refs https://github.com/TryGhost/Team/issues/1728
- previously, we allowed a member to be mapped to multiple tiers simultaneously as an edge case, in case they managed to signup via another subscription
- since this was always an edge case and not supported, to simplify the flows going forward now that complimentary members can also upgrade, in case of an active subscription we'll always just attach the associated tier to member and remove all other tiers mapped to it
- this endpoint returns the Ghost version, of which the minor just hit
double digits
- because of this, the content-length size changed, and the snapshot was
incorrect
- we've previously allowed overrides for the content-length to be any number (see
1fadbacdec)
- this commit allows the header to be any number so it doesn't fail when
the Ghost version is incremented
closes: https://github.com/TryGhost/Ghost/issues/15252
- comments are deleted when posts are deleted. Without cascade delete on parent_id, replies cannot be deleted
- this change means that deleting a post will delete all comments and replies without error
refs: https://github.com/TryGhost/Ghost/issues/15252
- all columns with a foreign key (references prop) must have a deletion strategy
- we just found a bug with this in the comments table - see referenced issue
- this fix adjusts the schema and migration for this change before its released so we don't have to write a horrible migration later
closes: https://github.com/TryGhost/Ghost/issues/15251
- getAdminUrl is a config helper that returns the direct value from config
- change to using urlJoin and urlFor helpers to generate the correct URL
- not entirely happy with this fix as there's no test, but I've verified it locally
refs https://github.com/TryGhost/Team/issues/1727
- adds alpha feature flag for expiring comped subs
- adds duration to complimentary subs to set expiry
- shows details for expiring comped subs
refs https://github.com/TryGhost/Team/issues/1727
- runs a daily cron job at start of the day to cleanup all expired comped subs
- removes `members<>products` mapping for expired entries, and updates status for corresponding members
- also adds status events for members going back from comp -> free as a result of expiry
- scope for future optimisation on how the scheduled job is ran or does the cleanup
refs https://github.com/TryGhost/Team/issues/1727
- if feature flag is enabled, handles storing expiry date on complimentary subscriptions in `expiry_at` column of `members_products`
- updates the expiry value on both member edit or add with tiers
- expiry is passed as `expiry_at` in `tiers` list of a member
- includes `expiry_at` on tiers data of a member when flag is enabled
- if a model is being deleted, `attributes` is an empty object and the
data we actually need is in `_previousAttributes`
- because of this, only fetching the type using `.get` returned the
wrong value, and the Action model validator would throw an error
because we tried to insert an empty type
- we can access the previous value using `.previous(..)`
- this commit fixes saving actions when deleting a post by fetching the
type from the previous attributes if the current attributes is empty
refs: 22fd7f289c
- in the mentioned commit I changed the tests so that we don't need to update snapshots for every labs flag change
- this commit does the same for content-length which didn't get picked up locally, but does on CI for some reason
- the goal is to allow the team to add and remove flags without needing to update a random snapshot
- it turns out we're running the members migration job in tests, and
these run every time we boot Ghost. Given we wipe the DB each time,
this forces the job to run, which is just burning valuable test time
- the reason this block of code is slow is because it waits 500ms to see
if the job has completed
- we run this 55 times, as of writing, during the E2E tests, so that's
over 27s of idle time
- this commit gates running the migrations to outside of the test environment
refs https://github.com/TryGhost/Team/issues/1822
Exposing the values through the API is restricted behind the alpha flag.
We're exposing the values by default when the flag is enabled for now,
but can reconsider that later.
