no issue
- updates the attribute sanitiser of the posts importer to convert `post.page=true/false` to `post.type='page'/'post'
- gives precedence to `post.type` if an imported post somehow has both `post.page` and `post.type` attributes
no-issue
* Removed redundant options from permittedOptions
The column option is already permitted at the Base model level.
* Remove defaultColumnsToFetch from Base model
* Removed defaultColumnsToFetch from Post model
no-issue
The previous version of url-utils's absoluteToRelative function had some
implicit behaviour that was relied on here. Namely that when passed an
empty string as the url AND the withoutSubdirectory option set to true,
we would end up calling the `replace` method on `null` - this would
throw an error and cause the subscribers app to enter the error handling
- which was desired behaviour.
This updates to explicitly check, and execute the error handling based
on the missing location property.
refs #10922
The frontend UrlGenerator parses the filters itself rather than passing
to the api, so we need to replicate the conversion from page -> type
no-issue
This module was being shared between multiple parts of the codebase, the
core/shared directory is a stopgap before we move it out to mongo-utils.
no issue
- the `mobiledoc_revisions` table can grow very large in certain circumstances which can result in Out-Of-Memory errors when performing backups, resulting in failed upgrades
- adds `mobiledoc_revisions` to the exporter excluded tables list as a temporary solution until we have safer export creation and/or improved revision handling
no issue
- updates `@tryghost/url-utils` following an internal refactor of the package
- renames `makeAbsoluteUrls` to `htmlRelativeToAbsolute` to better reflect what the function is doing
- renames `getBlogUrl` to `getSiteUrl`
- updates UrlUtils test stubbing util to work with a class
- fixes use of invalid port numbers in tests (max port number is 65535, any higher is an invalid URL that will error with some parsers)
refs #8717
- The posts without slugs should not be taken into account when detecting duplicates as slug field is not required when importing.
- Ideal solution would require generating slug before duplicate detection phase. This would cause duplicate detection to take 'title' into account which didn't happen before.
no issue
Admin key token verification was using hardcoded audience check with v2 admin endpoint, this updates it to check against api version and api type of the request url
refs #10922
When rolling back the removal of the page column, we must re-add it, but
the definition for it has been removed from the schema, so we must
hardcode the definition.
refs #10922
This will allow us to pass through a customer transformer to replace
references to removed columns in the mongo query generated inside of NQL