stash

backend/authentication/api.py

@@ -130,7 +130,8 @@ def getUserInfo(request):
                     return Response({'profile_picture_id': 'File does not exist.'}, status=400)
         user.save()
 
-        if old_file and old_file != user.profile_picture and old_file.connected_items.count() == 0 and old_file.profile_picture_users.count() == 0:
+        if old_file and old_file != user.profile_picture and old_file.connected_items.count() == 0 \
+                and old_file.profile_picture_users.count() == 0 and old_file.staged_by_workflows.count() == 0:
             old_file.delete()
 
     return Response({

backend/files/media_urls.py

@@ -26,7 +26,8 @@ def media_urls(request, hash_path):
     try:
         file = File.objects.filter(
             Q(connected_items__owner__in=request.user.friends_or_self()) |
-            Q(profile_picture_users__in=request.user.friends_or_self())
+            Q(profile_picture_users__in=request.user.friends_or_self()) |
+            Q(staged_by_workflows__owner__in=request.user.user.all())
         ).distinct().get(
             file=hash_path)
 

backend/toolshed/api/files.py

@@ -7,7 +7,7 @@ from rest_framework.response import Response
 from authentication.signature_auth import SignatureAuthenticationLocal
 from files.models import File
 from files.serializers import FileSerializer
-from toolshed.models import InventoryItem
+from toolshed.models import InventoryItem, WorkflowInstance
 
 
 @api_view(['GET'])
@@ -30,6 +30,16 @@ def get_item_files(request, item_id):
 def post_item_file(request, item_id):
     try:
         item = InventoryItem.objects.get(id=item_id, owner=request.user)
+        if 'file_hash' in request.data:
+            # Attach a file the caller already staged on one of their own workflows, identified
+            # by its content hash (which the client already computed before ever uploading it),
+            # instead of re-uploading bytes that are already stored server-side.
+            try:
+                file = File.objects.get(hash=request.data['file_hash'], staged_by_workflows__owner=request.user)
+            except File.DoesNotExist:
+                return Response(status=status.HTTP_404_NOT_FOUND)
+            item.files.add(file)
+            return Response(FileSerializer(file).data, status=status.HTTP_201_CREATED)
         serializer = FileSerializer(data=request.data)
         if serializer.is_valid():
             file = serializer.save()
@@ -40,6 +50,31 @@ def post_item_file(request, item_id):
         return Response(status=status.HTTP_404_NOT_FOUND)
 
 
+def get_staged_files(request, workflow_id):
+    try:
+        workflow = WorkflowInstance.objects.get(id=workflow_id, owner=request.user)
+        # Staged files are private working state the client already holds in full (name, size,
+        # mime_type, base64 data) from the moment it read them off disk/camera - the only thing
+        # it can't already know is whether/under what hash the upload was persisted, so that's
+        # all this returns, unlike the fuller FileSerializer representation item_files uses.
+        return Response(list(workflow.staged_files.values_list('hash', flat=True)))
+    except WorkflowInstance.DoesNotExist:
+        return Response(status=status.HTTP_404_NOT_FOUND)
+
+
+def post_staged_file(request, workflow_id):
+    try:
+        workflow = WorkflowInstance.objects.get(id=workflow_id, owner=request.user)
+        serializer = FileSerializer(data=request.data)
+        if serializer.is_valid():
+            file = serializer.save()
+            workflow.staged_files.add(file)
+            return Response({'hash': file.hash}, status=status.HTTP_201_CREATED)
+        return Response(serializer.errors, status=status.HTTP_400_BAD_REQUEST)
+    except WorkflowInstance.DoesNotExist:
+        return Response(status=status.HTTP_404_NOT_FOUND)
+
+
 @api_view(['POST', 'GET'])
 @permission_classes([IsAuthenticated])
 @authentication_classes([SignatureAuthenticationLocal])
@@ -58,7 +93,8 @@ def delete_item_file(request, item_id, file_id, format=None):  # /item_files/
         item = InventoryItem.objects.get(id=item_id, owner=request.user)
         file = item.files.get(id=file_id)
         item.files.remove(file_id)
-        if file.connected_items.count() == 0:
+        if file.connected_items.count() == 0 and file.profile_picture_users.count() == 0 \
+                and file.staged_by_workflows.count() == 0:
             file.delete()
         return Response(status=status.HTTP_204_NO_CONTENT)
     except InventoryItem.DoesNotExist:
@@ -67,8 +103,38 @@ def delete_item_file(request, item_id, file_id, format=None):  # /item_files/
         return Response(status=status.HTTP_404_NOT_FOUND)
 
