From d96be4907ee659e38446b9e033a2089fb9f4ccf3 Mon Sep 17 00:00:00 2001
From: Kevin Ansfield <kevin@ghost.org>
Date: Mon, 5 Aug 2019 13:56:28 +0100
Subject: [PATCH] Fixed relative canonical_url values not being stored as
 root-relative (#10989)

no issue

- we try to store all urls as relative paths where possible in Ghost so that the `config.url` value can be changed
- all relative paths are stored as root-relative except for the `post.canonical_url` field which was storing subdirectory-relative paths
- adds a migration to put the subdirectory prefix onto any relative canonical_url paths
- updates the canonical_url input serialiser to keep the subdirectory rather than stripping it to match all other url fields
---
 .../v2/utils/serializers/input/utils/url.js   |  2 +-
 ...subdirectory-to-relative-canonical-urls.js | 98 +++++++++++++++++++
 2 files changed, 99 insertions(+), 1 deletion(-)
 create mode 100644 core/server/data/migrations/versions/2.27/3-add-subdirectory-to-relative-canonical-urls.js

diff --git a/core/server/api/v2/utils/serializers/input/utils/url.js b/core/server/api/v2/utils/serializers/input/utils/url.js
index 31c27bfb0a..a05360b2c7 100644
--- a/core/server/api/v2/utils/serializers/input/utils/url.js
+++ b/core/server/api/v2/utils/serializers/input/utils/url.js
@@ -12,7 +12,7 @@ const handleCanonicalUrl = (canonicalUrl) => {
     // Blog URL incl. the same protocol. This allows users to keep e.g. Facebook comments after
     // a http -> https switch
     if (absolute.startsWith(blogDomain) && isSameProtocol) {
-        return urlUtils.absoluteToRelative(canonicalUrl, {withoutSubdirectory: true});
+        return urlUtils.absoluteToRelative(canonicalUrl);
     }
 
     return canonicalUrl;
diff --git a/core/server/data/migrations/versions/2.27/3-add-subdirectory-to-relative-canonical-urls.js b/core/server/data/migrations/versions/2.27/3-add-subdirectory-to-relative-canonical-urls.js
new file mode 100644
index 0000000000..7dcfce7195
--- /dev/null
+++ b/core/server/data/migrations/versions/2.27/3-add-subdirectory-to-relative-canonical-urls.js
@@ -0,0 +1,98 @@
+const models = require('../../../../models');
+const common = require('../../../../lib/common');
+const config = require('../../../../config');
+const {URL} = require('url');
+
+module.exports.config = {
+    transaction: true
+};
+
+// We've been incorrectly saving canonical_url fields without a subdirectory.
+// We need this column to be consistent with all other relative URLs, so...
+
+// if we have a configured url with a subdirectory
+// find all posts that have a canonical_url starting with / but not //
+// prefix the subdirectory to the canonical_url and save the post
+module.exports.up = (options) => {
+    // normalize config url to always have a trailing-slash
+    let configUrl = config.get('url');
+    if (!configUrl.endsWith('/')) {
+        configUrl = `${configUrl}/`;
+    }
+    const url = new URL(configUrl);
+
+    const localOptions = Object.assign({
+        context: {internal: true}
+    }, options);
+
+    if (url.pathname === '/') {
+        common.logging.info('Skipping posts.canonical_url subdirectory fix: no subdirectory configured');
+        return Promise.resolve();
+    }
+
+    // perform a specific query for the type of canonical URLs we're looking for
+    // so we're not fetching and manually looping over a ton of post models
+    return models.Posts
+        .forge()
+        .query((qb) => {
+            qb.where('canonical_url', 'like', '/%');
+            qb.whereNot('canonical_url', 'like', '//%');
+        })
+        .fetch(localOptions)
+        .then((posts) => {
+            if (posts) {
+                return Promise.mapSeries(posts, (post) => {
+                    const canonicalUrl = post.get('canonical_url').replace('/', url.pathname);
+                    post.set('canonical_url', canonicalUrl);
+                    return post.save(null, localOptions);
+                }).then(() => {
+                    common.logging.info(`Added subdirectory prefix to canonical_url in ${posts.length} posts`);
+                });
+            }
+
+            common.logging.info('Skipping posts.canonical_url subdirectory fix: no canonical_urls to fix');
+            return Promise.resolve();
+        });
+};
+
+// if we have a configured url with a subdirectory
+// find all posts with a canonical_url starting with the subdirectory
+// remove it and save the post
+module.exports.down = (options) => {
+    // normalize config url to always have a trailing-slash
+    let configUrl = config.get('url');
+    if (!configUrl.endsWith('/')) {
+        configUrl = `${configUrl}/`;
+    }
+    const url = new URL(configUrl);
+
+    const localOptions = Object.assign({
+        context: {internal: true}
+    }, options);
+
+    if (url.pathname === '/') {
+        common.logging.info('Skipping posts.canonical_url subdirectory fix: no subdirectory configured');
+        return Promise.resolve();
+    }
+
+    return models.Posts
+        .forge()
+        .query((qb) => {
+            qb.where('canonical_url', 'LIKE', `${url.pathname}%`);
+        })
+        .fetch()
+        .then((posts) => {
+            if (posts) {
+                return Promise.mapSeries(posts, (post) => {
+                    const canonicalUrl = post.get('canonical_url').replace(url.pathname, '/');
+                    post.set('canonical_url', canonicalUrl);
+                    return post.save(null, localOptions);
+                }).then(() => {
+                    common.logging.info(`Removed subdirectory prefix from canonical_url in ${posts.length} posts`);
+                });
+            }
+
+            common.logging.info('Skipping posts.canonical_url subdirectory fix: no canonical_urls to fix');
+            return Promise.resolve();
+        });
+};