Add get language from amazon urls

microlinkhq · Dec 30, 2017 · f6de069 · f6de069
1 parent 5ec9472
commit f6de069
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 0 deletions.
diff --git a/packages/metascraper-amazon/__snapshots__/index.js.snap-shot b/packages/metascraper-amazon/__snapshots__/index.js.snap-shot
@@ -1,4 +1,5 @@
 exports['product url 1'] = {
+  "lang": "en",
   "author": "Mat Follas",
   "title": "Vegetable Perfection: 100 delicious recipes for roots, bulbs, shoots and stems",
   "publisher": "Amazon",
@@ -9,6 +10,7 @@ exports['product url 1'] = {
 }
 
 exports['ansi url 1'] = {
+  "lang": "en",
   "author": "SainSmart",
   "title": "SainSmart 4-Channel Relay Module",
   "publisher": "Amazon",
@@ -19,6 +21,7 @@ exports['ansi url 1'] = {
 }
 
 exports['product url 2'] = {
+  "lang": "en",
   "author": "David Baldacci",
   "title": "The Whole Truth (A. Shaw Book 1) Kindle Edition",
   "publisher": "Amazon",
@@ -29,6 +32,7 @@ exports['product url 2'] = {
 }
 
 exports['product url 3'] = {
+  "lang": "es",
   "author": "Cecotec",
   "title": "Robot aspirador Conga Excellence 990 de Cecotec. Friega el suelo, Barre, aspira y pasa la mopa. Programable 24h. 5 modos de limpieza. Base de carga. Silencioso. Potente. Filtro HEPA.",
   "publisher": "Amazon",

diff --git a/packages/metascraper-amazon/index.js b/packages/metascraper-amazon/index.js
@@ -1,18 +1,41 @@
 'use strict'
 
 const { titleize, isUrl } = require('@metascraper/helpers')
+const { URL } = require('url')
 
 const REGEX_AMAZON_URL = /https?:\/\/(.*amazon\..*\/.*|.*amzn\..*\/.*|.*a\.co\/.*)/i
 const isAmazonUrl = url => REGEX_AMAZON_URL.test(url)
 
+const SUFFIX_LANGUAGES = {
+  'ca': 'en',
+  'cn': 'zh',
+  'co.jp': 'ja',
+  'co.uk': 'en',
+  'com.mx': 'es',
+  'com': 'en',
+  'de': 'de',
+  'es': 'es',
+  'fr': 'fr',
+  'in': 'en',
+  'it': 'it'
+}
+
 const wrap = rule => ({ htmlDom, url }) => isAmazonUrl(url) && rule(htmlDom)
 
 const wrapUrl = rule => ({ htmlDom, url }) => {
   const value = wrap(rule)({htmlDom, url})
   return isUrl(value) && value
 }
 
+const getDomainLanguage = url => {
+  const {host} = new URL(url)
+  const suffix = host.replace('www.', '').split('.')
+  suffix.shift()
+  return SUFFIX_LANGUAGES[suffix.join('.')]
+}
+
 module.exports = () => ({
+  lang: [({ htmlDom: $, meta, url }) => getDomainLanguage(url)],
   author: [
     wrap($ => titleize($('.contributorNameID').text())),
     wrap($ => titleize($('#bylineInfo').text())),

diff --git a/packages/metascraper-logo-favicon/__snapshots__/index.js.snap-shot b/packages/metascraper-logo-favicon/__snapshots__/index.js.snap-shot
@@ -1,4 +1,5 @@
 exports['create an absolute faivcon url if the logo is not present 1'] = {
+  "lang": "en",
   "author": "Mat Follas",
   "title": "Vegetable Perfection: 100 delicious recipes for roots, bulbs, shoots and stems",
   "publisher": "Amazon",