 
+@api_view(['POST', 'GET'])
+@permission_classes([IsAuthenticated])
+@authentication_classes([SignatureAuthenticationLocal])
+def staged_files(request, workflow_id, format=None):  # /staged_files/
+    if request.method == 'GET':
+        return get_staged_files(request, workflow_id)
+    elif request.method == 'POST':
+        return post_staged_file(request, workflow_id)
+
+
+@api_view(['DELETE'])
+@permission_classes([IsAuthenticated])
+@authentication_classes([SignatureAuthenticationLocal])
+def delete_staged_file(request, workflow_id, file_hash, format=None):  # /staged_files/
+    try:
+        workflow = WorkflowInstance.objects.get(id=workflow_id, owner=request.user)
+        file = workflow.staged_files.get(hash=file_hash)
+        workflow.staged_files.remove(file)
+        if file.connected_items.count() == 0 and file.profile_picture_users.count() == 0 \
+                and file.staged_by_workflows.count() == 0:
+            file.delete()
+        return Response(status=status.HTTP_204_NO_CONTENT)
+    except WorkflowInstance.DoesNotExist:
+        return Response(status=status.HTTP_404_NOT_FOUND)
+    except File.DoesNotExist:
+        return Response(status=status.HTTP_404_NOT_FOUND)
+
+
 urlpatterns = [
     path('files/', list_all_files),
     path('item_files/<int:item_id>/', item_files),
     path('item_files/<int:item_id>/<int:file_id>/', delete_item_file),
+    path('staged_files/<int:workflow_id>/', staged_files),
+    path('staged_files/<int:workflow_id>/<str:file_hash>/', delete_staged_file),
 ]

backend/toolshed/api/inventory.py

@@ -7,6 +7,7 @@ from rest_framework.response import Response
 
 from authentication.models import ToolshedUser, KnownIdentity
 from authentication.signature_auth import SignatureAuthentication
+from files.models import File
 from toolshed.models import InventoryItem, StorageLocation, WorkflowInstance
 from toolshed.serializers import InventoryItemSerializer, StorageLocationSerializer, WorkflowInstanceSerializer
 
@@ -120,7 +121,12 @@ class WorkflowInstanceViewSet(viewsets.ModelViewSet):
 
     def perform_destroy(self, instance):
         if instance.owner == self.request.user.user.get():
+            staged_file_ids = list(instance.staged_files.values_list('id', flat=True))
             instance.delete()
+            for file in File.objects.filter(id__in=staged_file_ids):
+                if file.connected_items.count() == 0 and file.profile_picture_users.count() == 0 \
+                        and file.staged_by_workflows.count() == 0:
+                    file.delete()
 
 
 router.register(r'inventory_items', InventoryItemViewSet, basename='inventory_items')

backend/toolshed/migrations/0011_workflowinstance_staged_files.py

@@ -0,0 +1,19 @@
+# Generated by Django 4.2.2
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('files', '0001_initial'),
+        ('toolshed', '0010_rename_name_workflowinstance_slug'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='workflowinstance',
+            name='staged_files',
+            field=models.ManyToManyField(blank=True, related_name='staged_by_workflows', to='files.file'),
+        ),
+    ]

backend/toolshed/models.py

@@ -139,10 +139,11 @@ class WorkflowInstance(models.Model):
     state = models.CharField(max_length=255)
     payload = models.TextField(default='', blank=True) # an opaque, frontend-serialized JSON string on the backend.
     owner = models.ForeignKey(ToolshedUser, on_delete=models.CASCADE, related_name='workflows')
+    staged_files = models.ManyToManyField(File, related_name='staged_by_workflows', blank=True)
     created_at = models.DateTimeField(auto_now_add=True)
     updated_at = models.DateTimeField(auto_now=True)
 
     def __str__(self):
-        return f"{self.name} ({self.state})"
+        return f"{self.slug} ({self.state})"
 
 

backend/toolshed/offlinedata.py

@@ -284,14 +284,15 @@ def delete_user_account(user):
 def _delete_orphaned_files(file_ids):
     """Delete File rows (and their underlying blobs) in `file_ids` that are no longer referenced.
 
-    A File is considered orphaned once no InventoryItem and no ToolshedUser (profile picture)
+    A File is considered orphaned once no InventoryItem, no ToolshedUser (profile picture), and no
-    references it anymore. Returns the number of files deleted.
+    WorkflowInstance (staged file) references it anymore. Returns the number of files deleted.
     """
     from files.models import File
 
     deleted = 0
     for file_obj in File.objects.filter(id__in=file_ids):
-        if file_obj.connected_items.exists() or file_obj.profile_picture_users.exists():
+        if file_obj.connected_items.exists() or file_obj.profile_picture_users.exists() \
+                or file_obj.staged_by_workflows.exists():
             continue
         file_obj.file.delete(save=False)
         file_obj.delete()

backend/toolshed/serializers.py

@@ -203,9 +203,16 @@ class InventoryItemSerializer(serializers.ModelSerializer):
 
 class WorkflowInstanceSerializer(serializers.ModelSerializer):
     owner = serializers.StringRelatedField(read_only=True)
+    # The client already holds the full file (name, size, mime_type, base64 data) for anything it
+    # staged itself - hash is the only thing it can't already know, so that's all this exposes,
+    # unlike InventoryItemSerializer.files which needs the fuller FileSerializer representation.
+    staged_files = serializers.SerializerMethodField()
 
     class Meta:
         model = WorkflowInstance
-        fields = ['id', 'slug', 'state', 'payload', 'owner', 'created_at', 'updated_at']
+        fields = ['id', 'slug', 'state', 'payload', 'owner', 'staged_files', 'created_at', 'updated_at']
-        read_only_fields = ['owner', 'created_at', 'updated_at']
+        read_only_fields = ['owner', 'staged_files', 'created_at', 'updated_at']
+
+    def get_staged_files(self, obj):
+        return list(obj.staged_files.values_list('hash', flat=True))