break(url): remove toDecode parameter (#175)

* break(url): remove `toDecode` param * chore(url): update bench results
lukeed · Aug 27, 2021 · e45fe88 · e45fe88
1 parent 6ac6498
commit e45fe88
Show file tree

Hide file tree

Showing 6 changed files with 77 additions and 291 deletions.
diff --git a/packages/url/bench/index.js b/packages/url/bench/index.js
@@ -17,7 +17,7 @@ const isVerbose = process.argv.includes('--verbose');
 
 /**
  * @typedef Contender
- * @type {(req: Request, toDecode: boolean) => any}
+ * @type {(req: Request) => any}
  */
 
 /**
@@ -26,22 +26,20 @@ const isVerbose = process.argv.includes('--verbose');
  * @type {Record<string, Contender>}
  */
 const contenders = {
-	'url.parse': (r, d) => {
+	'url.parse': r => {
 		let out = native.parse(r.url, true); // query as object
-		if (d) out.pathname = decodeURIComponent(out.pathname);
 		// returns `null` if no value
 		out.search = out.search || '';
 		return out;
 	},
 
-	'new URL()': (r, d) => {
+	'new URL()': r => {
 		let url = r.url;
 		let { pathname, search, searchParams } = new URL(url, 'http://x.com');
-		if (d) pathname = decodeURIComponent(pathname);
 		return { url, pathname, search, query: searchParams };
 	},
 
-	'parseurl': (r, d) => {
+	'parseurl': r => {
 		/** @type {Record<string, string>} */ // @ts-ignore
 		let out = parseurl(r);
 
@@ -51,35 +49,31 @@ const contenders = {
 		// @ts-ignore - always returns `query` as string|null
 		if (out.query) out.query = qs.parse(out.query);
 
-		// never decodes, do bare minimum for compat
-		if (d) out.pathname = decodeURIComponent(out.pathname);
-
 		return out;
 	},
 
-	'@polka/url': (r, d) => {
+	'@polka/url': r => {
 		// @ts-ignore - request
-		return polka(r, d);
+		return polka(r);
 	},
 };
 
 /**
  * @param {object} config
  * @param {string} config.url
- * @param {boolean} config.decode
  * @param {Record<string, unknown>} config.expect
  * @param {boolean} [config.repeat]
  */
 function runner(config) {
-	let { url, expect, repeat, decode=false } = config;
-	let title = repeat ? 'repeat' : decode ? 'decode' : 'normal';
+	let { url, expect, repeat } = config;
+	let title = repeat ? 'repeat' : 'normal';
 
 	console.log('\nValidation: (%s) "%s"', title, url);
 	Object.keys(contenders).forEach(name => {
 		let key, fn=contenders[name];
 
 		try {
-			let output = fn({ url }, decode);
+			let output = fn({ url });
 
 			for (key in expect) {
 				let tmp = output[key];
@@ -107,7 +101,7 @@ function runner(config) {
 		let req = repeat && { url };
 
 		bench.add(name + ' '.repeat(16 - name.length), () => {
-			fn(req || { url }, decode);
+			fn(req || { url });
 		}, {
 			minSamples: 100
 		});
@@ -120,7 +114,6 @@ function runner(config) {
 
 runner({
 	url: '/foo/bar?user=tj&pet=fluffy',
-	decode: false,
 	expect: {
 		pathname: '/foo/bar',
 		search: '?user=tj&pet=fluffy',
@@ -134,7 +127,6 @@ runner({
 runner({
 	repeat: true,
 	url: '/foo/bar?user=tj&pet=fluffy',
-	decode: false,
 	expect: {
 		pathname: '/foo/bar',
 		search: '?user=tj&pet=fluffy',
@@ -147,7 +139,6 @@ runner({
 
 runner({
 	url: '/foo/bar',
-	decode: false,
 	expect: {
 		pathname: '/foo/bar',
 		search: '',
@@ -156,62 +147,8 @@ runner({
 
 runner({
 	url: '/',
-	decode: false,
 	expect: {
 		pathname: '/',
 		search: '',
 	}
 });
-
-// DECODES
-
-runner({
-	url: '/f%C3%B8%C3%B8%C3%9F%E2%88%82r',
-	decode: true,
-	expect: {
-		pathname: '/føøß∂r',
-		search: '',
-	}
-});
-
-runner({
-	url: '/f%C3%B8%C3%B8%C3%9F%E2%88%82r?phone=%2b393383123549',
-	decode: true,
-	expect: {
-		pathname: '/føøß∂r',
-		search: '?phone=%2b393383123549',
-		query: { phone: '+393383123549' },
-	}
-});
-
-runner({
-	repeat: true,
-	url: '/f%C3%B8%C3%B8%C3%9F%E2%88%82r?phone=%2b393383123549',
-	decode: true,
-	expect: {
-		pathname: '/føøß∂r',
-		search: '?phone=%2b393383123549',
-		query: { phone: '+393383123549' },
-	}
-});
-
-runner({
-	url: '/foo/bar?hello=123',
-	decode: true,
-	expect: {
-		pathname: '/foo/bar',
-		search: '?hello=123',
-		query: {
-			hello: '123',
-		}
-	}
-});
-
-runner({
-	url: '/foo/bar',
-	decode: true,
-	expect: {
-		pathname: '/foo/bar',
-		search: '',
-	}
-});
diff --git a/packages/url/bench/readme.md b/packages/url/bench/readme.md
@@ -1,124 +1,87 @@
 ## Benchmarks
 
