- as per https://github.com/bookshelf/bookshelf/wiki/Migrating-from-0.15.1-to-1.0.0#default-to-require-true-on-modelfetch-and-collectionfetchone, models will now default to `{require:true}` during a fetch, which changes how Bookshelf will respond when a models yields no results
- instead of passing a `null` result, it will reject with an error, so we'd need to switch to `.catch`ing everything
- our code is set up to handle all these null results and switching style is not currently on the cards so we want to use the existing behaviour for now
- to enable this, the `requireFetch` option needs to be added to the model definitions
no-issue
Writing code outside of Ghost which deals with the models is currently
done by passing the models which are needed to the external module,
rather than the instance of ghostBookshelf. This does not give us a way
to create transaction to run queries in. This method is designed as a
simple way to give all models the power to create a transaction for
themselves.
This will be used in @tryghost/members-api for example to ensure that
failures in communication with the Stripe API will rollback the related
inserts in the database.
refs https://github.com/TryGhost/Team/issues/946
This refactor pulls out the core logic so that we can easily add other
bulk operations without having to duplicate even more logic.
It also gives a consistent return value between bulk operations, renaming
`unsuccessfulIds` and `unsuccessfulRecords` to `unsuccessfulData`
We also add a bulkEdit method which will be used to bulk unsubscribe members
from the newsletter.
- This is a precursor to trying to split apart into:
- model events + webhooks system which makes sense
- frontend events which should be independent or removed
- maybe some concept of a settings manager that we can use in various places to bind logic 🤔
- other usages of events that should be refactored to not use events
refs 8a1fd1f57f
refs 5584430ddc
- The change to async/await in the original commit 558443 was causing problems in downstream dependencies (create-error package) where it was loosing a context of "this". It's not a direct dependency so I didn't go yak shaving into where exacly the context is lost.
- The fix to keep a correct context of "this" was sticking to an existing pattern using regular function returning promises. Once we need to redo them into async/await we can investigate if there's a way around create-error's context prolbem
closes https://github.com/TryGhost/Team/issues/817
refs 6d083ee00e/packages/bookshelf-pagination/lib/bookshelf-pagination.js (L256)
- The 500 error is not the best we can do in this situation and throwing a 400 just like we doo in a referenced commit would keep the convention
- The underlying problem of the bug is bigger - we allow the fields named the same way as relations to leak into the db query and that causes an incorrect SQL syntax. It's a bigger problem which would need a separate, holistic approach
refs 188de00489
- this fix was incorrect - the function should have been on the
prototype but I'd moved it over incorrectly into the static class functions
- this commit moves `defaultColumnsToFetch` to the prototype functions
and reverts the referenced commit back to `this.prototype` as expected
- this wasn't including the custom columns from the `post` model, which
was causing tests to fail
- pro tip: run tests!
refs a457631a20
- `defaultColumnsToFetch` was moved to the CRUD plugin in the referenced
commit, which makes it a function on `this` instead of `this.prototype`
- this means the function doesn't exist and Admin throws an error when
you start typing in the search bar because the API 500s
- this commit switches it to `this.defaultColumnsToFetch()`
refs https://github.com/TryGhost/Team/issues/808
- see referenced issue for context, but turning this function into
async-await seems to have broken error handling when deleting things
that don't exist
- i'm really not sure why - but my running theory is that it's something
to do with Bluebird Promises vs native Promises
- this should keep the same functionality until I can investigate what
is going on
no issue
- `setId` is only used within the `events` plugin and it makes sense to
keep code together
- we don't lose anything by putting it here, but it should make it
easier to test in the future
no issue
- `formatOnWrite` doesn't override anything in Bookshelf but we use it
within the `override` plugin and sub-models may override it, so it's
easier to keep these things together
no issue
- we were only importing the `db` to access the `knex` instance, but
we can get this through the Bookshelf instance
- switches to pulling out `knex` from Bookshelf so we can remove the
remaining local require
no issue
- this commit extracts code relating to bulk DB operations into a
separate plugin
- it __could__ go into the CRUD one but these operations are a little
more involved
no issue
- this commit extracts event related code from the Base model into a plugin
- in particular:
- events initialization
- the `on*` events
- `emitChange` - I'm not sure about this one but it __is__ event
related
no issue
- this commit extracts code related to Actions from the Base model into
a separate plugin
- `api-key.js` contained the exact same helper function as the Base
model so that has been de-duplicated
no issue
- I'm working on pulling apart the base index.js and this code is
specific to setting up Bookshelf + the plugins, which is pretty
contained and can stay in one file
- it only has one local require so it might be a good candidate for
extracting out of Ghost in the future
no issue
- we're going to pull this out into the framework monorepo but
refactoring it here first makes it a lot easier to extract without
losing the history
no issue
The only pieces of Ghost-Ignition used in Ghost were debug and
logging. Both of these modules have been superceded by the Framework
monorepo, and all usages of Ignition have now been removed, replaced
with @tryghost/debug and @tryghost/logging.
