refs #10318
If migrating from a previous version that does not include the setting
being migrated from we can safely not update the new setting, and just
rely on its default value being present. When rolling back we can use
defaults if the new setting does not exist.
refs https://github.com/TryGhost/Ghost/issues/10318
- Updates the type field for a setting to specify its type explicitly
- Type can be one of (string, number, boolean, array)
- `object` is allowed a temp type till all the object settings are removed
- existing `type` value for setting was misleading and referring to `group`
- we moved the current `type` value for a setting to `group` here - 4a9e57c170
refs #10318
As populateDefaults is run _before_ migrations, the new settings will
already be inserted in the database, so we just need to update their
values and then delete the old settings.
refs https://github.com/TryGhost/Ghost/issues/10318
`Settings.populateDefaults()` is run before migrations during Ghost's startup. This can cause problems when new settings table columns are added (and populated in `default-settings.json`) because `populateDefaults()` was using the model layer which assumes that those columns are available, resulting in `ER_BAD_FIELD_ERROR: Unknown column` type errors.
- query the database for the available `settings` table columns
- switch to using raw knex queries without Bookshelf for insertions so that we're in control of the columns that are added
- use `_.pick` to skip any properties in `default-settings.json` that do not match to an available column - those columns will be added and populated by later migrations
- moving away from using the model to insert settings has the side-effect of not emitting `settings.added/edited` and `settings.x.added/edited` events, this should be fine because `populateDefaults()` is called before anything else is set up and listening
- added a call to `populateDefaults()` in our knex-migrator "before migration" hook so that we have consistent db state across both startup initialised migrations and manually triggered knex migrations
no-issue
These utils are the first steps toward getting the models out of our
migrations! The utils here interact directly with the database and where
possible don't reinvent the wheel, by using smaller building blocks to
build more comples ones.
refs https://github.com/TryGhost/Ghost/issues/10318
- maps old `settings.type` values to new `settings.type/group` values
- uses an explicit map so that we don't lose information and can safely roll back even though we're modifying `settings.type` too
- updates `settings.type` values too to keep code working while we switch to using `settings.group`
- sets the `settings.group` value for all settings which are keeping the same group as their current type
- adapts `settings.type` validations to match new groups
- adds flags to specific settings, both in the migration for existing settings records and in default-settings.json for new settings records
refs https://github.com/TryGhost/Ghost/issues/10318
- `group`
- to replace the `type` column, provides a more descriptive name for the columns use
- for existing sites it will be populated by migrating data from the `type` column in a later migration
- for new sites a minimal update has been added to `parseDefaultSettings()` to populate the `group` field when settings are created during startup - fixes the NOT NULL constraint on `settings.group`
- `flags`
- signifies special handling that is different to other settings in a group
- eg, `PUBLIC,RO` would indicate that the setting is available via unauthenticated endpoints and is read-only
no issue
- Previous migration in commit - 2614565d5a - was using incorrect json structure for brand -> primary_color
- Fixes the up/down migrations to work on correct existing `brand: primary_color` structure
refs https://github.com/TryGhost/Ghost/issues/10318
- Adds following renames to settings table keys:
'default_locale' -> 'lang'
'active_timezone' -> 'timezone'
'ghost_head' -> 'codeinjection_head'
'ghost_foot' -> 'codeinjection_foot'
'brand.publicationColor' -> 'accent_color'
- The renames are done to match revised naming conventions and naming
exposed through APIs
- Supersedes this revert - 1eeb5a60b8 and #11946
refs https://github.com/TryGhost/Ghost/issues/10318
- Adds following renames to settings table keys:
'default_locale' -> 'lang'
'active_timezone' -> 'timezone'
'ghost_head' -> 'codeinjection_head'
'ghost_foot' -> 'codeinjection_foot'
'brand.publicationColor' -> 'accent_color'
- The renames are done to match revised naming conventions and naming
exposed through APIs
refs https://github.com/TryGhost/Ghost/issues/11414
We want to allow adding custom domains to member's from address as long as the new email is verified using the magic link flow and not saved directly to DB. Since we currently don't store domain as part of fromAddress, this PR -
