From 0845028b70a41e58b70db503000326c361bc1a47 Mon Sep 17 00:00:00 2001
From: Nick Otter <notter@kochava.com>
Date: Thu, 1 Oct 2020 09:13:10 -0700
Subject: [PATCH] Add FormatStatsPath to ochttp Transport to reduce stats path
 cardinality

Updates ochttp.Transport to accept an optional FormatStatsPath which is supplied
to ochttp.statsTransport. FormatStatsPath can override a request URL's path parameter
to reduce cardinality for stats similar to FormatSpanName.
---
 plugin/ochttp/client.go       | 12 ++++++++-
 plugin/ochttp/client_stats.go | 12 ++++++---
 plugin/ochttp/example_test.go | 50 +++++++++++++++++++++++++++++++++++
 3 files changed, 70 insertions(+), 4 deletions(-)

diff --git a/plugin/ochttp/client.go b/plugin/ochttp/client.go
index da815b2a7..241840d8a 100644
--- a/plugin/ochttp/client.go
+++ b/plugin/ochttp/client.go
@@ -56,6 +56,10 @@ type Transport struct {
 	// name equals the URL Path.
 	FormatSpanName func(*http.Request) string
 
+	// FormatStatsPath holds the function to use for standardizing the path
+	// supplied to metrics. By default the path equals the URL path.
+	FormatStatsPath func(*http.Request) string
+
 	// NewClientTrace may be set to a function allowing the current *trace.Span
 	// to be annotated with HTTP request event information emitted by the
 	// httptrace package.
@@ -95,7 +99,13 @@ func (t *Transport) RoundTrip(req *http.Request) (*http.Response, error) {
 		formatSpanName: spanNameFormatter,
 		newClientTrace: t.NewClientTrace,
 	}
-	rt = statsTransport{base: rt}
+
+	statsPathFormatter := t.FormatStatsPath
+	if statsPathFormatter == nil {
+		statsPathFormatter = statsPath
+	}
+
+	rt = statsTransport{base: rt, pathFormatter: statsPathFormatter}
 	return rt.RoundTrip(req)
 }
 
diff --git a/plugin/ochttp/client_stats.go b/plugin/ochttp/client_stats.go
index 17142aabe..f54e8787f 100644
--- a/plugin/ochttp/client_stats.go
+++ b/plugin/ochttp/client_stats.go
@@ -28,7 +28,8 @@ import (
 
 // statsTransport is an http.RoundTripper that collects stats for the outgoing requests.
 type statsTransport struct {
-	base http.RoundTripper
+	base          http.RoundTripper
+	pathFormatter func(*http.Request) string
 }
 
 // RoundTrip implements http.RoundTripper, delegating to Base and recording stats for the request.
@@ -36,8 +37,8 @@ func (t statsTransport) RoundTrip(req *http.Request) (*http.Response, error) {
 	ctx, _ := tag.New(req.Context(),
 		tag.Upsert(KeyClientHost, req.Host),
 		tag.Upsert(Host, req.Host),
-		tag.Upsert(KeyClientPath, req.URL.Path),
-		tag.Upsert(Path, req.URL.Path),
+		tag.Upsert(KeyClientPath, t.pathFormatter(req)),
+		tag.Upsert(Path, t.pathFormatter(req)),
 		tag.Upsert(KeyClientMethod, req.Method),
 		tag.Upsert(Method, req.Method))
 	req = req.WithContext(ctx)
@@ -141,3 +142,8 @@ func (t *tracker) Close() error {
 	t.end()
 	return t.body.Close()
 }
+
+// statsPath gets the path from a request url
+func statsPath(req *http.Request) string {
+	return req.URL.Path
+}
diff --git a/plugin/ochttp/example_test.go b/plugin/ochttp/example_test.go
index bb115abb9..3069bc2b8 100644
--- a/plugin/ochttp/example_test.go
+++ b/plugin/ochttp/example_test.go
@@ -17,6 +17,7 @@ package ochttp_test
 import (
 	"log"
 	"net/http"
+	"regexp"
 
 	"go.opencensus.io/plugin/ochttp"
 	"go.opencensus.io/plugin/ochttp/propagation/b3"
@@ -54,6 +55,55 @@ func ExampleTransport() {
 	_ = client
 }
 
+// FormatStatsPath allows paths to be standardized before the tag is assigned
+//
+// For example https://jsonplaceholder.typicode.com/posts/1 and https://jsonplaceholder.typicode.com/posts/2
+// can have the path  `/posts/:id` while https://jsonplaceholder.typicode.com/posts/1/comments has the path
+// `/posts/:id/comments`
+//
+// Using FormatStatsPath can reduce cardinality when using KeyClientPath
+func ExampleTransport_statsPathFormatter() {
+	// import (
+	// 		"go.opencensus.io/plugin/ochttp"
+	//		"go.opencensus.io/stats/view"
+	// )
+
+	// Add KeyClientPath and KeyClientHost tags to a few views.
+	ochttp.ClientRoundtripLatencyDistribution.TagKeys = append(ochttp.ClientRoundtripLatencyDistribution.TagKeys, ochttp.KeyClientPath, ochttp.KeyClientHost)
+	ochttp.ClientCompletedCount.TagKeys = append(ochttp.ClientCompletedCount.TagKeys, ochttp.KeyClientPath, ochttp.KeyClientHost)
+
+	if err := view.Register(
+		// Register a few default views.
+		ochttp.ClientCompletedCount,
+		ochttp.ClientRoundtripLatencyDistribution,
+	); err != nil {
+		log.Fatal(err)
+	}
+
+	postRegexp := regexp.MustCompile(`^\/?posts\/[0-9]+\/?$`)
+	postCommentsRegexp := regexp.MustCompile(`^\/?posts\/[0-9]+\/comments\/?$`)
+
+	client := &http.Client{
+		Transport: &ochttp.Transport{
+			FormatStatsPath: func(req *http.Request) string {
+				if req.URL.Host == "jsonplaceholder.typicode.com" {
+					if postRegexp.MatchString(req.URL.Path) { // format the post endpoint
+						return "/posts/:id"
+					} else if postCommentsRegexp.MatchString(req.URL.Path) { // format the post comment endpoint
+						return "/posts/:id/comments"
+					}
+				}
+
+				// use the URL path as the default path
+				return req.URL.Path
+			},
+		},
+	}
+
+	// Use client to perform requests.
+	_ = client
+}
+
 var usersHandler http.Handler
 
 func ExampleHandler() {