idreplacer.go

package config

import (
	"net/http"
	"regexp"
	"strings"
)

// idPatterns describes patterns of IDs in URLs. These patterns are matched
// against an entire URL. These patterns are used as given without any
// additional prefix or suffix, so to match against only one element of a path
// (e.g., only “bar” in “https://foo/bar/baz”), the patterns must include its
// own checks for path separators and the end of the URL.
var idPatterns = []struct {
	pattern     *regexp.Regexp
	placeholder []byte
}{
	// Strictly numeric IDs (e.g., “/123”).
	{
		regexp.MustCompile(`/\d+(/|$)`),
		[]byte("/id$1"),
	},
	// UUIDs (e.g., “00000000-0000-0000-0000-000000000000”).
	{
		regexp.MustCompile(`/\w{8}-\w{4}-\w{4}-\w{4}-\w{12}(/|$)`),
		[]byte("/uuid$1"),
	},
}

// IDReplacer attempts to generate the resource name for an HTTP request by
// replacing portions which look like IDs with generic placeholders. Because we
// don't have access to the actual route pattern, we'll synthesize it by
// replacing anything in the URL which looks like a numeric ID or a UUID with a
// placeholder token (“id” or “uuid”, respectively). E.g., a GET request for:
//
//     /api/foo/f9849b8d-455b-41fc-bec5-a45bc822023d/bar/42
//
// will be transformed into:
//
//     GET /api/foo/uuid/bar/id
//
// This may differ from the actual route definition; for example:
//
//     /api/foo/:fooID/bar/:barID
//
// ...but it's close enough for tracing.
func IDReplacer(req *http.Request) string {
	resource := req.URL.Path
	for _, idPattern := range idPatterns {
		resource = string(idPattern.pattern.ReplaceAll(
			[]byte(resource),
			idPattern.placeholder))
	}

	// Route normalization: remove any trailing slashes so traces for
	// extension-less routes are combined regardless of whether the request had
	// a trailing slash (e.g., “/foo vs. “/foo/”).
	if resource != "/" {
		resource = strings.TrimRight(resource, "/")
	}

	return req.Method + " " + resource
}