Added Canonical URL support to posts&pages in Admin & Content API v2 (#10594)

refs #10593

- Added `canonical_url` field to post&pages resources in Admin & Content APIs
- Support for canonical URL on metadata layer (used in {{ghost_head}} helper)
- Made sure the new field is not accessible from API v0.1 
- Added handling same domain relative and absolute URLs

core/server/api/v2/utils/serializers/input/utils/url.js

@@ -1,13 +1,26 @@
 const _ = require('lodash');
 const {absoluteToRelative, getBlogUrl, STATIC_IMAGE_URL_PREFIX} = require('../../../../../../services/url/utils');
 
+const handleCanonicalUrl = (url) => {
+    const blogDomain = getBlogUrl().replace(/^http(s?):\/\//, '').replace(/\/$/, '');
+    const absolute = url.replace(/^http(s?):\/\//, '');
+
+    if (absolute.startsWith(blogDomain)) {
+        return absoluteToRelative(url, {withoutSubdirectory: true});
+    }
+
+    return url;
+};
+
 const handleImageUrl = (imageUrl) => {
     const blogDomain = getBlogUrl().replace(/^http(s?):\/\//, '').replace(/\/$/, '');
     const imageUrlAbsolute = imageUrl.replace(/^http(s?):\/\//, '');
     const imagePathRe = new RegExp(`^${blogDomain}/${STATIC_IMAGE_URL_PREFIX}`);
+
     if (imagePathRe.test(imageUrlAbsolute)) {
         return absoluteToRelative(imageUrl);
     }
+
     return imageUrl;
 };
 
@@ -45,6 +58,10 @@ const forPost = (attrs, options) => {
         attrs.twitter_image = handleImageUrl(attrs.twitter_image);
     }
 
+    if (attrs.canonical_url) {
+        attrs.canonical_url = handleCanonicalUrl(attrs.canonical_url);
+    }
+
     if (options && options.withRelated) {
         options.withRelated.forEach((relation) => {
             if (relation === 'tags' && attrs.tags) {

core/server/api/v2/utils/serializers/output/utils/mapper.js

@@ -1,3 +1,4 @@
+const _ = require('lodash');
 const utils = require('../../../index');
 const url = require('./url');
 const date = require('./date');
@@ -25,7 +26,11 @@ const mapTag = (model, frame) => {
 };
 
 const mapPost = (model, frame) => {
-    const jsonModel = model.toJSON(frame.options);
+    const extendedOptions = Object.assign(_.cloneDeep(frame.options), {
+        extraProperties: ['canonical_url']
+    });
+
+    const jsonModel = model.toJSON(extendedOptions);
 
     url.forPost(model.id, jsonModel, frame.options);
 

core/server/api/v2/utils/serializers/output/utils/url.js

@@ -16,6 +16,10 @@ const forPost = (id, attrs, options) => {
         attrs.twitter_image = urlService.utils.urlFor('image', {image: attrs.twitter_image}, true);
     }
 
+    if (attrs.canonical_url) {
+        attrs.canonical_url = urlService.utils.relativeToAbsolute(attrs.canonical_url);
+    }
+
     if (attrs.html) {
         const urlOptions = {
             assetsOnly: true

core/server/api/v2/utils/validators/input/schemas/pages.json

@@ -103,6 +103,11 @@
                     "type": ["string", "null"],
                     "maxLength": 100
                 },
+                "canonical_url": {
+                    "type": ["string", "null"],
+                    "format": "uri-reference",
+                    "maxLength": 2000
+                },
                 "authors": {
                     "$ref": "#/definitions/page-authors"
                 },

core/server/api/v2/utils/validators/input/schemas/posts.json

@@ -103,6 +103,11 @@
                     "type": ["string", "null"],
                     "maxLength": 100
                 },
+                "canonical_url": {
+                    "type": ["string", "null"],
+                    "format": "uri-reference",
+                    "maxLength": 2000
+                },
                 "authors": {
                     "$ref": "#/definitions/post-authors"
                 },

core/server/data/meta/canonical_url.js

@@ -1,8 +1,14 @@
-var urlService = require('../../services/url'),
-    getUrl = require('./url');
+const _ = require('lodash');
+const urlService = require('../../services/url');
+const getUrl = require('./url');
 
