feat: Allow URL encoded dataset in libhoney endpoint paths

route/route.go

@@ -168,6 +168,7 @@ func (r *Router) LnS(incomingOrPeer string) {
 
 	// require an auth header for events and batches
 	authedMuxxer := muxxer.PathPrefix("/1/").Methods("POST").Subrouter()
+	authedMuxxer.UseEncodedPath()
 	authedMuxxer.Use(r.apiKeyChecker)
 
 	// handle events and batches
@@ -388,8 +389,10 @@ func (r *Router) requestToEvent(req *http.Request, reqBod []byte) (*types.Event,
 		sampleRate = 1
 	}
 	eventTime := getEventTime(req.Header.Get(types.TimestampHeader))
-	vars := mux.Vars(req)
-	dataset := vars["datasetName"]
+	dataset, err := getDatasetFromRequest(req)
+	if err != nil {
+		return nil, err
+	}
 
 	apiHost, err := r.Config.GetHoneycombAPI()
 	if err != nil {
@@ -678,8 +681,10 @@ func (r *Router) batchedEventToEvent(req *http.Request, bev batchedEvent, apiKey
 	eventTime := bev.getEventTime()
 	// TODO move the following 3 lines outside of this loop; they could be done
 	// once for the entire batch instead of in every event.
-	vars := mux.Vars(req)
-	dataset := vars["datasetName"]
+	dataset, err := getDatasetFromRequest(req)
+	if err != nil {
+		return nil, err
+	}
 	apiHost, err := r.Config.GetHoneycombAPI()
 	if err != nil {
 		return nil, err
@@ -969,3 +974,15 @@ func (r *Router) AddOTLPMuxxer(muxxer *mux.Router) {
 	otlpMuxxer.HandleFunc("/logs", r.postOTLPLogs).Name("otlp_logs")
 	otlpMuxxer.HandleFunc("/logs/", r.postOTLPLogs).Name("otlp_logs")
 }
+
+func getDatasetFromRequest(req *http.Request) (string, error) {
+	dataset := mux.Vars(req)["datasetName"]
+	if dataset == "" {
+		return "", fmt.Errorf("missing dataset name")
+	}
+	dataset, err := url.PathUnescape(dataset)
+	if err != nil {
+		return "", err
+	}
+	return dataset, nil
+}

route/route_test.go

@@ -9,6 +9,7 @@ import (
 	"io"
 	"net/http"
 	"net/http/httptest"
+	"net/url"
 	"strings"
 	"testing"
 	"time"
@@ -612,3 +613,109 @@ func TestGRPCHealthProbeWatch(t *testing.T) {
 	sentMessage := mockServer.GetSentMessages()[0]
 	assert.Equal(t, grpc_health_v1.HealthCheckResponse_SERVING, sentMessage.Status)
 }
+
+func TestGetDatasetFromRequest(t *testing.T) {
+	testCases := []struct {
+		name                string
+		datasetName         string
+		expectedDatasetName string
+		expectedError       error
+	}{
+		{
+			name:          "empty dataset name",
+			datasetName:   "",
+			expectedError: fmt.Errorf("missing dataset name"),
+		},
+		{
+			name:          "dataset name with invalid URL encoding",
+			datasetName:   "foo%2",
+			expectedError: url.EscapeError("%2"),
+		},
+		{
+			name:                "normal dataset name",
+			datasetName:         "foo",
+			expectedDatasetName: "foo",
+		},
+		{
+			name:                "dataset name with numbers",
+			datasetName:         "foo123",
+			expectedDatasetName: "foo123",
+		},
+		{
+			name:                "dataset name with hyphen",
+			datasetName:         "foo-bar",
+			expectedDatasetName: "foo-bar",
+		},
+		{
+			name:                "dataset name with underscore",
+			datasetName:         "foo_bar",
+			expectedDatasetName: "foo_bar",
+		},
+		{
+			name:                "dataset name with tilde",
+			datasetName:         "foo~bar",
+			expectedDatasetName: "foo~bar",
+		},
+		{
+			name:                "dataset name with period",
+			datasetName:         "foo.bar",
+			expectedDatasetName: "foo.bar",
+		},
+		{
+			name:                "dataset name with URL encoded hyphen",
+			datasetName:         "foo%2Dbar",
+			expectedDatasetName: "foo-bar",
+		},
+		{
+			name:                "dataset name with URL encoded underscore",
+			datasetName:         "foo%5Fbar",
+			expectedDatasetName: "foo_bar",
+		},
+		{
+			name:                "dataset name with URL encoded tilde",
+			datasetName:         "foo%7Ebar",
+			expectedDatasetName: "foo~bar",
+		},
+		{
+			name:                "dataset name with URL encoded period",
+			datasetName:         "foo%2Ebar",
+			expectedDatasetName: "foo.bar",
+		},
+		{
+			name:                "dataset name with URL encoded forward slash",
+			datasetName:         "foo%2Fbar",
+			expectedDatasetName: "foo/bar",
+		},
+		{
+			name:                "dataset name with URL encoded colon",
+			datasetName:         "foo%3Abar",
+			expectedDatasetName: "foo:bar",
+		},
+		{
+			name:                "dataset name with URL encoded square brackets",
+			datasetName:         "foo%5Bbar%5D",
+			expectedDatasetName: "foo[bar]",
+		},
+		{
+			name:                "dataset name with URL encoded parentheses",
+			datasetName:         "foo%28bar%29",
+			expectedDatasetName: "foo(bar)",
+		},
+		{
+			name:                "dataset name with URL encoded curly braces",
+			datasetName:         "foo%7Bbar%7D",
+			expectedDatasetName: "foo{bar}",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			req, _ := http.NewRequest("GET", "/1/events/dataset", nil)
+			req = mux.SetURLVars(req, map[string]string{"datasetName": tc.datasetName})
+
+			dataset, err := getDatasetFromRequest(req)
+			assert.Equal(t, tc.expectedError, err)
+			assert.Equal(t, tc.expectedDatasetName, dataset)
+		})
+	}
+}