refs https://github.com/TryGhost/Team/issues/3151
- adds a new vite + typescript + storybook + TW package for setting up admin settings in react with base config that works with Ghost monorepo
- includes base components/design system for new settings UI
- adds eslint rule config to the package to match rest of Ghost codebase
- this is an experimental package as we figure out the best patterns for new admin packages in Ghost monorepo
Co-authored-by: Peter Zimon <zimo@ghost.org>
These versions use the latest version of @tryghost/errors, which uses
the correct import for @stdlib/utils-copy. This should hopefully stop
missing module errors when running locally.
As discussed with the product team we want to enforce kebab-case file names for
all files, with the exception of files which export a single class, in which
case they should be PascalCase and reflect the class which they export.
This will help find classes faster, and should push better naming for them too.
Some files and packages have been excluded from this linting, specifically when
a library or framework depends on the naming of a file for the functionality
e.g. Ember, knex-migrator, adapter-manager
refs 9d104c8511
- we've seen recurring instances where Ghost will hog memory after image
uploads
- we use `jemalloc` to try and help this, but it still seems to happen
- according to the sharp thread referenced in my commit above, memory
fragmentation can also be helped by reducing the concurrency within
sharp
- this is a bit of an experiment and we can revert if it causes issues
refs 27e4523aec
- we no longer use `oembed-parser`, so we can remove it from
package.json
- also pins the `@extractus/oembed-extractor` package and adds it into
`@tryghost/oembed-service` where it was missing
refs TryGhost/Ghost#16048
- When attempting to embed a Youtube video that has had embedding
disabled by its owner/author, Ghost displayed a generic error message
that didn't indicate the reason for the failed emebed.
- This change updated the error message when Youtube (or any provider)
returns 401: Unauthorized to indicate that the owner of the resource has
explicitly disabled embedding.
This is an initial start to using TypeScript in our non-core Ghost packages.
- Adds a prepare script to build the project after installing deps
- Adds an initial tsconfig.json which is compatible with our node env
- Migrates all of the code to TypeScript, including tests
- Updates tests to use ts-node so that we don't need to compile the tests
- ts-node is installed at the top level because the env is weird with lerna and
doesn't work otherwise
- Updates the yarn dev script to build the project with the --all and --revisions flag
fixes https://github.com/TryGhost/Team/issues/2385
The Sentry version has been locked to v7.11.1 for some time because Sentry still used a legacy Node feature, called domains. Due to a bug or change in in Noide 16+, those domains broke handling uncaught promise execptions. So Ghost crashed when a promise exception wasn't caught. But that shouldn't be the case because we have a global uncaught exception handler.
Luckily Sentry switched to AsyncLocalStorage in v7.48.0. This fixes the issue as demonstrated in c0cd62184c
refs TryGhost/Team#2904
<!-- Leave the line below if you'd like GitHub Copilot to generate a
summary from your commit -->
<!--
copilot:summary
-->
### <samp>🤖 Generated by Copilot at b3f5423</samp>
This pull request adds support for multiple formats of snippet content,
especially the `lexical` format, to the Ghost CMS. It modifies the
snippets API, model, and test files to handle the format conversion,
filtering, and serialization of snippets.
no issue
Rough prototype only, current limitations:
- **No persistence**. Docs are in-memory only, YJS state will be lost on server restart although it could be re-populated by clients if they reconnect without closing their local doc (needs testing/investigation)
- **No tie-in with saved lexical state**. Lexical state is updated in the post model via normal API requests from Admin which can mean the multiplayer doc and the saved lexical state become out of sync but there's no detection/indication of that state at present. Will also trigger the "someone else is editing" errors because multiplayer doesn't yet override the default post update collision detection
- **New posts don't start in multiplayer**. New posts don't have an ID and so can't have a respective YJS doc, after initial save we don't transition to multiplayer because the React component in Ember doesn't re-render on prop changes yet
- **No tests**. Experimental code just to get something working and help answer questions for what's next
Changes:
- added `lexicalMultiplayer` labs flag
- updated `<KoenigLexicalEditor>` to pass through the required `<KoenigComposer>` props for multiplayer when enabled
- added `lexical-multiplayer` service
- `init()` called during boot, used to set up the `enable()` and `disable()` methods so the flag can be toggled without restarts
- when enabled it adds `upgrade` request handling to the base Ghost server
- returns 404 if the URL doesn't match `/ghost/api/admin/posts/multiplayer/*`
- returns 401 if a valid session cookie is not present
- if everything is good, hands off to code in `y-websocket.js` that handles YJS doc creation, awareness, keepalive, etc
- uses doc names in the format `${post.id}/${docId}` where `docId` is `main` for the primary document and a GUID for any sub-documents like captions and nested editors in cards
- updated `SettingsBREADService` to check if the `labs` setting is changed, and enables/disables the `lexical-multiplayer` service as needed so the websockets server can be started and shutdown when toggling without requiring a restart
refs https://github.com/TryGhost/Team/issues/2691
- This bump changes the "sentEmailCount" method to a more descriptive "assertSentEmailCount" and adds chaining to this method.
refs TryGhost/Team#2691
- The bump adds possibility to make email's html/text snapshots with dynamic content. The breaking change here is with separate "matchPlaintextSnapshot" method extracted out of "matchMetadataSnapshot" to handle dynamic content in "text" part of the sent email.
- we previously used `@stdlib/utils` instead of the child package
`@stdlib/copy`, which is a lot smaller and contains our only use of
the parent
- this saves 140+MB of dependencies
- we keep ending up with multiple versions of the depedency in our tree,
and it's causing problems when comparing instances
- the workaround I'm implementing for now is to bump the package
everywhere and set a resolution so we only have 1 shared instance
- hopefully we can come up with a better method down the line
refs https://github.com/TryGhost/Team/issues/2845
We needed to update the html out of the cards to include images for light
and dark mode, and then we've used CSS to show/hide them
Co-authored-by: Fabien "egg" O'Carroll <fabien@allou.is>
no issue
- Renovate merged in a breaking change to ember-cli-code-coverage which
broke our coverage reporting for the admin app
- This commit fixes the issue by pinning the version of
ember-cli-code-coverage to the last working version and telling renovate
to ignore it in the future
- It also adds html coverage reporting to make it easier to run locally
and see your coverage before pushing
refs TryGhost/Team#2840
- moves the `entities.decode()` step to the `LinkReplacer` class so that
it's applied to all links, not just the ones that are replaced in the
email service
- adds a test case to `LinkReplacer` to ensure that the
`entities.decode()` step is applied to all links correctly, decoding any
URLs with HTML entities in them
---------
Co-authored-by: Chris Raible <chris@ghost.org>
refs https://github.com/TryGhost/Team/issues/2805
When we render mobiledoc to HTML, it automatically escapes HTML entities in the process, so a button or directly pasted link with href="https://example.com?code=test" will be rendered as href="https://example.com?code=test" as the url is encoded in the rendered HTML. Our link tracking was using the encoded URL as the redirect URL in newsletters, causing certain links to break.
This change updates the link tracking to decode the URL with `entities.decode(url)` so we store the correct redirect URL in our DB and ensure link tracking redirects to the correct url from newsletters.
---------
Co-authored-by: Rishabh <zrishabhgarg@gmail.com>
- by default, got retries failed requests, which is causing issues in
tests because we've disabled the network with `nock`
- this is causing huge idle time because got pauses before retrying
- this change disables the retries if we're running tests, so things are
more stable
