Rewrite URL parser to run in linear time

Author: gregjacobs

README.md

@@ -1,6 +1,6 @@
 {
     "name": "autolinker",
-    "version": "3.16.2",
+    "version": "4.0.0",
     "description": "Utility to automatically link the URLs, email addresses, phone numbers, hashtags, and mentions (Twitter, Instagram) in a given block of text/HTML",
     "main": "./dist/commonjs/index.js",
     "typings": "./dist/commonjs/index.d.ts",
@@ -18,6 +18,7 @@
         "start": "webpack serve --open",
         "test": "npm-run-all test:unit test:integration",
         "test:unit": "node --require=ts-node/register node_modules/jasmine/bin/jasmine.js \"tests/**/*.spec.ts\"",
+        "test:unit:debug": "node --inspect-brk --require=ts-node/register node_modules/jasmine/bin/jasmine.js \"tests/**/*.spec.ts\"",
         "test:integration": "ts-node scripts/test-integration.ts",
         "update-tld-regex": "ts-node scripts/update-tld-regex.ts",
         "prepare": "husky install"
@@ -46,6 +47,7 @@
     "devDependencies": {
         "@rollup/plugin-commonjs": "^21.0.0",
         "@rollup/plugin-node-resolve": "^13.1.3",
+        "@types/cli-table": "^0.3.0",
         "@types/dedent": "^0.7.0",
         "@types/fs-extra": "^9.0.13",
         "@types/jasmine": "^3.10.3",
@@ -54,6 +56,7 @@
         "@types/node": "^16.11.26",
         "@types/webpack": "^5.28.0",
         "axios": "^0.26.0",
+        "cli-table": "^0.3.11",
         "css-loader": "^6.6.0",
         "dedent": "^0.7.0",
         "fast-glob": "^3.2.11",

package-lock.json

@@ -15,10 +15,10 @@ const es2015OutputDir = './dist/es2015';
 fse.outputFileSync(
     `${pkgDir}/src/version.ts`,
     dedent`
-		// Important: this file is generated from the 'build' script and should not be
-		// edited directly
-		export const version = '${pkg.version}';
-	`,
+        // Important: this file is generated from the 'build' script and should not be
+        // edited directly
+        export const version = '${pkg.version}';
+    ` + '\n',
     'utf-8'
 );
 
@@ -119,15 +119,17 @@ async function checkMinifiedFileSize() {
     const sizeInKb = stats.size / 1000;
     const maxExpectedSizeInKb = 47;
 
+    console.log(`dist/autolinker.min.js size: ${sizeInKb}kb`);
+
     if (sizeInKb > maxExpectedSizeInKb) {
         throw new Error(dedent`
-			Minified file size of ${sizeInKb.toFixed(2)}kb is greater than max 
-			expected minified size of ${maxExpectedSizeInKb}kb
-			
-			This check is to make sure that a dependency is not accidentally 
-			added which significantly inflates the size of Autolinker. If 
-			additions to the codebase have been made though and a higher size 
-			is expected, bump the 'maxExpectedSizeInKb' number in ${__filename}
-		`);
+            Minified file size of ${sizeInKb.toFixed(2)}kb is greater than max 
+            expected minified size of ${maxExpectedSizeInKb}kb
+            
+            This check is to make sure that a dependency is not accidentally 
+            added which significantly inflates the size of Autolinker. If 
+            additions to the codebase have been made though and a higher size 
+            is expected, bump the 'maxExpectedSizeInKb' number in ${__filename}
+        `);
     }
 }

package.json

@@ -22,8 +22,6 @@ import { execSync } from './util/exec-sync';
     fse.copySync(testsSrcDir, testsOutputDir);
 
     // Create a .tar.gz output file like the one that would be downloaded from npm
-    // TODO: Was using 'yarn' - does npm have a --filename arg?
-    //await exec( `./node_modules/.bin/yarn pack --filename ./.tmp/tests-integration/autolinker.tar.gz`, {
     execSync(`${pkgRoot}/node_modules/.bin/npm pack --pack-destination ${testsOutputDir}`, {
         cwd: pkgRoot,
     });

scripts/build.ts

@@ -16,12 +16,16 @@ export async function updateTldRegex() {
         responseType: 'text',
     });
 
