feat: Added performance enhancements + improved benchmark display

README.md

@@ -35,31 +35,39 @@ npm i -S node-html-parser
 
 ## Benchmarks
 ```
-node-html-makrdown (reused instance): 47.2261 ms/file ± 29.1069 (avg bytes/sec: 2.086)
-node-html-markdown                  : 55.2654 ms/file ± 38.2978 (avg bytes/sec: 1.782)
-turndown (reused instance)          : 66.8692 ms/file ± 30.9974 (avg bytes/sec: 1.473)
-turndown                            : 68.9757 ms/file ± 35.0430 (avg bytes/sec: 1.428)
+--------------------------------------------------------------------------------------
 
-Comparison to fastest (node-html-makrdown (reused instance)):
-  node-html-markdown: -14.55%
-  turndown (reused instance): -29.38%
-  turndown: -31.53%
+node-html-makrdown (reused instance): 46.0008 ms/file ± 27.3827 (avg bytes/sec: 2.141)
+node-html-markdown                  : 47.0208 ms/file ± 28.4478 (avg bytes/sec: 2.095)
+turndown                            : 80.7753 ms/file ± 33.3519 (avg bytes/sec: 1.219)
+turndown (reused instance)          : 65.9743 ms/file ± 32.3411 (avg bytes/sec: 1.493)
+
+--------------------------------------------------------------------------------------
 
 Estimated processing times (fastest to slowest):
 
   [node-html-makrdown (reused instance)]
-    100 kB:  0.05sec
-    1 MB:    0.50sec
-    50 MB:   25.14sec
-    1 GB:    8min, 34.80sec
-    50 GB:   7hr, 8min, 59.91sec
+    100 kB:  48ms
+    1 MB:    490ms
+    50 MB:   24.48sec
+    1 GB:    8min, 21sec
+    50 GB:   6hr, 57min, 52sec
 
   [turndown (reused instance)]
-    100 kB:  0.07sec
-    1 MB:    0.71sec
-    50 MB:   35.59sec
-    1 GB:    12min, 8.92sec
-    50 GB:   10hr, 7min, 26.09sec
+    100 kB:  69ms
+    1 MB:    702ms
+    50 MB:   35.12sec
+    1 GB:    11min, 59sec
+    50 GB:   9hr, 59min, 18sec
+
+--------------------------------------------------------------------------------------
+
+Comparison to fastest (node-html-makrdown (reused instance)):
+  node-html-markdown: -2.17%
+  turndown (reused instance): -30.27%
+  turndown: -43.05%
+
+--------------------------------------------------------------------------------------
 ```
 
 ## Usage

benchmark/_run.js

@@ -16,7 +16,8 @@ process.on('message', function (item) {
 		total: Benchmark.TOTAL,
 		complete: '=',
 		incomplete: ' ',
-		width: 50
+		width: 50,
+    clear: true
 	});
 
   const parser = require(item.parser);

benchmark/execute.js

@@ -19,6 +19,8 @@ const wrappers = fs.readdirSync(path.join(__dirname, 'wrapper'))
 
 const MAX_WIDTH = Math.max(...wrappers.map(wrapper => wrapper.name.length));
 
+const SEPARATOR = '\n' + '-'.repeat(MAX_WIDTH + 50) + '\n';
+
 // endregion
 
 
@@ -50,7 +52,9 @@ function humanTime(seconds) {
 
   for (const n of [ hours, minutes, s ]) if (!isFinite(n) || isNaN(n)) return 'N/A';
 
-  return `${hours ? hours + 'hr, ' : ''}${minutes ? minutes + 'min, ' : ''}${s.toFixed(2)}sec`;
+  return (!hours && !minutes && seconds < 1) ? `${Math.round((s % 1) * 1000)}ms` :
+         (!hours && !minutes) ? `${s.toFixed(2)}sec` :
+         `${hours ? hours + 'hr, ' : ''}${minutes ? minutes + 'min, ' : ''}${Math.round(s)}sec`;
 }
 
 // endregion
@@ -64,10 +68,12 @@ function humanTime(seconds) {
   if (!quickMode) console.log('NOTE: Large mode is generally less reliable in most environments!');
   const stats = [];
 
+  console.log(SEPARATOR);
+
   async.eachSeries(
     wrappers,
     function (item, done) {
-      const runner = fork(path.join(__dirname, '_run.js'), void 0, { env: { QUICK_MODE: quickMode }});
+      const runner = fork(path.join(__dirname, '_run.js'), void 0, { env: { QUICK_MODE: quickMode, LOG_PERF: true }});
       runner.send(item);
       runner.on('message', function (stat) {
         const name = formatName(item.name);
@@ -76,7 +82,7 @@ function humanTime(seconds) {
         const avgBytesPerSec = (stat.avgBytesPerMs / 1000).toPrecision(4);
 
         stats.push({ name, ...stat });
-        console.log(`\n${name}: ${mean} ms/file ± ${sd} (avg bytes/sec: ${avgBytesPerSec})`);
+        console.log(`${name}: ${mean} ms/file ± ${sd} (avg bytes/sec: ${avgBytesPerSec})`);
       });
 
       runner.on('close', function (n) {
@@ -85,10 +91,14 @@ function humanTime(seconds) {
       });
     },
     function () {
+      console.log(SEPARATOR);
       console.log(
-        `\nTotal Files: ${stats[0].totalFiles}\nAvg. file size: ${humanFileSize(stats[0].avgFileSize)}\n`);
+        `Total Files: ${stats[0].totalFiles}\n`+
+        `Avg. file size: ${humanFileSize(stats[0].avgFileSize)}`
+      );
 