- adds migration to update all existing fromAddress by appending site domain
closes https://github.com/TryGhost/Ghost/issues/11841
- Migration adds mapping between permissions and roles for email_preview send test mail
- Only owner previously had correct permission to send test emails
- Fixture existed to allow Admin/Editor/Integrations to send test mails but had missing migration
- Adds tests for roles to send test email
- Represents that logging is shared across all parts of Ghost at present
* moved core/server/lib/common/logging to core/shared/logging
* updated logging path for generic imports
* updated migration and schema imports of logging
* updated tests and index logging import
* 🔥 removed logging from common module
* fixed tests
* moved `server/config` to `shared/config`
* updated config import paths in server to use shared
* updated config import paths in frontend to use shared
* updated config import paths in test to use shared
* updated config import paths in root to use shared
* trigger regression tests
* of course the rebase broke tests
- All var declarations are now const or let as per ES6
- All comma-separated lists / chained declarations are now one declaration per line
- This is for clarity/readability but also made running the var-to-const/let switch smoother
- ESLint rules updated to match
How this was done:
- npm install -g jscodeshift
- git clone https://github.com/cpojer/js-codemod.git
- git clone git@github.com:TryGhost/Ghost.git shallow-ghost
- cd shallow-ghost
- jscodeshift -t ../js-codemod/transforms/unchain-variables.js . -v=2
- jscodeshift -t ../js-codemod/transforms/no-vars.js . -v=2
- yarn
- yarn test
- yarn lint / fix various lint errors (almost all indent) by opening files and saving in vscode
- grunt test-regression
- sorted!
no issue
- moved `mobiledoc.renderers.mobiledocHtmlRenderer` to `mobiledoc.mobiledocHtmlRenderer` so that it's easier for the getter to access the parent objects getters
- removed all tests and dependencies that now live in @tryghost/mobiledoc-dom-renderer
- kept the `mobiledocHtmlRenderer` test because that's testing that we've correctly wired up our cards and atoms and the output is what we expect
no issue
- the blank document we use in Ghost is not specific to the html renderer
- renamed from `structure` to `document` to better represent its intent
- allows for easier extraction of `mobiledocHtmlRenderer`
no issue
- Migrations within a minor have to be named with numbered prefixes like 01-, 02-, 03-.
- These two migrations were merged into master in the same time window which lead to having incorrect naming
no issue
- The flag has not been used and can be removed, to make the `members_subscription_settings` JSON record in `settings` table easier to read.
- It used to indicate Stripe configuration being present. Currently that is checked by looking up if Stripe config's `public_token` and `secret_token` values are present (example - https://github.com/TryGhost/Ghost/blob/3.11.0/core/frontend/helpers/ghost_head.js#L54)
refs https://github.com/TryGhost/Ghost/issues/11648
- Removes Stripe plan entries from settings that are not formatted correctly.
- Incorrect formatting was caused by a bug in 3.10.0 Admin-Client where it wasn't able to find complimentary plan. Related fix for this here - 9e7a6b801a
no issue
- Adds 'GET /members/:id/signin_urls' endpoint to Admin API allowing to fetch login URL for member. This URL allows to log in as a member which is useful in situations when you need to impersonate a member (for example to debug some issue they are having)
- Added member_signin_urls permission with migrations. Only the "Owner" user can read "signin_urls" resource. Admin and other users will be denied access
no issue
- 3.6.0 contained incorrect references in the `schema.js` file for the `members_label` table that was added in that version. On MySQL knex created a foreign key constraint for that reference which stopped member labels from being createable
- this fixes the schema file and has a migration to drop and recreate the table. Knex handles removal and addition of foreign keys during table drop/create
no issue
* Updated sendEmailWithMagicLink syntax
* Updated label name selection from theme
* Updated migration version for labels
* Added labels to export/import of members
* Added member labels sanitization for case-insensitive duplicates
* Fixed tests
* Fixed label serialization bug on import
* Bumped @tryghost/members-api to 0.15.0
* Fixed lint
* Cleanup
no issue
Since we added `email_subject` to `posts_meta` table in `3.1`, the migration tries to add `email_subject` column from post table, which does not exist and thus tries adding `undefined` value for column. Since sqlite expects default values while inserting new columns, this breaks any migration directly from `1.x`/`2.x` to 3.x.
