fix: remove www from hostname; strip &nbsp

Author: nerdstep

package.json

@@ -1,6 +1,6 @@
 {
   "name": "recipe-scrapers-js",
-  "version": "0.1.0-alpha.3",
+  "version": "0.1.0-alpha.4",
   "license": "MIT",
   "description": "A recipe scrapers library",
   "author": {

src/plugins/__tests__/html-stripper.processor.test.ts

@@ -20,6 +20,9 @@ describe('HtmlStripperPlugin', () => {
     expect(plugin.process('title', '<span>Test &lt;tag&gt;</span>')).toBe(
       'Test <tag>',
     )
+    expect(plugin.process('description', '<span>Hello&nbsp;World</span>')).toBe(
+      'Hello World',
+    )
   })
 
   it('strips HTML from instructions Set<string>', () => {

src/plugins/html-stripper.processor.ts

@@ -69,6 +69,7 @@ export class HtmlStripperPlugin extends PostProcessorPlugin {
     return html
       .replace(/<[^>]*>/g, '') // Remove HTML tags
       .replace(/&amp;/g, '&') // Decode common entities
+      .replace(/&nbsp;/g, ' ')
       .replace(/&lt;/g, '<')
       .replace(/&gt;/g, '>')
       .replace(/&quot;/g, '"')

src/utils/__tests__/index.test.ts

@@ -1,5 +1,6 @@
 import { describe, expect, it } from 'bun:test'
 import {
+  getHostName,
   isDefined,
   isFunction,
   isNumber,
@@ -92,3 +93,36 @@ describe('isString', () => {
     expect(isString({})).toBe(false)
   })
 })
+describe('getHostName', () => {
+  it('should return the host for a standard URL', () => {
+    expect(getHostName('https://www.example.com/path')).toBe('example.com')
+  })
+
+  it('should return the host for a URL with a subdomain', () => {
+    expect(getHostName('http://sub.domain.co.uk/page?q=1')).toBe(
+      'sub.domain.co.uk',
+    )
+  })
+
+  it('should return the host for a URL without a path', () => {
+    expect(getHostName('https://anothersite.org')).toBe('anothersite.org')
+  })
+
+  it('should throw an error for an invalid URL string', () => {
+    const invalidUrl = 'not a url'
+    expect(() => getHostName(invalidUrl)).toThrow(
+      new Error(`Invalid URL: ${invalidUrl}`),
+    )
+  })
+
+  it('should throw an error for an empty string', () => {
+    expect(() => getHostName('')).toThrow(new Error('Invalid URL: '))
+  })
+
+  it('should throw an error for a string that looks like a host but lacks a protocol', () => {
+    const urlWithoutProtocol = 'example.com'
+    expect(() => getHostName(urlWithoutProtocol)).toThrow(
+      new Error(`Invalid URL: ${urlWithoutProtocol}`),
+    )
+  })
+})

src/utils/index.ts

@@ -26,11 +26,13 @@ export function isString(value: unknown): value is string {
 }
 
 /**
- * Extracts the host name from a URL string.
+ * Extracts the host name from a URL string
+ * and removes 'www.' prefix if present.
+ * Throws an error if the input is not a valid URL.
  */
 export function getHostName(value: string) {
   try {
-    const url = new URL(value)
+    const url = new URL(value.replace('www.', ''))
     return url.host
   } catch {
     throw new Error(`Invalid URL: ${value}`)