 function getCanonicalUrl(data) {
-    var url = urlService.utils.urlJoin(urlService.utils.urlFor('home', true), getUrl(data, false));
+    if ((_.includes(data.context, 'post') || _.includes(data.context, 'page'))
+        && data.post && data.post.canonical_url) {
+        return data.post.canonical_url;
+    }
+
+    let url = urlService.utils.urlJoin(urlService.utils.urlFor('home', true), getUrl(data, false));
 
     if (url.indexOf('/amp/')) {
         url = url.replace(/\/amp\/$/i, '/');

core/server/data/migrations/versions/2.17/2-posts-add-canonical-url.js

@@ -0,0 +1,39 @@
+const Promise = require('bluebird'),
+    common = require('../../../../lib/common'),
+    commands = require('../../../schema').commands,
+    table = 'posts',
+    columns = ['canonical_url'],
+    _private = {};
+
+_private.handle = function handle(options) {
+    let type = options.type,
+        isAdding = type === 'Adding',
+        operation = isAdding ? commands.addColumn : commands.dropColumn;
+
+    return function (options) {
+        let connection = options.connection;
+
+        return connection.schema.hasTable(table)
+            .then(function (exists) {
+                if (!exists) {
+                    return Promise.reject(new Error('Table does not exist!'));
+                }
+
+                return Promise.each(columns, function (column) {
+                    return connection.schema.hasColumn(table, column)
+                        .then(function (exists) {
+                            if (exists && isAdding || !exists && !isAdding) {
+                                common.logging.warn(`${type} column ${table}.${column}`);
+                                return Promise.resolve();
+                            }
+
+                            common.logging.info(`${type} column ${table}.${column}`);
+                            return operation(table, column, connection);
+                        });
+                });
+            });
+    };
+};
+
+module.exports.up = _private.handle({type: 'Adding'});
+module.exports.down = _private.handle({type: 'Dropping'});

core/server/data/schema/schema.js

@@ -56,7 +56,8 @@ module.exports = {
         twitter_image: {type: 'string', maxlength: 2000, nullable: true},
         twitter_title: {type: 'string', maxlength: 300, nullable: true},
         twitter_description: {type: 'string', maxlength: 500, nullable: true},
-        custom_template: {type: 'string', maxlength: 100, nullable: true}
+        custom_template: {type: 'string', maxlength: 100, nullable: true},
+        canonical_url: {type: 'text', maxlength: 2000, nullable: true}
     },
     users: {
         id: {type: 'string', maxlength: 24, nullable: false, primary: true},

core/server/models/post.js

@@ -572,6 +572,13 @@ Post = ghostBookshelf.Model.extend({
         // CASE: never expose the revisions
         delete attrs.mobiledoc_revisions;
 
+        // expose canonical_url only for API v2 calls
+        // NOTE: this can be removed when API v0.1 is dropped. A proper solution for field
+        //       differences on resources like this would be an introduction of API output schema
+        if (!_.get(unfilteredOptions, 'extraProperties', []).includes('canonical_url')) {
+            delete attrs.canonical_url;
+        }
+
         // If the current column settings allow it...
         if (!options.columns || (options.columns && options.columns.indexOf('primary_tag') > -1)) {
             // ... attach a computed property of primary_tag which is the first tag if it is public, else null

core/server/services/url/utils.js

@@ -460,11 +460,20 @@ function absoluteToRelative(urlToModify, options) {
     return relativePath;
 }
 
+function relativeToAbsolute(url) {
+    if (!url.startsWith('/') || url.startsWith('//')) {
+        return url;
+    }
+
+    return createUrl(url, true);
+}
+
 function deduplicateDoubleSlashes(url) {
     return url.replace(/\/\//g, '/');
 }
 
 module.exports.absoluteToRelative = absoluteToRelative;
+module.exports.relativeToAbsolute = relativeToAbsolute;
 module.exports.makeAbsoluteUrls = makeAbsoluteUrls;
 module.exports.getProtectedSlugs = getProtectedSlugs;
 module.exports.getSubdir = getSubdir;

core/test/regression/api/v0.1/utils.js

@@ -22,6 +22,7 @@ const expectedProperties = {
         .without('mobiledoc', 'plaintext')
         // swaps author_id to author, and always returns computed properties: url, comment_id, primary_tag, primary_author
         .without('author_id').concat('author', 'url', 'primary_tag', 'primary_author')
+        .without('canonical_url')
         .value(),
     user: {
         default: _(schema.users).keys().without('password').without('ghost_auth_access_token').value(),

core/test/regression/api/v2/admin/posts_spec.js

@@ -151,6 +151,32 @@ describe('Posts API', function () {
                 });
         });
 
