Support resolve URL

Author: Kikobeats

src/get-data/rules/url.js

@@ -1,22 +1,21 @@
 'use strict'
 
-const normalizeUrl = require('normalize-url')
+const sanetizeUrl = require('normalize-url')
 const isString = require('lodash.isstring')
 
+const normalizeUrl = url => sanetizeUrl(url, {stripWWW: false})
+
 /**
  * Wrap a rule with validation and formatting logic.
  *
  * @param {Function} rule
  * @return {Function} wrapped
  */
 
-const wrap = rule => $ => {
-  let value = rule($)
+const wrap = rule => (htmlDom, url) => {
+  const value = rule(htmlDom)
   if (!isString(value)) return
-
-  return normalizeUrl(value, {
-    stripWWW: false
-  })
+  return normalizeUrl(value)
 }
 
 /**
@@ -27,5 +26,5 @@ module.exports = [
   wrap($ => $('meta[property="og:url"]').attr('content')),
   wrap($ => $('meta[name="twitter:url"]').attr('content')),
   wrap($ => $('link[rel="canonical"]').attr('href')),
-  wrap(($, url) => url)
+  ($, url) => isString(url) ? normalizeUrl(url) : null
 ]

test/web/astier/index.js

@@ -9,7 +9,7 @@ const fs = require('fs')
 const getMetaData = require('../../..')
 const readFile = promisify(fs.readFile)
 
-const url = 'https://anisse.astier.eu'
+const url = 'https://anisse.astier.eu/awk-driven-iot.html'
 
 it('astier', async () => {
   const html = await readFile(resolve(__dirname, 'input.html'))

test/web/astier/output.json

@@ -5,5 +5,5 @@
   "image": "https://anisse.astier.eu/images/anisse.jpg",
   "publisher": null,
   "title": "Linux Engineer's random thoughts - awk driven IoT",
-  "url": null
+  "url": "https://anisse.astier.eu/awk-driven-iot.html"
 }