The fix adds a default `null` value for any post_schema entry which doesn't has a value.
no issue
- The helper allows generating HTML needed to cancel or continue the member's subscription depending on subscription state.
- Added public members endpoint to allow updating subscription's `cancel_at_period_end` attribute available at: `PUT /api/canary/members/subscriptions/:id/`
- Added client-side hook to allow calling subscription cancellation. Allows to create elements with `data-members-cancel-subscription` / `data-members-continue-subscription` attributes which would call subscription update.
- Updated schema and added migration for `current_period_end` column
- As discussed we only add a single column to subscriptions table to avoid preoptimizing for future cases
- Added {{cancel_link}} helper
- Added error handling for {{cancel_link}} when members are disabled
- Added test coverage for {{cancel_link}} helper
- Bumped @tryghost/members-api version to 0.10.2. Needed to use `updateSubscription` middleware
- Bumped gscan to 3.2.0. Needed to recognize new {{cancel_link}} helper
no issue
- additional migration for the column added since the last 3.1 beta release to allow beta upgrades without rollbacks
- will be a no-op for upgrades from 3.0 as it's covered by `3.1/05-add-emails-table.js`
We want to allow admin users to trigger a retry of failed emails without having to go through the unpublish/republish dance.
- fixed resource identifier in email permissions migration so email permissions are added correctly
- added new email permissions migration so that beta releases can be upgraded without rollback (will be a no-op for any non-beta upgrades)
- added `/emails/:id/retry/` canary Admin API endpoint
- follows same URL pattern as theme activation
- only triggers mega service retry endpoint if the email has a `'failed'` status
refs https://github.com/TryGhost/Ghost/pull/11270
- Fixed 3.0/11-update-posts-html migration which failed in scenario when more than 999 posts with posts_meta relation were present
- The issue was originally spotted here: https://github.com/TryGhost/Ghost/pull/11270#issuecomment-546248308
- The main problem is in the `SELECT` statement which is generated for `findAll` method in Bookshelf which creates `WHERE IN(post_ids_here)` statement with all posts in the database
- Using knex directly as that's a preferred way to write migrations (does not depend on the model layer)
closes#11263
- Fixed `3.0/05-populate-posts-meta-table.js` migration failure when having >999 posts with metadata in the database
- The issue here is with hitting SQLite's internal SQLITE_LIMIT_VARIABLE_NUMBER limit when updating with a large amount of posts having metadata fields set (ref.: https://sqlite.org/limits.html#max_variable_number)
- Transforming migration to iterative method avoided inserting lots of records at once
no issue
We split `posts` table into 2 in v3 with a new `posts_meta` table. Since migrations always use the version of code which is being migrated to - in this case the Post model - which in v3 relies on the posts_meta table, `2.x` migrations relying on post model will fail as it doesn't exist in the expected state. This PR updates all 2.x migrations using `models.Post` to use knex queries directly to access database and perform operations.
no issue
We added 2 new member subscription settings - `allowSelfSignup` and `fromAddress`- with defaults as `true` and `noreply`, this migration sets default values for both settings for users migrating from previous version and cleans up intermediate naming for `allowSelfSignup`.
no issue
Since we removed subscribers code in v3, we cannot use `models.Subscribers` for migration, and instead switch to using db directly for fetching existing subscribers before migrating them to members.