no issue
- we're going to pull this out into the `framework` monorepo but
refactoring it here first makes it a lot easier to extract without
losing the history
- note: this is very temporary and will be extracted soon
refs:
- cf15f60085
- dd20cc649b
- ccf27f7009
- abf146d61f
- 2b54c92a14
- bb029a53f6
- 95bd7ee675
- 9018b4df22
- df01a6e5f4
- d313726b34
- these plugins were in a state where they were independent enough to be
pulled out into their own packages, which is what we did in the
referenced commits above
- each package is named like `@tryghost/bookshelf-<plugin>`
- to avoid requiring multiple packages into Ghost, we've also created a
wrapper package called `@tryghost/bookshelf-plugins` which re-exports
all these plugins, so the changes in Ghost are very simple - dbebdd43b5
- this commit deletes the plugins + tests, and replaces with our new
package with some minor code changes
- This is a really specific piece of code related to validating models against our internal schema.js format
- This doesn't make sense without a schema.js file
- It does depend on the internal validator and validate tools - but those are used elsewhere too, and can reasonably be moved out of the codebase
- I don't see schema.js moving out of the codebase any time soon. We can move the validator but it would be a class that requires schema via DI
- For now my focus is on getting the data/validation tooling separated and making clear sense
- Improving data/schema can come later :)
refs d783a8d2d4
- we're removing i18n from Ghost core because it no longer meets our
needs
- this switches out i18n in the base Bookshelf model for our
`tryghost/tpl` package with a `messages` object of strings sprinkled
through the code
refs 829e8ed010
- i18n is used everywhere but only requires shared or external packages, therefore it's a good candidate for living in shared
- this reduces invalid requires across frontend and server, and lets us use it everywhere until we come up with a better option
- Having these as destructured from the same package is hindering refactoring now
- Events should really only ever be used server-side
- i18n should be a shared module for now so it can be used everywhere until we figure out something better
- Having them seperate also allows us to lint them properly
refs c873899e49
- as of `bson-objectid` v2.0.0, this library exports the function
to generate an ObjectID directly, and then you need to use `.toHexString()`
to get the 24 character hex string - 6696f27d82
- this commit removes all uses of `.generate()` and replaces with this
change
closes https://github.com/TryGhost/Ghost/issues/12791
closes https://github.com/TryGhost/Team/issues/566https://github.com/TryGhost/Ghost/pull/12787 introduced a significant performance regression due to a misunderstanding of when Bookshelf calls `.format()` ([related upstream issue](https://github.com/bookshelf/bookshelf/issues/668)). We expected `.format()` to only be called on save but it's also called when Bookshelf performs fetching and eager loading which happens frequently. `.format()` can be a heavy method as it needs to parse and serialize html and markdown so it should be performed as infrequently as possible.