       /* Get speed estimates */
+      console.log(SEPARATOR);
       console.log(`Estimated processing times (fastest to slowest):`);
       const sortedStats = [ ...stats ].sort((a,b) => b.avgBytesPerMs - a.avgBytesPerMs)
       sortedStats.forEach(({ name, avgBytesPerMs }) => {
@@ -103,13 +113,14 @@ function humanTime(seconds) {
       });
 
       /* Get comparisons */
-      console.log(`\nComparison to fastest (${sortedStats[0].name.trim()}): `);
+      console.log(SEPARATOR);
+      console.log(`Comparison to fastest (${sortedStats[0].name.trim()}): \n`);
       const fastestMean = sortedStats[0].mean;
       sortedStats.slice(1).forEach(({ name, mean }) =>
         console.log(`  ${name.trim()}: -${((1 - (fastestMean / mean)) * 100).toFixed(2)}%`)
       );
 
-      console.log('');
+      console.log(SEPARATOR);
     }
   );
 })();

benchmark/package.json

@@ -12,9 +12,9 @@
   },
   "dependencies": {
     "async": "^3.1.0",
+    "node-html-markdown": "link:../",
     "summary": "^1.0.0",
-    "turndown": "^7.0.0",
-    "node-html-markdown": "link:../"
+    "turndown": "^7.0.0"
   },
   "devDependencies": {
     "progress": "^2.0.3"

package.json

@@ -14,7 +14,7 @@
     "test": "jest",
     "test:coverage": "jest --collect-coverage",
     "-------------  ": "-------------",
-    "prepare": "cd benchmark && yarn install"
+    "prepare": "ts-patch patch tsc --silent && cd benchmark && yarn install"
   },
   "files": [
     "README.md",
@@ -52,6 +52,7 @@
     "jest": "^26.4.2",
     "ts-jest": "^26.4.1",
     "ts-node": "^9.0.0",
+    "ts-patch": "^1.3.1",
     "typescript": "^4.0.3"
   }
 }

src/config.ts

@@ -20,6 +20,8 @@ export const defaultIgnoreElements = [
   'SOURCE', 'STYLE', 'TRACK', 'WBR'
 ];
 
+export const contentlessElements = [ 'BR', 'HR', 'IMG' ];
+
 // endregion
 
 

src/nodes.ts

@@ -1,19 +1,22 @@
 import * as NHParser from 'node-html-parser';
-import { CommentNode, NodeType, TextNode } from 'node-html-parser';
+import { CommentNode, NodeType } from 'node-html-parser';
 
 
 /* ****************************************************************************************************************** */
 // region: Types
 /* ****************************************************************************************************************** */
 
-export { TextNode, NodeType, CommentNode }
+export { NodeType, CommentNode }
 
 /* ********************************************************* *
  * Merged Nodes - Unions of node-html-parser and common DOM
  * ********************************************************* */
 
-export type HtmlNode = NHParser.Node | Node
-export type ElementNode = NHParser.HTMLElement | HTMLElement
+type NodeBase = { preserve?: boolean }
+
+export type HtmlNode = (NHParser.Node | Node) & NodeBase
+export type ElementNode = (NHParser.HTMLElement | HTMLElement) & NodeBase
+export type TextNode = (NHParser.TextNode) & NodeBase
 
 // endregion
 

src/utilities.ts

@@ -1,5 +1,5 @@
 import { NodeHtmlMarkdownOptions } from './options';
-import { ElementNode } from './nodes';
+import { ElementNode, HtmlNode } from './nodes';
 import { nodeHtmlParserConfig } from './config';
 
 
@@ -78,6 +78,7 @@ export function parseHTML(html: string, options: NodeHtmlMarkdownOptions): Eleme
   let nodeHtmlParse: ReturnType<typeof getNodeHtmlParser>;
 
   /* If specified, try to parse with native engine, fallback to node-html-parser */
+  perfStart('parse');
   let el: ElementNode | undefined;
   if (options.preferNativeParser) {
     try {
@@ -86,12 +87,44 @@ export function parseHTML(html: string, options: NodeHtmlMarkdownOptions): Eleme
     catch (e) {
       nodeHtmlParse = getNodeHtmlParser();
       if (nodeHtmlParse) console.warn('Native DOM parser encountered an error during parse', e);
-      throw e;
+      else throw e;
     }
   }
   else nodeHtmlParse = getNodeHtmlParser();
 
-  return el || nodeHtmlParse!(html, nodeHtmlParserConfig).removeWhitespace();
+  if (!el) el = nodeHtmlParse!(html, nodeHtmlParserConfig).removeWhitespace();
+  perfStop('parse');
+
+  return el;
+}
+
+// endregion
+
+
+/* ****************************************************************************************************************** */
+// region: General
+/* ****************************************************************************************************************** */
+
+export function getChildNodes<T extends HtmlNode | Node>(node: T): T[]
+export function getChildNodes(node: HtmlNode | Node): (Node | HtmlNode)[] {
+  if (!isNodeList(node.childNodes)) return node.childNodes;
+
+  const res: (ChildNode)[] = [];
+  node.childNodes.forEach(n => res.push(n));
+
+  return res;
+
+  function isNodeList(v: any): v is NodeListOf<ChildNode> {
+    return (v != null) || (typeof v[Symbol.iterator] === 'function');
+  }
+}
+
+export function perfStart(label: string) {
+  if (process.env.LOG_PERF) console.time(label);
+}
+
+export function perfStop(label: string) {
+  if (process.env.LOG_PERF) console.timeEnd(label);
 }
 
 // endregion