-> Running on Node v14.15.3
+> Running on Node v16.8.0
 
 ***Modifications:***
 
-Each candidate is _slightly_ modified to match the `pathname` and `query` values that `@polka/url` returns. This is done in an honest/best-effort manner to normalize all candidates (and pass the validation steps), especially considering that the typical user _wants_ `pathname`s to be normalized and _wants_ `query` values to be parsed (and decoded) into an object.
+Each candidate is _slightly_ modified to match the `search` and `query` values that `@polka/url` returns. This is done in an honest/best-effort manner to normalize all candidates (and pass the validation steps), especially considering that the typical user wants `query` to be parsed (and decoded) into an object.
 
-Please see the [Raw Performance](#raw-performance) benchmarks for results ***without*** any modifications. However, do note that the benchmark is effectively useless since all candidates do different things by default.
+Please see the [Raw Performance](#raw-performance) benchmarks for results ***without*** any modifications. However, do note that the raw-benchmark is effectively useless since all candidates do different things by default.
 
 
-### Without Decoding
+## Normalized (minimal)
 
 > **Important:** All candidates listed pass validation – sometimes due to normalization.
 
 ```
 Benchmark: "/foo/bar?user=tj&pet=fluffy"
-  url.parse        x   1,437,988 ops/sec ±0.15% (193 runs sampled)
-  new URL()        x     258,158 ops/sec ±0.17% (193 runs sampled)
-  parseurl         x   2,074,250 ops/sec ±0.22% (193 runs sampled)
-  @polka/url       x   2,584,006 ops/sec ±0.26% (194 runs sampled)
+  url.parse        x   1,766,192 ops/sec ±0.59% (189 runs sampled)
+  new URL()        x     254,742 ops/sec ±0.93% (188 runs sampled)
+  parseurl         x   2,296,634 ops/sec ±0.43% (188 runs sampled)
+  @polka/url       x   3,096,770 ops/sec ±0.56% (189 runs sampled)
 
 Benchmark: (REPEAT) "/foo/bar?user=tj&pet=fluffy"
-  url.parse        x   1,457,258 ops/sec ±0.17% (193 runs sampled)
-  new URL()        x     257,459 ops/sec ±0.24% (189 runs sampled)
-  parseurl         x  26,687,772 ops/sec ±0.40% (192 runs sampled)
-  @polka/url       x 117,737,558 ops/sec ±0.23% (193 runs sampled)
+  url.parse        x   1,794,821 ops/sec ±1.01% (188 runs sampled)
+  new URL()        x     258,587 ops/sec ±0.63% (189 runs sampled)
+  parseurl         x  30,616,846 ops/sec ±0.47% (191 runs sampled)
+  @polka/url       x 314,370,079 ops/sec ±0.36% (189 runs sampled)
 
 Benchmark: "/foo/bar"
-  url.parse        x   5,928,214 ops/sec ±0.33% (191 runs sampled)
-  new URL()        x     290,799 ops/sec ±0.17% (192 runs sampled)
-  parseurl         x  17,597,099 ops/sec ±0.62% (189 runs sampled)
-  @polka/url       x  34,097,146 ops/sec ±0.55% (192 runs sampled)
+  url.parse        x   7,373,531 ops/sec ±0.63% (188 runs sampled)
+  new URL()        x     291,642 ops/sec ±0.83% (189 runs sampled)
+  parseurl         x  21,946,341 ops/sec ±0.93% (186 runs sampled)
+  @polka/url       x  48,697,030 ops/sec ±0.49% (189 runs sampled)
 
 Benchmark: "/"
-  url.parse        x   8,933,877 ops/sec ±0.70% (190 runs sampled)
-  new URL()        x     333,268 ops/sec ±0.19% (193 runs sampled)
-  parseurl         x  27,358,354 ops/sec ±0.76% (188 runs sampled)
-  @polka/url       x  43,529,456 ops/sec ±1.99% (172 runs sampled)
-```
-
-
-### With Decoding
-
-> **Important:** All candidates listed pass validation – sometimes due to normalization.
-
-```
-Benchmark: "/f%C3%B8%C3%B8%C3%9F%E2%88%82r"
-  url.parse        x   1,038,724 ops/sec ±0.12% (193 runs sampled)
-  new URL()        x     229,125 ops/sec ±0.17% (192 runs sampled)
-  parseurl         x   1,370,300 ops/sec ±0.20% (192 runs sampled)
-  @polka/url       x   1,540,894 ops/sec ±0.18% (192 runs sampled)
-
-Benchmark: "/f%C3%B8%C3%B8%C3%9F%E2%88%82r?phone=%2b393383123549"
-  url.parse        x     514,280 ops/sec ±0.40% (193 runs sampled)
-  new URL()        x     187,672 ops/sec ±0.67% (192 runs sampled)
-  parseurl         x     618,801 ops/sec ±0.10% (189 runs sampled)
-  @polka/url       x     696,125 ops/sec ±0.14% (191 runs sampled)
-
-Benchmark: (REPEAT) "/f%C3%B8%C3%B8%C3%9F%E2%88%82r?phone=%2b393383123549"
-  url.parse        x     425,909 ops/sec ±0.29% (193 runs sampled)
-  new URL()        x     187,735 ops/sec ±0.14% (194 runs sampled)
-  parseurl         x   1,770,147 ops/sec ±0.15% (193 runs sampled)
-  @polka/url       x 197,963,726 ops/sec ±0.18% (194 runs sampled)
-
-Benchmark: "/foo/bar?hello=123"
-  url.parse        x   1,133,709 ops/sec ±0.28% (190 runs sampled)
-  new URL()        x     236,384 ops/sec ±0.20% (193 runs sampled)
-  parseurl         x   1,431,879 ops/sec ±0.19% (191 runs sampled)
-  @polka/url       x   3,883,489 ops/sec ±0.37% (192 runs sampled)
-
-Benchmark: "/foo/bar"
-  url.parse        x   1,824,376 ops/sec ±0.54% (192 runs sampled)
-  new URL()        x     252,204 ops/sec ±0.18% (192 runs sampled)
-  parseurl         x   2,329,132 ops/sec ±0.26% (193 runs sampled)
-  @polka/url       x  19,972,200 ops/sec ±0.64% (188 runs sampled)
+  url.parse        x  10,744,706 ops/sec ±0.47% (188 runs sampled)
+  new URL()        x     315,725 ops/sec ±0.86% (184 runs sampled)
+  parseurl         x  46,863,886 ops/sec ±1.06% (189 runs sampled)
+  @polka/url       x  72,862,914 ops/sec ±0.54% (190 runs sampled)
 ```
 