- during a refactor, I moved the `BATCH_SIZE` variable around
- putting the variable export above a more general export means it get
overwritten and the value is `undefined` outside of the module
- when we chunk the emails, we were chunking in sized of `undefined`,
so I'm guessing it just defaulted to 1
- this means the email batches were of size 1 instead of 1000 - oops
refs https://github.com/TryGhost/Team/issues/1808
refs https://github.com/TryGhost/Team/issues/1809
refs https://github.com/TryGhost/Team/issues/1820
refs https://github.com/TryGhost/Team/issues/1814
### Changes in `member-events` package
- Added MemberCreatedEvent (event, not model)
- Added SubscriptionCreatedEvent (event, not model)
### Added `member-attribution` package (new)
- Added the AttributionBuilder class which is able to convert a url history to an attribution object (exposed as getAttribution on the service itself, which handles the dependencies)
```
[{
"path": "/",
"time": 123
}]
```
to
```
{
"url": "/",
"id": null,
"type": "url"
}
```
- event handler listens for MemberCreatedEvent and SubscriptionCreatedEvent and creates the corresponding models in the database.
### Changes in `members-api` package
- Added urlHistory to `sendMagicLink` endpoint body + convert the urlHistory to an attribution object that is stored in the tokenData of the magic link (sent by Portal in this PR: https://github.com/TryGhost/Portal/pull/256).
- Added urlHistory to `createCheckoutSession` endpoint + convert the urlHistory to attribution keys that are saved in the Stripe Session metadata (sent by Portal in this PR: https://github.com/TryGhost/Portal/pull/256).
- Added attribution data property to member repository's create method (when a member is created)
- Dispatch MemberCreatedEvent with attribution
### Changes in `members-stripe-service` package (`ghost/stripe`)
- Dispatch SubscriptionCreatedEvent in WebhookController on subscription checkout (with attribution from session metadata)
refs https://github.com/TryGhost/Toolbox/issues/356
- even though I was hiding the button, it's better to prevent going to
the route entirely by redirecting to the homepage if the labs flag is
not enabled
- this was due to the fact that we use the same model for pages as we do
for posts, so the hardcoded `post` key was not accurate
- this commit adds support for providing a function to return the key for the
action type
refs https://github.com/TryGhost/Toolbox/issues/356
- some resources don't have titles, like users, so we should optionally
see if they have a name
- this is starting to become unwieldy so I'll likely pull this logic out
in an upcoming commit
refs https://github.com/TryGhost/Toolbox/issues/356
- this allows you to click on the title and go to whatever resource was
acted upon
- if the resource has not be found, it'll just show the resource ID
- cleaned up unused dependencies
- adds missing dependencies that are used in the code
- this should help us be more explicit about the dependencies a package
uses
closes https://github.com/TryGhost/Team/issues/1715
refs https://github.com/TryGhost/Members/pull/407
- when comped boolean was brought back for backwards compatibility, it accidentally opened a path from admin to set comped stripe subscriptions on members on save, as Admin always sends back `comped` value
- since we don't allow stripe comped subscriptions anymore, this caused weird behavior with assigning the default tier on comped stripe subscription.
- this change removes the `json.comped` value to be sent up from Admin to avoid any weird comped changes to the member
no issue
- switched `parse-audit-log-event` to a class helper to get access to dependency injection
- added `get actor()` to the parsed event object
- uses the store to find the related user record, returning an already loaded record if it exists otherwise fetches the record
- allows use of `event.actor.*` in templates because they are promise-aware and will be filled in once the record is loaded
refs. https://github.com/TryGhost/Team/issues/1813
- All the groups of the Portal settings accordion was closed by default. Opening the first group when showing the modal makes the workflow faster.
- The top heading on the content area had a white background which made the screen visually a bit cluttered.
refs. https://github.com/TryGhost/Team/issues/1812
- The "Copy link" button on the Offer detail screen was only enabled once certain fields were filled. This makes unnecessary cognitive load on the end user.
