fix(main): remove HTML before content slug calc

index.js

@@ -6,6 +6,9 @@ var _ = require('lodash');
 var urlModule = require('url');
 var yaml = require('js-yaml');
 var strftime = require('strftime');
+var ent = require('ent');
+
+var htmlRegexp = /<[^>]+>/g;
 
 var Formatter = function () {};
 
@@ -46,7 +49,15 @@ Formatter.prototype._formatSlug = function (data) {
     return _.kebabCase(data.properties.slug[0]);
   }
 
-  var name = (data.properties.name || data.properties.content || [''])[0].trim();
+  var name;
+
+  if (data.properties.name) {
+    name = data.properties.name[0].trim();
+  }
+  if (!name && data.properties.content) {
+    name = data.properties.content[0].trim();
+    name = ent.decode(name.replace(htmlRegexp, ''));
+  }
 
   if (name) {
     name = name.split(/\s+/);

package.json

@@ -27,6 +27,7 @@
     "sinon": "^1.15.4"
   },
   "dependencies": {
+    "ent": "^2.2.0",
     "js-yaml": "^3.3.1",
     "lodash": "^3.10.0",
     "strftime": "^0.9.2"

test/formatter.spec.js

@@ -143,6 +143,14 @@ describe('Formatter', function () {
       formatter._formatSlug(baseMicroformatData).should.equal('hello-world');
     });
 
+    it('should ignore html-tags when basing slug on content', function () {
+      delete baseMicroformatData.properties.name;
+      baseMicroformatData.properties.content = ['<h1>Foo</h1> Bar &amp; <strong>Abc</strong>'];
+      // Test twice so that we don't get a non-reusable regexp!
+      formatter._formatSlug(baseMicroformatData).should.equal('foo-bar-abc');
+      formatter._formatSlug(baseMicroformatData).should.equal('foo-bar-abc');
+    });
+
     it('should ulimately fall back to publish time', function () {
       delete baseMicroformatData.properties.name;
       delete baseMicroformatData.properties.content;