 
 ## Raw Performance
 
 These are the results of the _unmodified_ candidates. In other words, there is **zero consistency** in the candidates outputs. For example:
 
-* `url.parse#1` uses [`url.parse`](https://nodejs.org/api/url.html#url_url_parse_urlstring_parsequerystring_slashesdenotehost) with `parseQueryString` enabled<br>_It decodes the `query` correctly, but all other segments are still encoded segments._
+* `url.parse#1` uses [`url.parse`](https://nodejs.org/api/url.html#url_url_parse_urlstring_parsequerystring_slashesdenotehost) with `parseQueryString` enabled<br>_It converts the `query` into a decoded object, but everything else remains encoded._
 
 * `url.parse#2` is the same as `url.parse#1`, except `parseQueryString` is disabled.<br>_It leaves the `query` as a string & does no decoding whatsoever._
 
-* `new URL()` does what it describes :)<br>_The `pathname` is never decoded, but `searchParams` is always an decoded `URLSearchParams` instance._
+* `new URL()` does what it describes :)<br>_Everything remains encoded, except for `searchParams`, which is always an `URLSearchParams` instance with decoded values._
 
 * `parseurl` never decodes any value segments and `query` is always a string.
 
-* `@polka/url` only decodes the `pathname` when asked and `query` is always a decoded object.
+* `@polka/url` never decodes any value segments except `query`, which is always a decoded object.
 
