🎨 Util to convert relative urls in absolute (#7264)

no issue

This PR takes the existing function `processUrls` in `data/xml/rss` and refactors it to be a stand-alone util.
The change is needed, as this functionality will be accessed from `apps/amp` to convert relative URLs.

core/server/data/xml/rss/index.js

@@ -1,15 +1,14 @@
-var _        = require('lodash'),
+var _           = require('lodash'),
-    cheerio  = require('cheerio'),
+    crypto      = require('crypto'),
-    crypto   = require('crypto'),
+    downsize    = require('downsize'),
-    downsize = require('downsize'),
+    RSS         = require('rss'),
-    RSS      = require('rss'),
+    config      = require('../../../config'),
-    url      = require('url'),
+    errors      = require('../../../errors'),
-    config   = require('../../../config'),
+    filters     = require('../../../filters'),
-    errors   = require('../../../errors'),
+    processUrls = require('../../../utils/make-absolute-urls'),
-    filters  = require('../../../filters'),
 
     // Really ugly temporary hack for location of things
-    fetchData = require('../../../controllers/frontend/fetch-data'),
+    fetchData   = require('../../../controllers/frontend/fetch-data'),
 
     generate,
     generateFeed,
@@ -65,48 +64,6 @@ function getBaseUrl(req, slugParam) {
     return baseUrl;
 }
 
-function processUrls(html, siteUrl, itemUrl) {
-    var htmlContent = cheerio.load(html, {decodeEntities: false});
-    // convert relative resource urls to absolute
-    ['href', 'src'].forEach(function forEach(attributeName) {
-        htmlContent('[' + attributeName + ']').each(function each(ix, el) {
-            var baseUrl,
-                attributeValue,
-                parsed;
-
-            el = htmlContent(el);
-
-            attributeValue = el.attr(attributeName);
-
-            // if URL is absolute move on to the next element
-            try {
-                parsed = url.parse(attributeValue);
-
-                if (parsed.protocol) {
-                    return;
-                }
-
-                // Do not convert protocol relative URLs
-                if (attributeValue.lastIndexOf('//', 0) === 0) {
-                    return;
-                }
-            } catch (e) {
-                return;
-            }
-
-            // compose an absolute URL
-
-            // if the relative URL begins with a '/' use the blog URL (including sub-directory)
-            // as the base URL, otherwise use the post's URL.
-            baseUrl = attributeValue[0] === '/' ? siteUrl : itemUrl;
-            attributeValue = config.urlJoin(baseUrl, attributeValue);
-            el.attr(attributeName, attributeValue);
-        });
-    });
-
-    return htmlContent;
-}
-
 getFeedXml = function getFeedXml(path, data) {
     var dataHash = crypto.createHash('md5').update(JSON.stringify(data)).digest('hex');
     if (!feedCache[path] || feedCache[path].hash !== dataHash) {

core/server/utils/make-absolute-urls.js

@@ -0,0 +1,57 @@
+var cheerio  = require('cheerio'),
+    url      = require('url'),
+    config   = require('../config');
+
+/**
+ * Make absolute URLs
+ * @param {string} html
+ * @param {string} siteUrl (blog URL)
+ * @param {string} itemUrl (URL of current context)
+ * @returns {object} htmlContent
+ * @description Takes html, blog url and item url and converts relative url into
+ * absolute urls. Returns an object. The html string can be accessed by calling `html()` on
+ * the variable that takes the result of this function
+ */
+function makeAbsoluteUrls(html, siteUrl, itemUrl) {
+    var htmlContent = cheerio.load(html, {decodeEntities: false});
+    // convert relative resource urls to absolute
+    ['href', 'src'].forEach(function forEach(attributeName) {
+        htmlContent('[' + attributeName + ']').each(function each(ix, el) {
+            var baseUrl,
+                attributeValue,
+                parsed;
+
+            el = htmlContent(el);
+
+            attributeValue = el.attr(attributeName);
+
+            // if URL is absolute move on to the next element
+            try {
+                parsed = url.parse(attributeValue);
+
+                if (parsed.protocol) {
+                    return;
+                }
+
+                // Do not convert protocol relative URLs
+                if (attributeValue.lastIndexOf('//', 0) === 0) {
+                    return;
+                }
+            } catch (e) {
+                return;
+            }
+
+            // compose an absolute URL
+
+            // if the relative URL begins with a '/' use the blog URL (including sub-directory)
+            // as the base URL, otherwise use the post's URL.
+            baseUrl = attributeValue[0] === '/' ? siteUrl : itemUrl;
+            attributeValue = config.urlJoin(baseUrl, attributeValue);
+            el.attr(attributeName, attributeValue);
+        });
+    });
+
+    return htmlContent;
+}
+
+module.exports = makeAbsoluteUrls;

core/test/unit/utils/make-absolute-urls_spec.js

@@ -0,0 +1,39 @@
+var makeAbsoluteUrls    = require('../../../server/utils/make-absolute-urls'),
+    configUtils         = require('../../utils/configUtils');
+
+describe('Make absolute URLs ', function () {
+    var siteUrl = 'http://my-ghost-blog.com',
+        itemUrl = 'my-awesome-post';
+    beforeEach(function () {
+        configUtils.set({url: 'http://my-ghost-blog.com'});
+    });
+
+    afterEach(function () {
+        configUtils.restore();
+    });
+
+    it('does not convert absolute URLs', function () {
+        var html = '<a href="http://my-ghost-blog.com/content/images" title="Absolute URL">',
+            result = makeAbsoluteUrls(html, siteUrl, itemUrl).html();
+
+        result.should.match(/<a href="http:\/\/my-ghost-blog.com\/content\/images" title="Absolute URL">/);
+    });
+    it('does not convert protocol relative `//` URLs', function () {
+        var html = '<a href="//my-ghost-blog.com/content/images" title="Absolute URL">',
+            result = makeAbsoluteUrls(html, siteUrl, itemUrl).html();
+
+        result.should.match(/<a href="\/\/my-ghost-blog.com\/content\/images" title="Absolute URL">/);
+    });
+    it('succesfully converts a relative URL', function () {
+        var html = '<a href="/about#nowhere" title="Relative URL">',
+            result = makeAbsoluteUrls(html, siteUrl, itemUrl).html();
+
+        result.should.match(/<a href="http:\/\/my-ghost-blog.com\/about#nowhere" title="Relative URL">/);
+    });
+    it('succesfully converts a relative URL including subdirectories', function () {
+        var html = '<a href="/about#nowhere" title="Relative URL">',
+            result = makeAbsoluteUrls(html, 'http://my-ghost-blog.com/blog', itemUrl).html();
+
+        result.should.match(/<a href="http:\/\/my-ghost-blog.com\/blog\/about#nowhere" title="Relative URL">/);
+    });
+});