-    const tldRegex = domainsToRegex(tldsFile.data);
-    let outputFile = dedent`
-        // NOTE: THIS IS A GENERATED FILE\n// To update with the latest TLD list, run \`npm run update-tld-regex\`\n\n
-        ${tldRegex}
-    `;
-    fse.writeFile('./src/matcher/tld-regex.ts', outputFile);
+    const tldRegexStr = domainsToRegex(tldsFile.data);
+    let outputFile =
+        dedent`
+        // NOTE: THIS IS A GENERATED FILE\n// To update with the latest TLD list, run \`npm run update-tld-regex\`
+        
+        export const tldRegexStr = '(?:${tldRegexStr})';\n
+
+        export const tldRegex = new RegExp('^' + tldRegexStr + '$');
+    ` + '\n';
+    await fse.outputFile('./src/parser/tld-regex.ts', outputFile);
 }
 
 function domainsToRegex(contents: string): string {
@@ -33,8 +37,7 @@ function domainsToRegex(contents: string): string {
         .filter(s => !!s) // remove empty elements;
         .sort(compareLengthLongestFirst);
 
-    const domainsRegexStr = `export const tldRegex = /(?:${domains.join('|')})/;\n`;
-    return domainsRegexStr;
+    return domains.join('|');
 }
 
 function notCommentLine(line: string): boolean {

scripts/test-integration.ts

@@ -1,9 +1,9 @@
-import { Match } from './match/match';
 import { HtmlTag } from './html-tag';
 import { TruncateConfigObj } from './autolinker';
 import { truncateSmart } from './truncate/truncate-smart';
 import { truncateMiddle } from './truncate/truncate-middle';
 import { truncateEnd } from './truncate/truncate-end';
+import { AbstractMatch } from './match/abstract-match';
 
 /**
  * @protected
@@ -64,11 +64,10 @@ export class AnchorTagBuilder {
      * Generates the actual anchor (<a>) tag to use in place of the
      * matched text, via its `match` object.
      *
-     * @param {Autolinker.match.Match} match The Match instance to generate an
-     *   anchor tag from.
-     * @return {Autolinker.HtmlTag} The HtmlTag instance for the anchor tag.
+     * @param match The Match instance to generate an anchor tag from.
+     * @return The HtmlTag instance for the anchor tag.
      */
-    build(match: Match) {
+    public build(match: AbstractMatch) {
         return new HtmlTag({
             tagName: 'a',
             attrs: this.createAttrs(match),
@@ -81,11 +80,10 @@ export class AnchorTagBuilder {
      *   tag being generated.
      *
      * @protected
-     * @param {Autolinker.match.Match} match The Match instance to generate an
-     *   anchor tag from.
-     * @return {Object} A key/value Object (map) of the anchor tag's attributes.
+     * @param match The Match instance to generate an anchor tag from.
+     * @return A key/value Object (map) of the anchor tag's attributes.
      */
-    protected createAttrs(match: Match) {
+    protected createAttrs(match: AbstractMatch) {
         let attrs: { [attrName: string]: string } = {
             href: match.getAnchorHref(), // we'll always have the `href` attribute
         };
@@ -122,13 +120,13 @@ export class AnchorTagBuilder {
      * - "myLink myLink-mention myLink-twitter"  // mention match with Twitter service
      *
      * @protected
-     * @param {Autolinker.match.Match} match The Match instance to generate an
+     * @param match The Match instance to generate an
      *   anchor tag from.
-     * @return {String} The CSS class string for the link. Example return:
+     * @return The CSS class string for the link. Example return:
      *   "myLink myLink-url". If no {@link #className} was configured, returns
      *   an empty string.
      */
-    protected createCssClass(match: Match) {
+    protected createCssClass(match: AbstractMatch): string {
         let className = this.className;
 
         if (!className) {
@@ -149,11 +147,11 @@ export class AnchorTagBuilder {
      * {@link #truncate} config.
      *
      * @private
-     * @param {String} anchorText The anchor tag's text (i.e. what will be
+     * @param anchorText The anchor tag's text (i.e. what will be
      *   displayed).
-     * @return {String} The processed `anchorText`.
+     * @return The processed `anchorText`.
      */
-    private processAnchorText(anchorText: string) {
+    private processAnchorText(anchorText: string): string {
         anchorText = this.doTruncate(anchorText);
 
         return anchorText;
@@ -166,11 +164,11 @@ export class AnchorTagBuilder {
      * `location` property. See {@link #truncate} for details.
      *
      * @private
-     * @param {String} anchorText The anchor tag's text (i.e. what will be
+     * @param anchorText The anchor tag's text (i.e. what will be
      *   displayed).
-     * @return {String} The truncated anchor text.
+     * @return The truncated anchor text.
      */
-    private doTruncate(anchorText: string) {
+    private doTruncate(anchorText: string): string {
         let truncate = this.truncate;
         if (!truncate || !truncate.length) return anchorText;