+        it('canonical_url', function () {
+            return request
+                .get(localUtils.API.getApiQuery(`posts/${testUtils.DataGenerator.Content.posts[0].id}/`))
+                .set('Origin', config.get('url'))
+                .expect(200)
+                .then((res) => {
+                    return request
+                        .put(localUtils.API.getApiQuery('posts/' + testUtils.DataGenerator.Content.posts[0].id + '/'))
+                        .set('Origin', config.get('url'))
+                        .send({
+                            posts: [{
+                                canonical_url: `/canonical/url`,
+                                updated_at: res.body.posts[0].updated_at
+                            }]
+                        })
+                        .expect('Content-Type', /json/)
+                        .expect('Cache-Control', testUtils.cacheRules.private)
+                        .expect(200);
+                })
+                .then((res) => {
+                    should.exist(res.body.posts);
+                    should.exist(res.body.posts[0].canonical_url);
+                    res.body.posts[0].canonical_url.should.equal(`${config.get('url')}/canonical/url`);
+                });
+        });
+
         it('update dates & x_by', function () {
             const post = {
                 created_by: ObjectId.generate(),

core/test/unit/data/meta/canonical_url_spec.js

@@ -23,7 +23,7 @@ describe('getCanonicalUrl', function () {
         sinon.restore();
     });
 
-    it('should return canonical url', function () {
+    it('should return default canonical url', function () {
         const post = testUtils.DataGenerator.forKnex.createPost();
 
         getUrlStub.withArgs(post, false).returns('/post-url/');
@@ -36,6 +36,17 @@ describe('getCanonicalUrl', function () {
         getUrlStub.calledOnce.should.be.true();
     });
 
+    it('should return canonical url field if present', function () {
+        const post = testUtils.DataGenerator.forKnex.createPost({canonical_url: 'https://example.com/canonical'});
+
+        getCanonicalUrl({
+            context: ['post'],
+            post: post
+        }).should.eql('https://example.com/canonical');
+
+        getUrlStub.called.should.equal(false);
+    });
+
     it('should return canonical url for amp post without /amp/ in url', function () {
         const post = testUtils.DataGenerator.forKnex.createPost();
 

core/test/unit/data/schema/integrity_spec.js

@@ -19,7 +19,7 @@ var should = require('should'),
  */
 describe('DB version integrity', function () {
     // Only these variables should need updating
-    const currentSchemaHash = '7c5d34376392d01c274700350de228c1';
+    const currentSchemaHash = 'fda0398e93a74b2dc435cb4c026679ba';
     const currentFixturesHash = '42e15796b3c9bdcf0d0ec7eb66a1abf5';
 
     // If this test is failing, then it is likely a change has been made that requires a DB version bump,

core/test/unit/services/url/utils_spec.js

@@ -43,6 +43,30 @@ describe('Url', function () {
         });
     });
 
+    describe('relativeToAbsolute', function () {
+        it('default', function () {
+            configUtils.set('url', 'http://myblog.com/');
+            urlService.utils.relativeToAbsolute('/test/').should.eql('http://myblog.com/test/');
+        });
+
+        it('with subdir', function () {
+            configUtils.set('url', 'http://myblog.com/blog/');
+            urlService.utils.relativeToAbsolute('/test/').should.eql('http://myblog.com/blog/test/');
+        });
+
+        it('should not convert absolute url', function () {
+            urlService.utils.relativeToAbsolute('http://anotherblog.com/blog/').should.eql('http://anotherblog.com/blog/');
+        });
+
+        it('should not convert absolute url', function () {
+            urlService.utils.relativeToAbsolute('http://anotherblog.com/blog/').should.eql('http://anotherblog.com/blog/');
+        });
+
+        it('should not convert schemeless url', function () {
+            urlService.utils.relativeToAbsolute('//anotherblog.com/blog/').should.eql('//anotherblog.com/blog/');
+        });
+    });
+
     describe('getProtectedSlugs', function () {
         it('defaults', function () {
             urlService.utils.getProtectedSlugs().should.eql(['ghost', 'rss', 'amp']);