feat: remove node-html-parser from browser bundle

Author: SettingDust

build.ts

@@ -9,7 +9,7 @@ import { nodeExternalsPlugin } from 'esbuild-node-externals';
     target: 'es2017',
     logLevel: 'info',
     color: true,
-    // minify: true,
+    minify: true,
     sourcemap: true,
     legalComments: 'external'
   }
@@ -20,20 +20,26 @@ import { nodeExternalsPlugin } from 'esbuild-node-externals';
     outfile: 'dist/node-html-markdown.js',
     format: 'esm',
     platform: 'node',
+    define: {
+      '__IS_BROWSER__': 'false'
+    },
     plugins: [ nodeExternalsPlugin({
       packagePath: './package.json'
     }) ]
-  }).then()
+  })
 
   await build({
     ...commonOptions,
     outfile: 'dist/node-html-markdown.cjs',
     format: 'cjs',
     platform: 'node',
+    define: {
+      '__IS_BROWSER__': 'false'
+    },
     plugins: [ nodeExternalsPlugin({
       packagePath: './package.json'
     }) ]
-  }).then()
+  })
 
   /**
    * Browser bundles are use for import/require directly from browser scripts.
@@ -45,17 +51,23 @@ import { nodeExternalsPlugin } from 'esbuild-node-externals';
     format: 'esm',
     platform: 'browser',
     define: {
-      'process.env.LOG_PERF': process.env.LOG_PERF ?? 'false'
-    }
-  }).then()
+      // This will remove the perfStart/End after minified by esbuild.
+      'process.env.LOG_PERF': process.env.LOG_PERF ?? 'false',
+      // This will remove the node-html-parser usage
+      '__IS_BROWSER__': 'true'
+    },
+    external: ['node-html-parser']
+  })
 
   await build({
     ...commonOptions,
     outfile: 'dist/node-html-markdown.browser.cjs',
     format: 'cjs',
     platform: 'browser',
     define: {
-      'process.env.LOG_PERF': process.env.LOG_PERF ?? 'false'
-    }
-  }).then()
+      'process.env.LOG_PERF': process.env.LOG_PERF ?? 'false',
+      '__IS_BROWSER__': 'true'
+    },
+    external: ['node-html-parser']
+  })
 })()

jest.config.ts

@@ -13,6 +13,9 @@ const config: Config = {
       }
     ]
   },
+  globals: {
+    '__IS_BROWSER__': false
+  },
   modulePaths: [ '<rootDir>' ],
   testTimeout: 10000,
   roots: [ '<rootDir>' ],

src/config.ts

@@ -1,7 +1,7 @@
 import { isWhiteSpaceOnly, splitSpecial, surround, tagSurround, trimNewLines } from './utilities';
 import { PostProcessResult, TranslatorConfigObject } from './translator';
 import { NodeHtmlMarkdownOptions } from './options';
-import { Options as NodeHtmlParserOptions } from 'node-html-parser'
+import type { Options as NodeHtmlParserOptions } from 'node-html-parser'
 
 
 /* ****************************************************************************************************************** */

src/nodes.ts

@@ -1,13 +1,10 @@
-import * as NHParser from 'node-html-parser';
-import { CommentNode, NodeType } from 'node-html-parser';
-
+import type * as NHParser from 'node-html-parser';
+import type { CommentNode } from 'node-html-parser';
 
 /* ****************************************************************************************************************** */
 // region: Types
 /* ****************************************************************************************************************** */
 
-export { NodeType, CommentNode }
-
 /* ********************************************************* *
  * Merged Nodes - Unions of node-html-parser and common DOM
  * ********************************************************* */
@@ -25,6 +22,12 @@ export type TextNode = (NHParser.TextNode) & NodeBase
 // region: TypeGuards
 /* ****************************************************************************************************************** */
 
+const NodeType = {
+  ELEMENT_NODE: 1,
+  TEXT_NODE: 3,
+  COMMENT_NODE: 8
+}
+
 export const isTextNode = (node: HtmlNode): node is TextNode => node.nodeType === NodeType.TEXT_NODE;
 export const isCommentNode = (node: HtmlNode): node is CommentNode => node.nodeType === NodeType.COMMENT_NODE;
 export const isElementNode = (node: HtmlNode): node is ElementNode => node.nodeType === NodeType.ELEMENT_NODE;

src/utilities.ts

@@ -2,7 +2,6 @@ import { NodeHtmlMarkdownOptions } from './options';
 import { ElementNode, HtmlNode } from './nodes';
 import { nodeHtmlParserConfig } from './config';
 
-
 /* ****************************************************************************************************************** */
 // region: String Utils
 /* ****************************************************************************************************************** */
@@ -123,6 +122,9 @@ export const truthyStr = (v: any, value?: string): string => v ? ((value !== und
 /* ****************************************************************************************************************** */
 // region: Parser
 /* ****************************************************************************************************************** */
+// For esbuild removing code
+declare global {const __IS_BROWSER__: boolean}
+
 
 function tryParseWithNativeDom(html: string): ElementNode | undefined {
   try {
@@ -169,24 +171,25 @@ const getNodeHtmlParser = () => {
 export function parseHTML(html: string, options: NodeHtmlMarkdownOptions): ElementNode {
   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) {
+  /* If specified, try to parse with native engine, fallback to node-html-parser */
+  if (__IS_BROWSER__ || options.preferNativeParser) {
     try {
       el = tryParseWithNativeDom(html);
     }
     catch (e) {
+      if (__IS_BROWSER__) throw e;
       nodeHtmlParse = getNodeHtmlParser();
       if (nodeHtmlParse) console.warn('Native DOM parser encountered an error during parse', e);
       else throw e;
     }
   } else nodeHtmlParse = getNodeHtmlParser();
 
-  if (!el) el = nodeHtmlParse!(html, nodeHtmlParserConfig);
+  if (!__IS_BROWSER__ && !el) el = nodeHtmlParse!(html, nodeHtmlParserConfig);
   perfStop('parse');
 
-  return el;
+  return el!;
 }
 
 // endregion