 ***Results***
 
 ```
 Benchmark: (normal) "/foo/bar?user=tj&pet=fluffy"
-  url.parse#1      x   1,462,942 ops/sec ±0.16% (193 runs sampled)
-  url.parse#2      x   3,059,883 ops/sec ±0.19% (193 runs sampled)
-  new URL()        x     273,878 ops/sec ±0.11% (194 runs sampled)
-  parseurl         x   7,742,605 ops/sec ±0.25% (192 runs sampled)
-  @polka/url       x   2,611,970 ops/sec ±0.18% (190 runs sampled)
+  url.parse#1      x   1,757,099 ops/sec ±0.67% (188 runs sampled)
+  url.parse#2      x   4,487,853 ops/sec ±0.70% (185 runs sampled)
+  new URL()        x     284,153 ops/sec ±0.68% (187 runs sampled)
+  parseurl         x   9,848,571 ops/sec ±0.97% (186 runs sampled)
+  @polka/url       x   3,040,460 ops/sec ±0.79% (188 runs sampled)
 
 Benchmark: (repeat) "/foo/bar?user=tj&pet=fluffy"
-  url.parse#1      x   1,477,740 ops/sec ±0.28% (194 runs sampled)
-  url.parse#2      x   3,121,952 ops/sec ±0.12% (193 runs sampled)
-  new URL()        x     271,608 ops/sec ±0.24% (192 runs sampled)
-  parseurl         x 112,501,081 ops/sec ±3.88% (177 runs sampled)
-  @polka/url       x  73,331,596 ops/sec ±2.44% (180 runs sampled)
+  url.parse#1      x   1,827,115 ops/sec ±0.41% (190 runs sampled)
+  url.parse#2      x   4,442,871 ops/sec ±0.72% (183 runs sampled)
+  new URL()        x     286,803 ops/sec ±0.30% (189 runs sampled)
+  parseurl         x  78,897,892 ops/sec ±1.51% (182 runs sampled)
+  @polka/url       x 291,908,732 ops/sec ±4.97% (179 runs sampled)
 
 Benchmark: (normal) "/foo/bar"
-  url.parse#1      x   6,109,524 ops/sec ±0.39% (190 runs sampled)
-  url.parse#2      x   6,741,743 ops/sec ±0.38% (190 runs sampled)
-  new URL()        x     304,240 ops/sec ±0.15% (192 runs sampled)
-  parseurl         x  17,809,555 ops/sec ±0.55% (190 runs sampled)
-  @polka/url       x  28,314,612 ops/sec ±1.14% (175 runs sampled)
+  url.parse#1      x   7,824,747 ops/sec ±0.65% (186 runs sampled)
+  url.parse#2      x   9,015,704 ops/sec ±0.68% (188 runs sampled)
+  new URL()        x     320,978 ops/sec ±0.41% (188 runs sampled)
+  parseurl         x  25,611,676 ops/sec ±0.45% (189 runs sampled)
+  @polka/url       x  48,554,610 ops/sec ±0.48% (190 runs sampled)
 
 Benchmark: (normal) "/"
-  url.parse#1      x   9,370,629 ops/sec ±0.37% (187 runs sampled)
-  url.parse#2      x  11,248,825 ops/sec ±0.55% (190 runs sampled)
-  new URL()        x     343,111 ops/sec ±0.18% (193 runs sampled)
-  parseurl         x  28,455,610 ops/sec ±0.92% (187 runs sampled)
-  @polka/url       x  41,677,905 ops/sec ±1.07% (188 runs sampled)
+  url.parse#1      x  11,682,323 ops/sec ±0.81% (187 runs sampled)
+  url.parse#2      x  15,679,363 ops/sec ±0.61% (188 runs sampled)
+  new URL()        x     348,880 ops/sec ±0.35% (189 runs sampled)
+  parseurl         x  34,522,603 ops/sec ±0.73% (190 runs sampled)
+  @polka/url       x  71,136,459 ops/sec ±0.58% (189 runs sampled)
 ```
diff --git a/packages/url/index.d.ts b/packages/url/index.d.ts
@@ -7,4 +7,4 @@ export interface ParsedURL {
 	raw: string;
 }
 
-export function parse(req: IncomingMessage, toDecode?: boolean): ParsedURL;
+export function parse(req: IncomingMessage): ParsedURL;
diff --git a/packages/url/index.js b/packages/url/index.js
@@ -8,21 +8,19 @@ import * as qs from 'querystring';
 /**
  * @typedef Request
  * @property {string} url
- * @property {string} _decoded
  * @property {ParsedURL} _parsedUrl
  */
 
 /**
  * @param {Request} req
- * @param {boolean} [toDecode]
  * @returns {ParsedURL|void}
  */
-export function parse(req, toDecode) {
+export function parse(req) {
 	let raw = req.url;
 	if (raw == null) return;
 
-	let prev=req._parsedUrl, encoded=!req._decoded;
-	if (prev && prev.raw === raw && !toDecode === encoded) return prev;
+	let prev = req._parsedUrl;
+	if (prev && prev.raw === raw) return prev;
 
 	let pathname=raw, search='', query;
 
@@ -36,15 +34,6 @@ export function parse(req, toDecode) {
 				query = qs.parse(search.substring(1));
 			}
 		}
-
-		if (!!toDecode && encoded) {
-			if (pathname.indexOf('%') === -1) {
-				req._decoded = pathname;
-			} else {
-				try { pathname = req._decoded = decodeURIComponent(pathname) }
-				catch (e) { /* URI malformed */ }
-			}
-		}
 	}
 
 	return req._parsedUrl = { pathname, search, query, raw };