refs https://github.com/TryGhost/Toolbox/issues/356
- we had a function called `getAction` in every model where we were
collecting CRUD actions to store in the DB
- this function has the same boilerplate code - make sure it's not
internal and then construct the object to return
- as we add more actions to more models, we probably want to pull this
out and just configure the things specific to the model
- this commit pulls out the function into the actions plugin and adds a
couple of keys to the models to indicate we'd like to store CRUD
actions, along with the model name
refs https://github.com/TryGhost/Toolbox/issues/363
- this commit switches us to using the official and maintained
`mailgun.js` SDK, and updates the `mailgun-client` code to reflect the
changes between the two
- some of the existing tests return `items` as an empty array
- the upcoming change to switch `mailgun-js` to `mailgun.js` means it's
more strict about requiring `paging` too
- this commit adds a new empty-response fixture so we can standardize
using that across tests
no issue
- we allowed line breaks to be created in the text-replacement html input used inside email cards but they were removed as soon as the card's edit mode was left
- fixed the clean-text-replacement-html routine so `<br>` atoms aren't stripped
- added the soft-return parser to the `<KoenigTextReplacementHtmlInput>` editor options so the `<br>` elements are correctly parsed back into atoms when entering edit mode
refs https://github.com/TryGhost/Team/issues/1804
- Adds a script that is only injected when the member attribution alpha flag is enabled
- This script builds a history and saves it in localStorage as `ghost-history` that contains something like this:
``` json
[
{
"time": 1660650730,
"path": "/about/"
},
{
"time": 1660651730,
"path": "/welcome/"
}
]
```
- Keeps track of the time of every page visit, so we can correctly remove old items. I also considered saving the time separately and clearing the whole history when the saved time is older than 24h, but that would have the side effect that items older than 24h might leak into the history if you visit every 12 hours (to give an example). Plus, having objects in the history might make it easier to add other attributes to the items if we ever want to do that in the future. We also have access to the time between visits.
- Added `.eslintrc` configuration for this new frontend script. This makes it easier to spot errors when developing, and follow the same syntax rules as other scripts. In the future it can allow us to require an older ECMA version in the browser script. If we like this pattern, we could also use it for other frontend scripts.
refs https://github.com/TryGhost/Toolbox/issues/364
refs 147ec91162
- This looks like a subtle bug that has gone unnoticed for years. Have checked if we rely on the logic anywhere (mostly used in image-dimensions frontend helper) - we don't access the "url" directly.
- There is no reasoning attached behind why the cached size was stored as a url (see refed commit)
- WHY is this even being fixed? Caches can store anything... does not mean we should! Inconsistent data becomes a real PITA if the cache is persisted and is hard to repopulate (e.g. to migrate the cached data format).
no refs.
- There was a bug on the tier cards action menu dropdown in Settings/Membership that the buttons in the dropdown were clickable even when it was closed.
refs. https://github.com/TryGhost/Team/issues/1806
- free trial indicator - which helps users identify tiers easier - was missing on tier cards in Settings / Membership in the Admin
refs https://github.com/TryGhost/Toolbox/issues/364
- The "new Map()" cache was a "hidden cache" that did not follow any specific pattern. Following the cache adapter pattern here makes it possible swapping out the cache for alternative implementations - e.g. Redis storage
refs https://github.com/TryGhost/Toolbox/issues/364
- Before introducing an image size caching mechanism we need to make existing caches explicit - makes sure caches are operating in multiple different instances.
- Explicit syntax also allows to pass in different configurations for each cache instance (e.g. use different Redis connections for caches with different purposes)
refs https://github.com/TryGhost/Toolbox/issues/364
- The InMemoryCache is an implementation of the cache adapter interface and allows to test cache in the works which is "close to the real world". Being able to do so in tests for image sizes cache manager proves we can use other cache adapters such as Redis based ones.
refs https://github.com/TryGhost/Toolbox/issues/364
- Doing the `.catch(errors.NotFoundError...` was throwing another error as this syntax did not work with native promises. Checking `instanceof` works 100% and is way more explicit/readable way to handle this type of error differently
refs https://github.com/TryGhost/Toolbox/issues/364
- Cache adapter does not have a "has" method, so using "get" instead to achieve the same to have full compatibility. Should allow to just drop in the cache adapter implementation instead of current Map instance.
