Arkindex 1.3.7
UI
-
Element, Project and Process UUIDs displayed in the frontend are now copyable.
-
Process logs now support color ANSI escape codes, allowing for a prettier/clearer display in the Process status page.
Workers
-
It is no longer possible to create Worker Runs, start Processes, apply Templates or build Task recipes using Worker Versions that are not
available, and warnings are displayed when trying to do so. -
All Worker Activity objects must now have a
process_idset. Deleting any process will delete all worker activities and not just the ones in aqueuedstate. -
When a process does not have a workflow, the frontend now displays an information notification, instead of a never-ending loading notification. This could occur when workflows expire.
S3 Import
-
PDF files are now imported from S3 buckets into Arkindex as folders, with one element per page. Any transcriptions they contain are imported as well.
API
Elements listing endpoints
-
The Element listing endpoints now support filtering with Worker Run IDs. This impacts the following endpoints:
ListElements,ListElementParents,ListElementChildrenas well as theirDestroycounterpoints. -
The
orderfield on theListElementChildren,ListElementsandListElementParentsendpoints now supports new sorting parameters. The retrieved children elements can be sorted by the centroid of their polygons, according to a "reading order":horizontal-lr,horizontal-rl,vertical-lrandvertical-rl. For detailed explanations, see the API documentation.
Model Version API
To support the logging of metrics during Machine Learning training processes, the Model Version API has been updated. - The CreateModelVersion endpoint, which no longer requires the hash, archive_hash and size fields. It now creates "empty" Model Versions, without an archive, as the model hasn’t finished training yet. - A new endpoint, ValidateModelVersion, validates the model archive and updates the Model Version to the available state.
Misc
-
Obsolete and unused fields (
width,heightandserver) have been removed from theCreateImageendpoint. -
The obsolete
datesandraw_datesfields have been removed from Entity API endpoints. -
The bulk classification creation endpoint
CreateClassificationsnow requires either a Worker Run ID or a Worker Version ID to be set. -
It is now possible to create Transcriptions on Elements that do not have a polygon and/or an image using the
CreateTranscriptionendpoint.