fix: correctly handle data URL with hashes. (#2475)

* fix: correctly handle data URL with hashes.

* fix: lint

* test: better name

* suggestion change

* perf: avoid substring

* fixup

* test: better

Author: tsctx

lib/fetch/dataURL.js

@@ -119,17 +119,14 @@ function dataURLProcessor (dataURL) {
  * @param {boolean} excludeFragment
  */
 function URLSerializer (url, excludeFragment = false) {
-  const href = url.href
-
   if (!excludeFragment) {
-    return href
+    return url.href
   }
 
-  const hash = href.lastIndexOf('#')
-  if (hash === -1) {
-    return href
-  }
-  return href.slice(0, hash)
+  const href = url.href
+  const hashLength = url.hash.length
+
+  return hashLength === 0 ? href : href.substring(0, href.length - hashLength)
 }
 
 // https://infra.spec.whatwg.org/#collect-a-sequence-of-code-points

test/fetch/data-uri.js

@@ -191,3 +191,24 @@ test('https://domain.com/?', (t) => {
   const serialized = URLSerializer(new URL(domain))
   t.equal(serialized, domain)
 })
+
+// https://github.com/nodejs/undici/issues/2474
+test('hash url', (t) => {
+  t.plan(1)
+  const domain = 'https://domain.com/#a#b'
+  const url = new URL(domain)
+  const serialized = URLSerializer(url, true)
+  t.equal(serialized, url.href.substring(0, url.href.length - url.hash.length))
+})
+
+// https://github.com/nodejs/undici/issues/2474
+test('data url that includes the hash', async (t) => {
+  t.plan(1)
+  const dataURL = 'data:,node#js#'
+  try {
+    const res = await fetch(dataURL)
+    t.equal(await res.text(), 'node')
+  } catch (error) {
+    t.fail(`failed to fetch ${dataURL}`)
+  }
+})