Add PDF and Page XML export Arkindex features

arkindex/process/management/commands/update_system_workers.py

@@ -90,6 +90,9 @@ class Command(BaseCommand):
         """
         assert worker.archived is None, "Cannot assign a version on an archived worker to a feature"
 
+        # Since there can be worker versions using the same Docker image but with different commands,
+        # we need to filter by command as well as by Docker image.
+        docker_command_filter = {"configuration__docker__command": docker_command} if docker_command else {"configuration__docker__command__isnull": True}
         # docker_image_iid is not unique, so we sort by state and take the most recent one.
         # Sorting by state means we prefer `available` versions first, then `created`, then `error`.
         worker_version = worker.versions.filter(
@@ -99,6 +102,7 @@ class Command(BaseCommand):
             configuration__user_configuration__isnull=True,
             gpu_usage=FeatureUsage.Disabled,
             model_usage=FeatureUsage.Disabled,
+            **docker_command_filter
         ).order_by("state", "-updated").first()
 
         if worker_version:

arkindex/process/models.py

@@ -666,6 +666,8 @@ class ArkindexFeature(Enum):
     InitElements = "init_elements"
     FileImport = "file_import"
     S3Ingest = "s3_ingest"
+    ExportPDF = "pdf_export"
+    ExportPageXML = "pagexml_export"
 
 
 class WorkerVersion(models.Model):

arkindex/system_workers.yml

@@ -11,3 +11,9 @@ features:
         command: worker-init-elements
     s3_ingest:
         image: registry.gitlab.teklia.com/arkindex/workers/import/s3:0.1.0
+    pdf_export:
+        image: registry.gitlab.teklia.com/arkindex/workers/export:0.1.0
+        command: worker-export-pdf
+    pagexml_export:
+        image: registry.gitlab.teklia.com/arkindex/workers/export:0.1.0
+        command: worker-export-pagexml