chore: add tests and docs

microlinkhq · Jun 24, 2024 · a4b3306 · a4b3306
1 parent 5d5136f
commit a4b3306
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 7 deletions.
diff --git a/README.md b/README.md
@@ -264,6 +264,12 @@ Type: `String`
 
 The HTML markup for extracting the content.
 
+##### htmlDom
+
+Type: `object`
+
+The DOM representation of the HTML markup. When it's not provided, it's get from the `html` parameter.
+
 #### rules
 
 Type: `Array`

diff --git a/packages/metascraper/src/index.js b/packages/metascraper/src/index.js
@@ -14,22 +14,21 @@ module.exports = rules => {
   return async ({
     url,
     html = '',
-    dom,
+    htmlDom,
     rules: inlineRules,
     validateUrl = true,
     ...props
   } = {}) => {
-    if (validateUrl && !isUrl(url)) {
+    if (!validateUrl && !isUrl(url)) {
       throw new MetascraperError({
         message: 'Need to provide a valid URL.',
         code: 'INVALID_URL'
       })
     }
-    const htmlDom = dom || load(html, { baseURI: url })
 
     return getData({
       url,
-      htmlDom,
+      htmlDom: htmlDom ?? load(html, { baseURI: url }),
       rules: mergeRules(inlineRules, loadedRules),
       ...props
     })

diff --git a/packages/metascraper/test/unit/interface.js b/packages/metascraper/test/unit/interface.js
@@ -4,6 +4,7 @@ const test = require('ava')
 
 const createMetascraper = require('../..')
 const titleRules = require('metascraper-title')()
+const { load } = require('cheerio')
 
 test('`url` is required', async t => {
   t.plan(9)
@@ -31,7 +32,7 @@ test('`url` is required', async t => {
   }
 })
 
-test('Disable URL validation using `validateUrl`', async t => {
+test('passing `{ validateUrl: false }`', async t => {
   const metascraper = createMetascraper([titleRules])
 
   const html = `
@@ -66,7 +67,7 @@ test('Disable URL validation using `validateUrl`', async t => {
   t.is(metadata.title, 'Document')
 })
 
-test('load extra `rules`', async t => {
+test('passing `rules`', async t => {
   const url = 'https://microlink.io'
 
   const html = `
@@ -104,7 +105,7 @@ test('load extra `rules`', async t => {
   t.is(metadata.foo, 'bar')
 })
 
-test('associate test function with rules', async t => {
+test('skip `rules` via `test` function', async t => {
   const url = 'https://microlink.io'
 
   const html = `
@@ -148,3 +149,36 @@ test('associate test function with rules', async t => {
   t.is(metadata.foo, null)
   t.true(isCalled)
 })
+
+test('passing `htmlDom`', async t => {
+  const url = 'https://microlink.io'
+
+  const htmlDom = load('<title>htmlDom</title>')
+
+  const html = `
+  <!DOCTYPE html>
+  <html lang="en">
+  <head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <meta http-equiv="X-UA-Compatible" content="ie=edge">
+    <title>Original HTML</title>
+  </head>
+  <body>
+    <div class="logos">
+      <img class="logo" href="https://microlink.io/logo.png">
+      <img class="logo" href="https://microlink.io/logo.png">
+      <img class="logo" href="https://microlink.io/logo.png">
+      <img class="logo" href="https://microlink.io/logo.png">
+    </div>
+
+    <img class="main-logo" href="https://microlink.io/logo.png">
+    <p>Hello World </p>
+  </body>
+  </html>
+  `
+
+  const metascraper = createMetascraper([titleRules])
+  const metadata = await metascraper({ url, htmlDom, html })
+  t.is(metadata.title, 'htmlDom')
+})