File import

  • You can now select a folder where new files will be imported in your corpus
  • Element types are auto-selected when configuring an import: by default Volume and Page are used when available.

File import improvements

Transcriptions management

We are improving our transcriptions management system, aiming for a simplified version in a few releases. This release has some preleminary work:

  • A new API endpoint is added to create transcriptions in bulk, but linked to an Element.
  • The Page XML import now creates sub-elements instead of creating transcriptions on the Page element.

Performance improvement

We are also improving our write performances, so that our Machine Learning workers and end users are as fast as possible.

  • Faster element creation
  • Faster element link creation (towards parents)
  • Speedup best classifications filters on ListElements
  • Store element selection in the database instead of the session

Workers

  • Support for new Workers mode in DataImport

Deprecation

  • No more DataImport.payload: all the JSON fields have been migrated to standard Model fields

Bugfixes

  • Fix classification not being saved when creating a new element (a hotfix was issued earlier for that patch)
  • Fix some stale read issues on entities, workers, transcriptions & classifications
  • Fix a few frontend silent type errors