Add language property

Closes #45

README.md

@@ -96,6 +96,9 @@ Here is a list of the metadata that **metascraper** collects by default:
 
 - **`image`** — eg. `https://assets.entrepreneur.com/content/3x2/1300/20160504155601-GettyImages-174457162.jpeg`<br/>
   An image URL that best represents the article.
+
+- **`lang`** — eg. `en`<br/>
+  An [ISO 639-1](https://en.wikipedia.org/wiki/ISO_639-1) representation of the url content language.
 
 - **`logo`** — eg. `https://entrepreneur.com/favicon180x180.png`<br/>
   An image URL that best represents the publisher brand.

packages/metascraper-lang/.npmrc

@@ -0,0 +1,4 @@
+unsafe-perm=true
+save-prefix=~
+shrinkwrap=false
+save=false

packages/metascraper-lang/README.md

@@ -0,0 +1,19 @@
+# metascraper-lang
+
+[![npm](https://img.shields.io/npm/v/metascraper-lang.svg?style=flat-square)](https://www.npmjs.com/package/metascraper-lang)
+[![Dependency Status](https://david-dm.org/microlinkhq/metascraper.svg?path=packages/metascraper-lang&style=flat-square)](https://david-dm.org/microlinkhq/metascraper?path=packages/metascraper-lang)
+
+> Get lang property from HTML markup.
+
+## Install
+
+```bash
+$ npm install metascraper-lang --save
+```
+
+## License
+
+**metascraper-lang** © [microlink.io](https://microlink.io), Released under the [MIT](https://github.com/microlinkhq/metascraper-lang/blob/master/LICENSE.md) License.<br>
+Authored and maintained by microlink.io with help from [contributors](https://github.com/microlinkhq/metascraper-lang/contributors).
+
+> [microlink.io](https://microlink.io) · GitHub [@microlink.io](https://github.com/microlinkhq) · Twitter [@microlinkhq](https://twitter.com/microlinkhq)

packages/metascraper-lang/index.js

@@ -0,0 +1,15 @@
+'use strict'
+
+const { isString, toLower } = require('lodash')
+
+const wrap = rule => ({ htmlDom }) => {
+  const value = rule(htmlDom)
+  return isString(value) && toLower(value.substring(0, 2))
+}
+
+module.exports = () => ({
+  lang: [
+    wrap($ => $('meta[property="og:locale"]').attr('content')),
+    wrap($ => $('html').attr('lang'))
+  ]
+})

packages/metascraper-lang/package.json

@@ -0,0 +1,34 @@
+{
+  "name": "metascraper-lang",
+  "description": "Get lang property from HTML markup",
+  "homepage": "https://metascraper.js.org",
+  "version": "0.0.0",
+  "main": "index.js",
+  "author": {
+    "email": "[email protected]",
+    "name": "Ian Storm Taylor"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/microlinkhq/metascraper/tree/master/packages/metascraper-lang"
+  },
+  "bugs": {
+    "url": "https://github.com/microlinkhq/metascraper/issues"
+  },
+  "dependencies": {
+    "lodash": "~4.17.4"
+  },
+  "devDependencies": {
+    "standard": "latest"
+  },
+  "engines": {
+    "node": ">= 8"
+  },
+  "files": [
+    "index.js"
+  ],
+  "scripts": {
+    "test": "exit 0"
+  },
+  "license": "MIT"
+}

packages/metascraper/package.json

@@ -64,6 +64,7 @@
     "metascraper-date": "^3.3.0",
     "metascraper-description": "^3.3.0",
     "metascraper-image": "^3.3.0",
+    "metascraper-lang": "^0.0.0",
     "metascraper-logo": "^3.3.0",
     "metascraper-publisher": "^3.3.0",
     "metascraper-title": "^3.3.0",

packages/metascraper/src/load-rules.js

@@ -22,7 +22,8 @@ const DEFAULT_RULES = [
   'metascraper-logo',
   'metascraper-publisher',
   'metascraper-title',
-  'metascraper-url'
+  'metascraper-url',
+  'metascraper-lang'
 ]
 
 let singletonConfig