refs https://github.com/TryGhost/Toolbox/issues/364
- It was using an outdated syntax and relied on Bluebird depencency. Updated the syntax to async/await and dropped the Bluebird dependency.
fixes https://github.com/TryGhost/Ghost/issues/11541
- due to an upstream bug, typing emojis using Safari would show square
boxes in the editor
- the upstream patch has been fixed, so I've pulled it into our
mobiledoc-kit fork
- this commit bumps that package to fix the issue in Ghost
refs https://github.com/TryGhost/Team/issues/1799
Rather than using the `adminAuthAssets` config which is not updated to
be aware of running in a different directory to the cwd, we use the
getContentPath method which handles all of the directory checking.
Without this, we were unable to serve the admin-auth iframe, as the
directory was incorrect for self hosters.
- we've had an optimization in CI that copies a fresh SQLite DB to a
file, and copies it back when we need to do a DB reset
- I originally only let this run in CI but we've had it around for a
while so we should GA it to run on local machines
- there may be edge cases, but we should fix them instead of letting
tests run slower for development
- this also makes sure we clean up any existing files before
initializing the DB
- the comments system lives at members/api/comments and was meant to be removed from here in an early refactor but was missed, possibly as a result of the huge rebase we did
- prior to this commit, if you add or remove a faeture flag, you also have to update the snapshots for the settings tests
- feature flags are intended to be very easy to add and remove, and so this extra step doesn't fit with our needs
- it's also unnecessary, we don't need to verify the exact contents of the labs setting
- because of how the npm scripts were set up, we were running the full
Admin integration tests during the unit tests phase of CI
- this commit renames the majority of `test` to `test:unit` in the
package.json files, and aliases `test` to `test:unit`
- special packages like Admin have no-op'd `test:unit` scripts so we
don't end up running its tests
- one of the reasons our tests are so slow is because we're running 10
rounds of bcrypt hashing on shared hardware, nearly 300 times during
the database tests
- we don't particularly care about password hash strength during tests
so this commit reduces the number of rounds to 1 if we're running in a
test environment
- this drops the time to produce an individual hash from ~140ms to ~3ms,
saving us a lot of time overall
- when I extracted `mailgun-client` to a separate package, I
accidentally removed the lazyloading for the `mailgun-js` library,
which takes a non-negligible amount of time to require on boot
- this fixes that by moving the require into the function where it's
used
refs https://github.com/TryGhost/Toolbox/issues/363
- the oembed service is completely standalone and could do with some
individual unit tests
- moving it out to a package allows us to draw the boundaries better and
allows us to remove some dependencies from the core package.json
- I'm not sure if we gain anything here except noise
- it was added after a bug in the oembed service but it was regarding
oembed-parser and not metascraper
- added core and builtin integrations to test fixtures
- allowed passing a custom api key id to generate JWT
- updated admin key auth test to make successful request with a `core` integration, which doesn't work atm because relations are not returned
fixes https://github.com/TryGhost/Team/issues/1787
- Options were not passed
- The member context variable was not passed to the count helper
- Liked was always false for replies
closes https://github.com/TryGhost/Team/issues/1784
- auto generates offer code for new offers from name, same as display name
- stops auto generation as soon as offer code is manually edited once
- if the API controller endpoint is a function, we early return as we
expect the function to handle the response but we still ended up
calculating the headers beforehand, only to be thrown away
- this commit moves the header fetching code down in the flow so it's
only executed when needed
- this doesn't really have a big effect for us because 99% of our
controllers follow the object pattern
- we need to initialize the submodules in case they weren't done when
cloning, otherwise Casper will be missing
- we can avoid doing an initial build because it'll get wiped when we do
`yarn dev` anyway
The limit applies to the replies relation is applies globally when
fetching a collection, which means only 3 replies in total will be
fetched across all comments.
This patches the findPage method to manually fetch the replies and
replies adjacent relations manually on each comment, applying the limit
on a comment-by-comment basis.
This is not optimised as we currently still make the initial request for
relations.
