fix: html entities in hrefs should not trigger noAnchorUrl

html-to-text · Dec 4, 2017 · 912536a · 912536a
1 parent 1dbebe1
commit 912536a
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 2 deletions.
diff --git a/lib/formatter.js b/lib/formatter.js
@@ -73,10 +73,10 @@ function formatAnchor(elem, fn, options) {
   if (!options.ignoreHref) {
     // Get the href, if present
     if (elem.attribs && elem.attribs.href) {
-      href = elem.attribs.href.replace(/^mailto\:/, '');
+      href = elem.attribs.href.replace(/^mailto:/, '');
     }
     if (href) {
-      if ((!options.noAnchorUrl) || (options.noAnchorUrl && href.indexOf('#') === -1)) {
+      if ((!options.noAnchorUrl) || (options.noAnchorUrl && href[0] !== '#')) {
         if (options.linkHrefBaseUrl && href.indexOf('/') === 0) {
           href = options.linkHrefBaseUrl + href;
         }

diff --git a/test/html-to-text.js b/test/html-to-text.js
@@ -195,6 +195,16 @@ describe('html-to-text', function() {
   });
 
   describe('a', function () {
+    it('should decode html attribute entities from href', function () {
+      var result = htmlToText.fromString('<a href="/foo?a&#x3D;b">test</a>');
+      expect(result).to.equal('test [/foo?a=b]');
+    });
+
+    it('should strip mailto: from email links', function () {
+      var result = htmlToText.fromString('<a href="mailto:foo@example.com">email me</a>');
+      expect(result).to.equal('email me [foo@example.com]');
+    });
+
     it('should return link with brackets', function () {
       var result = htmlToText.fromString('<a href="http://my.link">test</a>');
       expect(result).to.equal('test [http://my.link]');