/getquote"))
+	assert.Equal(t, "PUT /bar/test/test/*/files/*/test/*", csf.ClusterURL("PUT /bar/test/test/bar-attach-generic-product-apjkmyp/files/multi-test-version-jwbCm/test/some-file.txt"))
+	assert.Equal(t, "PUT /test/bar/test/test/*/files/*/test/*", csf.ClusterURL("PUT /test/bar/test/test/bar-attach-generic-registry-apjkmyp/files/push-metrics-test-OYboK/test/README.md"))
+	assert.Equal(t, "PUT /test/bar/test_plus/test.now/*/files/*/test/*", csf.ClusterURL("PUT /test/bar/test_plus/test.now/bar-attach-generic-registry-apjkmyp/files/push-metrics-test-OYboK/test/README.md"))
+	assert.Equal(t, "PUT /bar/test/test/*/files/*/test/*", csf.ClusterURL("PUT /bar/test/test/a----/files/-a-a-a--/test/some-file.txt"))
 	assert.Equal(t, "", csf.ClusterURL("?"))
 	assert.Equal(t, "*", csf.ClusterURL("attach12?"))
 	assert.Equal(t, "*", csf.ClusterURL("1?"))

test/integration/red_test_python_aws_s3.go

@@ -16,6 +16,7 @@ import (
 	"github.com/mariomac/guara/pkg/test"
 	"github.com/stretchr/testify/require"
 
+	ti "go.opentelemetry.io/obi/pkg/test/integration"
 	"go.opentelemetry.io/obi/test/integration/components/jaeger"
 )
 
@@ -35,6 +36,21 @@ func testPythonAWSS3(t *testing.T) {
 	waitForTestComponentsNoMetrics(t, address+"/health")
 	waitForTestComponentsNoMetrics(t, localstackAddress)
 
+	// Wait for /health to appear in jaeger
+	test.Eventually(t, testTimeout, func(t require.TestingT) {
+		ti.DoHTTPGet(t, "http://localhost:8381/health", 200)
+		resp, err := http.Get(jaegerQueryURL + "?service=python3.12&operation=GET%20%2Fhealth")
+		require.NoError(t, err)
+		if resp == nil {
+			return
+		}
+		require.Equal(t, http.StatusOK, resp.StatusCode)
+		var tq jaeger.TracesQuery
+		require.NoError(t, json.NewDecoder(resp.Body).Decode(&tq))
+		traces := tq.FindBySpan(jaeger.Tag{Key: "url.path", Type: "string", Value: "/health"})
+		require.Len(t, traces, 1)
+	}, test.Interval(1*time.Second))
+
 	s3Req(t, address+"/createbucket")
 	s3Req(t, address+"/createobject")
 	s3Req(t, address+"/listobjects")