From 6d675811bc25b972364ff0e194e942eb81ba3d5d Mon Sep 17 00:00:00 2001
From: Kevin Dew <kevin.dew@digital.cabinet-office.gov.uk>
Date: Sat, 10 Dec 2016 11:01:54 +0000
Subject: [PATCH] Only add schemes to URLs with hostnames

Fixes: https://github.com/yabwe/medium-editor/issues/1250

This changes the way a scheme is prefixed to a URL when linkValidation
is on. Previously any URL that did not match the scheme regex would be
prefixed with 'http://' which meant that an absolute path such as '/'
would be changed to 'http:///'

This changes it so that the 'http://' is only prefixed if the first part
of the path looks like a hostname, which allows absolute paths to be
unchanged and most relative paths to be unchanged.

I say most as a path such as "test.txt" would be matched as a hostname and
converted to 'http://test.txt' but this seems an acceptable trade off
and definitely better than trying to store a list of valid TLDs so
filenames and hostnames can be determined.
---
 spec/anchor.spec.js         | 57 +++++++++++++++++++++++++++++++++++++
 src/js/extensions/anchor.js | 29 ++++++++++++-------
 2 files changed, 76 insertions(+), 10 deletions(-)

diff --git a/spec/anchor.spec.js b/spec/anchor.spec.js
index fb62a9dcf..3a895c584 100644
--- a/spec/anchor.spec.js
+++ b/spec/anchor.spec.js
@@ -356,6 +356,63 @@ describe('Anchor Button TestCase', function () {
             expect(link.getAttribute('href')).toBe(validHashLink);
         });
 
+        it('should not add a scheme to an absolute path', function () {
+            var editor = this.newMediumEditor('.editor', {
+                    anchor: {
+                        linkValidation: true
+                    }
+                }),
+                absolutePath = '/test',
+                link,
+                anchorExtension = editor.getExtensionByName('anchor');
+
+            selectElementContentsAndFire(editor.elements[0]);
+            anchorExtension.showForm(absolutePath);
+            fireEvent(anchorExtension.getForm().querySelector('a.medium-editor-toolbar-save'), 'click');
+
+            link = editor.elements[0].querySelector('a');
+            expect(link).not.toBeNull();
+            expect(link.getAttribute('href')).toBe(absolutePath);
+        });
+
+        it('should not add a scheme to an obviously relative path', function () {
+            var editor = this.newMediumEditor('.editor', {
+                    anchor: {
+                        linkValidation: true
+                    }
+                }),
+                relativePath = 'test/file.html',
+                link,
+                anchorExtension = editor.getExtensionByName('anchor');
+
+            selectElementContentsAndFire(editor.elements[0]);
+            anchorExtension.showForm(relativePath);
+            fireEvent(anchorExtension.getForm().querySelector('a.medium-editor-toolbar-save'), 'click');
+
+            link = editor.elements[0].querySelector('a');
+            expect(link).not.toBeNull();
+            expect(link.getAttribute('href')).toBe(relativePath);
+        });
+
+        it('should add a scheme to a localhost url', function () {
+            var editor = this.newMediumEditor('.editor', {
+                    anchor: {
+                        linkValidation: true
+                    }
+                }),
+                localhostUrl = 'http://localhost',
+                link,
+                anchorExtension = editor.getExtensionByName('anchor');
+
+            selectElementContentsAndFire(editor.elements[0]);
+            anchorExtension.showForm('localhost');
+            fireEvent(anchorExtension.getForm().querySelector('a.medium-editor-toolbar-save'), 'click');
+
+            link = editor.elements[0].querySelector('a');
+            expect(link).not.toBeNull();
+            expect(link.getAttribute('href')).toBe(localhostUrl);
+        });
+
         it('should change spaces to %20 for a valid url if linkValidation option is set to true', function () {
             var editor = this.newMediumEditor('.editor', {
                     anchor: {
diff --git a/src/js/extensions/anchor.js b/src/js/extensions/anchor.js
index cfe25bea0..dd4388b3f 100644
--- a/src/js/extensions/anchor.js
+++ b/src/js/extensions/anchor.js
@@ -257,6 +257,8 @@
             // Matches common external protocols "mailto:" "tel:" "maps:"
             // Matches relative hash link, begins with "#"
             var urlSchemeRegex = /^([a-z]+:)?\/\/|^(mailto|tel|maps):|^\#/i,
+                hasScheme = urlSchemeRegex.test(value),
+                scheme = '',
                 // telRegex is a regex for checking if the string is a telephone number
                 telRegex = /^\+?\s?\(?(?:\d\s?\-?\)?){3,20}$/,
                 urlParts = value.match(/^(.*?)(?:\?(.*?))?(?:#(.*))?$/),
@@ -266,17 +268,24 @@
 
             if (telRegex.test(value)) {
                 return 'tel:' + value;
-            } else {
-                // Check for URL scheme and default to http:// if none found
-                return (urlSchemeRegex.test(value) ? '' : 'http://') +
-                    // Ensure path is encoded
-                    this.ensureEncodedUri(path) +
-                    // Ensure query is encoded
-                    (query === undefined ? '' : '?' + this.ensureEncodedQuery(query)) +
-                    // Include fragment unencoded as encodeUriComponent is too
-                    // heavy handed for the many characters allowed in a fragment
-                    (fragment === undefined ? '' : '#' + fragment);
             }
+
+            if (!hasScheme) {
+                var host = path.split('/')[0];
+                // if the host part of the path looks like a hostname
+                if (host.match(/.+(\.|:).+/) || host === 'localhost') {
+                    scheme = 'http://';
+                }
+            }
+
+            return scheme +
+                // Ensure path is encoded
+                this.ensureEncodedUri(path) +
+                // Ensure query is encoded
+                (query === undefined ? '' : '?' + this.ensureEncodedQuery(query)) +
+                // Include fragment unencoded as encodeUriComponent is too
+                // heavy handed for the many characters allowed in a fragment
+                (fragment === undefined ? '' : '#' + fragment);
         },
 
         doFormCancel: function () {