refs https://github.com/TryGhost/Ghost/pull/11152
- Added subscribers table drop migration
- Removed subscribers from schema
- Removed subscribers controllers/routes/regression tests
- Removed subscriber related API code
- Removed subscribers from internal apps
- Removed subscriber importer
- Removed subscriber model
- Removed subscriber related permissions
- Removed webhook code related to subscribers
- When upgrading to v3 it is on the site admin to migrate all zapps or any other webhook clients to use members
- Removed subscriber-specific translation
- Removed subscriber lab flag
closes#11207
MySQL doesn't allow unqiue keys with a length of more than 191 when using InnoDB with utfmb4. These changes will ensure any incorrect tables created are fixed and have the
correct length for customer_id
* Changed `customer_id` to non-unique column
* Nooped the 2.32 `members_stripe_customers` migration
* Added migration to recreate `members_stripe_customers` table
* sqlite doesn't allow `ALTER TABLE` queries so this is the cleanest solution considering the table is not yet in use
no issue
- Populates members table with existing subscribers. Only takes into account columns we know already exist and need to be copied i.e `name`/`email`
no issue
- bumps `knex-migrator` so it supports irreversible migrations
- marks the `03-drop-client-auth` migration as irreversible because it destroys data that is not recoverable and is required for earlier versions of Ghost to function
no issue
- rollbacks have switched to using transactions but the migration code was copied from an old migration coded before that switch
- `down()` is no longer called with an object that contains a `connection` key, it has `transacting` instead
no issue
- `knex-migrator` will run migrations in the order that nodejs provides when running `fs.readDirSync` which in most cases is strict alphabetical
- 3.0 will shortly have more than 9 migrations which was resulting in the migration order being 1, 10, 2
- prefixing all single-digit migrations with `0` means that strict alphabetical ordering results in the expected order
refs https://github.com/TryGhost/Ghost/issues/10922
- adds migrations to...
1. add `post.type` column
2. populate `post.type` column based on `post.page` value
3. drop `post.page` column
- updates all code paths to work with `post.type` in place of `post.page`
- adds `nql-map-key-values` transformer for mapping `page`->`type` in `filter` params when using the v2 API
- modifies importer to handle `post.page`->`post.type` transformation when importing older export files
NOTE: The post metadata table split is purely an internal optimization for v3 and doesn't require or expect any external actions including related API usage in v3
We keep running into issues adding new fields to the post table because there are too many fields making the post table "too wide". We have also hit MySQL limitations in how many bytes can be in a row (64kb) with post table.
In v3, we decided to split the 8 post fields (meta, twitter and og) used for meta data into a posts_meta table as these 8 fields are all "problem" `varchar` fields and make sense logically grouped together. The API layer is unaffected by the split as input/output serializers ensure the data flow works the same way as it was in v2. Only thing to note is json export in v3 will have slightly different structure with posts meta fields as separate.
- Creates new post_meta schema/table with 8 fields (2 meta_* , 3 twitter_* and 3 og_*)
- Update relations between post and post_meta table
- Update input/output serializers to keep existing API behavior
- Avoids new entry in post_meta table for post where all meta fields are null
- Keeps the current fields API param behavior
- Handles migration of existing posts to new table structure
- Updates importer/exporter to work seamlessly with table changes
no issue
- drops now-unused `accesstokens`, `refreshtokens`, `clients`, and `client_trusted_domains` tables
- no rollback because the db schema for the tables no longer exists
no issue
- v0.1 is ☠️ so there's no longer any use of client auth
- removes all code related to `clients` and `client_trusted_domains`
- noops the "add backup client" migration in 1.7 because the referenced fixture no longer exists causing migrations and consequently all regression tests to fail
no issue
- Drops `ghost_auth_access_token` and `ghost_auth_id` fields since not used anymore
- Adds migration for dropping these columns from users table
- Drops Auth strategy - `ghostStrategy` - since its not used anymore
no issue
- we recently started wrapping rollbacks in transactions (https://github.com/TryGhost/knex-migrator/pull/161)
- in a number of migrations we were calling `model.destroy()` without passing through the options which includes the current transaction
- for models which are using `bookshelf-relations` this could result in an internal `SQLITE_BUSY: database is locked` error because it tries to run queries against tables that have been locked by previous queries in the transaction
- by passing through the options when calling `.destroy()` it allows the `bookshelf-relations` to re-use the same transction avoiding the database lock problems
