diff --git a/package.json b/package.json
index 480e5c1..bec5b70 100644
--- a/package.json
+++ b/package.json
@@ -44,11 +44,11 @@
"html-encode": "~2.1.6",
"html-urls": "~2.4.55",
"is-html-content": "~1.0.0",
+ "localhost-url-regex": "~1.0.11",
"lodash": "~4.17.21",
"mri": "~1.2.0",
"p-cancelable": "~2.1.0",
"p-retry": "~4.6.0",
- "replace-string": "~3.1.0",
"tinyspawn": "~1.2.6",
"top-sites": "~1.1.205"
},
diff --git a/src/html.js b/src/html.js
index 27b9fe2..d0942e8 100644
--- a/src/html.js
+++ b/src/html.js
@@ -1,8 +1,8 @@
'use strict'
const { get, split, nth, castArray, forEach } = require('lodash')
+const localhostUrl = require('localhost-url-regex')
const { TAGS: URL_TAGS } = require('html-urls')
-const replaceString = require('replace-string')
const isHTML = require('is-html-content')
const cssUrl = require('css-url-regex')
const execall = require('execall')
@@ -95,7 +95,9 @@ const rewriteHtmlUrls = ({ $, url }) => {
const el = $(this)
const attr = el.attr(urlAttr)
- if (typeof attr === 'string' && !attr.startsWith('http')) {
+ if (localhostUrl().test(attr)) {
+ el.remove()
+ } else if (typeof attr === 'string' && !attr.startsWith('http')) {
try {
const newAttr = new URL(attr, url).toString()
el.attr(urlAttr, newAttr)
@@ -117,7 +119,7 @@ const rewriteCssUrls = ({ html, url }) => {
if (cssUrl.startsWith('/')) {
try {
const absoluteUrl = new URL(cssUrl, url).toString()
- html = replaceString(html, `url(${cssUrl})`, `url(${absoluteUrl})`)
+ html = html.replaceAll(`url(${cssUrl})`, `url(${absoluteUrl})`)
} catch (_) {}
}
})
diff --git a/test/html/index.js b/test/html/index.js
index 01d0384..598e80d 100644
--- a/test/html/index.js
+++ b/test/html/index.js
@@ -1,12 +1,7 @@
-/* eslint-disable prefer-regex-literals */
-
'use strict'
const cheerio = require('cheerio')
-const execall = require('execall')
-const path = require('path')
const test = require('ava')
-const fs = require('fs')
const { prettyHtml } = require('../util')
@@ -88,221 +83,6 @@ test('add video markup', t => {
t.snapshot(prettyHtml(output))
})
-test("'`rewriteCssUrls` don't modify html markup", t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://www.rubiomonocoatusa.com/blogs/blog/how-to-apply-oil-plus-2c-to-furniture',
- html: `
-
-
-
-
-
-
-`,
- headers: {
- 'content-type': 'text/html; charset=utf-8'
- }
- })
-
- t.true(
- output.includes(
- 'content="http://cdn.shopify.com/s/files/1/0260/4810/2497/articles/Applying-Oil-Plus-2C-to-a-table_600x.jpg?v=1616464305"'
- )
- )
-
- t.true(
- output.includes(
- 'url(https://cdn.shopify.com/s/files/1/0260/4810/2497/articles/Applying-Oil-Plus-2C-to-a-table_600x.jpg?v=1616464305)'
- )
- )
-
- t.snapshot(prettyHtml(output))
-})
-
-test('`rewriteHtmlUrls` rewrites relative root URLs inside html markup', t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://browserless.js.org',
- html: fs.readFileSync(
- path.resolve(__dirname, '../fixtures/browserless.html'),
- 'utf8'
- ),
- headers: {
- 'content-type': 'text/html; charset=utf-8'
- }
- })
-
- t.true(output.includes('https://browserless.js.org/static/main.min.js'))
- t.true(output.includes('https://unpkg.com/docsify/lib/docsify.min.js'))
-
- t.snapshot(prettyHtml(output))
-})
-
-test('`rewriteHtmlUrls` rewrites relative URLs inside html markup', t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://moovility.me/',
- html: `
-
-
-
-
-
- `,
- headers: {
- 'content-type': 'text/html; charset=utf-8'