- override `sync()` in the base model so we can call our own `.formatOnWrite()` method to transform attributes on `update` and `insert` operations
- this was the only feasible location in Bookshelf I could find that is low enough level to not require modifying model instance attributes
- gives models the option to perform heavy transform operations only when writing to the database compared to the usual `.format()` method that is also called on fetch in many situations
refs https://github.com/TryGhost/Ghost/issues/12567
- Introduced here @@UNIQUE_CONSTRAINTS@@ notation allows to create unique contraints over multiple database fields. This will be needed to change posts' table unique constraint from `slug` to `slug+type`.
- The notation is equivalent to SQL's: UNIQUE(column_name1,column_name2)
- Example use in schema:
posts: {
slug: {type: 'string', maxlength: 191, nullable: false},
type: {type: 'string', maxlength: 50, nullable: false, defaultTo: 'post', validations: {isIn: [['post', 'page']]}},
'@@UNIQUES@@': [
['slug', 'type']
]
}
refs c1d66f0b01
- fixed base model allowing '@@INDEXES@@' as a permitted attribute/order
- fixed base model automatically setting `@@INDEXES@@` to null on the model when creating
- added `doAuth('members:emails')`
- creates an `email_batch` record attached to the first email in the fixtures
- creates an `email_recipients` record for each member
- runs analytics aggregation so the email and member counts are as expected
- added acceptance test for `/member/:id/?include=email_recipients`
refs https://github.com/TryGhost/Ghost/issues/12420
- updated `order` bookshelf plugin's `parseOrderOption()` method to return multiple order-related properties
- `order` same as before, a key-value object of property-direction
- `orderRaw` new property that is a raw SQL order string generated from `orderRawQuery()` method in models
- `eagerLoad` new property that is an array of properties the `eagerLoad` plugin should use to join across
- updated `pagination.fetchAll()` to apply normal order + raw order if both are available and to handle eager loading / joins when `options.eagerLoad` is populated
- updated post model to include details for email relationship and to add `orderRawQuery()` that allows `email.open_rate` to be used as an order option
requires https://github.com/TryGhost/Ghost/pull/12192
- added initial `EmailBatch` and `EmailRecipient` model definitions with defaults and relationships
- added missing `post` relationship function to email model
- fetch member list without bookshelf
- bookshelf can add around 3x overhead when fetching the members list for an email
- we don't need full members at this point, only having the data is fine
- if we need full models later on we can push the model hydration into background jobs where recipient batches are fetched ready for an email to be sent
- bookshelf model instantiation of many models blocks the event loop, using knex directly keeps concurrent requests fast
- adds `getFilteredCollectionQuery` method to base model to facilitate getting a knex query based on our normal model filters along with transaction/forUpdate applied
- store recipient list before sending email
- chunk already-fetched members list into batches and insert records into the `email_recipients` table via knex
- chunked into batches of 1000 to match the number of emails that Mailgun accepts in a single API request but this may not be the absolute fastest batch size for recipient insertion:
| Batch size | Batch time | Total time |
| ---------- | ---------- | ---------- |
| 500 | 20ms | 4142ms |
| 1000 | 50ms | 4651ms |
| 5000 | 170ms | 3540ms |
| 10000 | 370ms | 3684ms |
- create an email_batch record before inserting recipient rows so we can effeciently fetch recipients by batch and store the overall batch status
refs #11729
- When ordering is done by fields from a relation (like post's `meta_title` that comes form `posts_meta` table), Bookshelf does not include those relations in the original query which caused errors. To support this usecase added a mechanism to detect fields from a relation and load those relations into query.
- Extended ordering to include table name in ordered field name. The information about the table name is needed to avoid using `tableName` within pagination plugin and gives path to having other than original table ordering fields (e.g. order by posts_meta table fields)
- Added test case to check ordering on posts_meta fields
- Added support for "eager loading" relations. Allows to extend query builder object with joins to related tables,
which could be used in ordering (possibly in filtering later). Bookshelf does not support ordering/filtering by proprieties coming from relations, that's why this kind of plugin and query expansion is needed
- Added note about lack of support for child relations with same property names.
