Merge pull request #48 from feelfreelinux/master

Add reading in of html "lang" property to html-metadata's parseGeneral method Bug: T148303
wikimedia · Dec 3, 2016 · 2fa47e7 · 2fa47e7
2 parents 8a97c42 + 3c9252d
commit 2fa47e7
Show file tree

Hide file tree

Showing 4 changed files with 39 additions and 3 deletions.
diff --git a/README.md b/README.md
@@ -103,6 +103,7 @@ The method parseGeneral obtains the following general metadata:
 <meta name ="robots" content="">
 <link rel="shortlink" href="">
 <title></title>
+<html lang="en">
 ```
 
 ## Tests

diff --git a/lib/index.js b/lib/index.js
@@ -288,6 +288,7 @@ exports.parseGeneral = BBPromise.method(function(chtml){
 		robots: chtml('meta[name=robots i]').first().attr('content'), //robots <meta name ="robots" content="">
 		shortlink: chtml('link[rel=shortlink i]').first().attr('href'), //short link <link rel="shortlink" href="">
 		title: chtml('title').first().text(), //title tag <title>
+		lang: chtml('html').first().attr('lang') //lang <html lang="">
 	};
 
 	// Copy key-value pairs with defined values to meta

diff --git a/test/lang.js b/test/lang.js
@@ -0,0 +1,34 @@
+'use strict';
+
+/**
+ * Tests using externally scraped websites. May fail if resource at
+ * location changes.
+ */
+
+var meta = require('../index');
+var assert = require('./utils/assert.js');
+
+// mocha defines to avoid JSHint breakage
+/* global describe, it, before, beforeEach, after, afterEach */
+
+describe('LangScraping', function() {
+
+	this.timeout(40000);
+
+	var url;
+
+	describe('Lang parameter from external website', function() {
+		it('should get hu parameter', function() {
+			var expected = "hu";
+			var options =  {
+    		url: "http://mno.hu/migr_1834/tellerlevel-cafolat-es-cafolat-700280",
+    		headers: {
+        	'User-Agent': 'webscraper'
+    		}
+			};
+			return meta(options, function(error, metadata){
+				assert.deepEqual(metadata.general.lang, expected);
+			});
+		});
+	});
+});
diff --git a/test/static.js b/test/static.js