- }
- })
-
- t.true(output.includes('https://moovility.me/img/icons/MOV/icon2-76.png'))
-
- t.snapshot(prettyHtml(output))
-})
-
-test('`rewriteHtmlUrls` rewrites relative URLs inside stylesheet', t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://kikobeats.com',
- html: `
-
-
-
-
-
-
- `,
- headers: {
- 'content-type': 'text/html; charset=utf-8'
- }
- })
-
- const results = execall(
- new RegExp('https://kikobeats.com/images/microlink.jpg', 'g'),
- output
- )
-
- t.is(results.length, 2)
- t.snapshot(prettyHtml(output))
-})
-
-test("`rewriteHtmlUrls` don't modify inline javascript", t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://www.latimes.com/opinion/story/2020-06-07/column-muralist-honors-african-americans-killed-by-police',
- html: `
-
-
-
-
-
-
-
-Print
-
-`,
- headers: {
- 'content-type': 'text/html;charset=UTF-8'
- }
- })
-
- t.true(
- output.includes(
- 'Print'
- )
- )
-
- t.snapshot(prettyHtml(output))
-})
-
-test("`rewriteHtmlUrls` don't modify non http protocols", t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://www.latimes.com/opinion/story/2020-06-07/column-muralist-honors-african-americans-killed-by-police',
- html: `
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-`,
- headers: {
- 'content-type': 'text/html;charset=UTF-8'
- }
- })
-
- t.true(output.includes(''))
- t.true(output.includes(''))
- t.true(output.includes(''))
- t.true(output.includes(''))
- t.true(output.includes(''))
-
- t.snapshot(prettyHtml(output))
-})
-
-test("`rewriteHtmlUrls` don't modify data URIs", t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://example.com',
- html: `
-
-
-
-
-
-
-
-
-
-`,
- headers: {
- 'content-type': 'text/html;charset=UTF-8'
- }
- })
-
- t.true(
- output.includes(
- ''
- )
- )
-
- t.snapshot(prettyHtml(output))
-})
-
-test("`rewriteHtmlUrls` don't modify undefined attributes", t => {
- const output = html({
- rewriteUrls: true,
- url: 'https://moovility.me',
- html: `
-
-
-
- Document
-
-
-
-
-`,
- headers: {
- 'content-type': 'text/html;charset=UTF-8'
- }
- })
-
- t.true(output.includes(""))
-
- t.snapshot(prettyHtml(output))
-})
-
test('styles injection', t => {
const output = html({
url: 'https://kikobeats.com',
diff --git a/test/html/rewrite-css-urls.js b/test/html/rewrite-css-urls.js
new file mode 100644
index 0000000..e1baeca
--- /dev/null
+++ b/test/html/rewrite-css-urls.js
@@ -0,0 +1,66 @@
+'use strict'
+
+const execall = require('execall')
+const test = require('ava')
+
+const { prettyHtml } = require('../util')
+
+const html = require('../../src/html')
+
+test("don't modify html markup", t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://www.rubiomonocoatusa.com/blogs/blog/how-to-apply-oil-plus-2c-to-furniture',
+ html: `
+
+
+
+
+
+
+`,
+ headers: {
+ 'content-type': 'text/html; charset=utf-8'
+ }
+ })
+
+ t.true(
+ output.includes(
+ 'content="http://cdn.shopify.com/s/files/1/0260/4810/2497/articles/Applying-Oil-Plus-2C-to-a-table_600x.jpg?v=1616464305"'
+ )
+ )
+
+ t.true(
+ output.includes(
+ 'url(https://cdn.shopify.com/s/files/1/0260/4810/2497/articles/Applying-Oil-Plus-2C-to-a-table_600x.jpg?v=1616464305)'
+ )
+ )
+
+ t.snapshot(prettyHtml(output))
+})
+
+test('rewrites relative URLs inside stylesheet', t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://kikobeats.com',
+ html: `
+
+
+
+
+
+
+ `,
+ headers: {
+ 'content-type': 'text/html; charset=utf-8'
+ }
+ })
+
+ const results = execall(
+ /https:\/\/kikobeats.com\/images\/microlink\.jpg/g,
+ output
+ )
+
+ t.is(results.length, 2)
+ t.snapshot(prettyHtml(output))
+})
diff --git a/test/html/rewrite-urls.js b/test/html/rewrite-urls.js
new file mode 100644
index 0000000..ba4f820
--- /dev/null
+++ b/test/html/rewrite-urls.js
@@ -0,0 +1,190 @@
+'use strict'
+
+const path = require('path')
+const test = require('ava')
+const fs = require('fs')
+
+const { prettyHtml } = require('../util')
+
+const html = require('../../src/html')
+
+test('remove localhost alike URLs', async t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://kikobeats.com',
+ html: `
+
+
+ kikobeats.com
+
+
+
+
+
+
+
+
+
+ `,
+ headers: { 'content-type': 'text/html; charset=utf-8' }
+ })
+
+ t.snapshot(prettyHtml(output))
+})
+
+test('rewrites relative root URLs inside html markup', t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://browserless.js.org',
+ html: fs.readFileSync(
+ path.resolve(__dirname, '../fixtures/browserless.html'),
+ 'utf8'
+ ),
+ headers: {
+ 'content-type': 'text/html; charset=utf-8'
+ }
+ })
+
+ t.true(output.includes('https://browserless.js.org/static/main.min.js'))
+ t.true(output.includes('https://unpkg.com/docsify/lib/docsify.min.js'))
+
+ t.snapshot(prettyHtml(output))
+})
+
+test('rewrites relative URLs inside html markup', t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://moovility.me/',
+ html: `
+
+
+
+
+
+ `,
+ headers: {
+ 'content-type': 'text/html; charset=utf-8'
+ }
+ })
+
+ t.true(output.includes('https://moovility.me/img/icons/MOV/icon2-76.png'))
+
+ t.snapshot(prettyHtml(output))
+})
+
+test(" don't modify inline javascript", t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://www.latimes.com/opinion/story/2020-06-07/column-muralist-honors-african-americans-killed-by-police',
+ html: `
+
+
+
+
+
+
+
+Print
+
+`,
+ headers: {
+ 'content-type': 'text/html;charset=UTF-8'
+ }
+ })
+
+ t.true(
+ output.includes(
+ 'Print'
+ )
+ )
+
+ t.snapshot(prettyHtml(output))
+})
+
+test("don't modify non http protocols", t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://www.latimes.com/opinion/story/2020-06-07/column-muralist-honors-african-americans-killed-by-police',
+ html: `
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+`,
+ headers: {
+ 'content-type': 'text/html;charset=UTF-8'
+ }
+ })
+
+ t.true(output.includes(''))
+ t.true(output.includes(''))
+ t.true(output.includes(''))
+ t.true(output.includes(''))
+ t.true(output.includes(''))
+
+ t.snapshot(prettyHtml(output))
+})
+
+test("don't modify data URIs", t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://example.com',
+ html: `
+
+
+
+
+
+
+
+
+
+`,
+ headers: {
+ 'content-type': 'text/html;charset=UTF-8'
+ }
+ })
+
+ t.true(
+ output.includes(
+ ''
+ )
+ )
+
+ t.snapshot(prettyHtml(output))
+})
+
+test("don't modify undefined attributes", t => {
+ const output = html({
+ rewriteUrls: true,
+ url: 'https://moovility.me',
+ html: `
+
+
+
+ Document
+
+
+
+
+`,
+ headers: {
+ 'content-type': 'text/html;charset=UTF-8'
+ }
+ })
+
+ t.true(output.includes(""))
+
+ t.snapshot(prettyHtml(output))
+})
diff --git a/test/html/snapshots/index.js.md b/test/html/snapshots/index.js.md
index c1df5b5..10974a3 100644
--- a/test/html/snapshots/index.js.md
+++ b/test/html/snapshots/index.js.md
@@ -110,205 +110,6 @@ Generated by [AVA](https://avajs.dev).
␊