feat(medcat-trainer): Speed up project upload

medcat-trainer/webapp/api/api/models.py

@@ -354,6 +354,14 @@ def __str__(self):
         return str(self.label)
 
 
+def update_project_last_modified(project: 'ProjectAnnotateEntities', last_modified):
+    # so that the local project last_modified is updated
+    project.last_modified = last_modified
+    # so that the project last_modified is updated in the database
+    ProjectAnnotateEntities.objects.filter(pk=project.id).update(
+        last_modified=last_modified
+    )
+
 class EntityRelation(models.Model):
     user = models.ForeignKey(settings.AUTH_USER_MODEL, on_delete=models.CASCADE)
     project = models.ForeignKey('Project', on_delete=models.CASCADE)
@@ -370,8 +378,7 @@ class Meta:
 
     def save(self, *args, **kwargs):
         super().save(*args, **kwargs)
-        self.project.last_modified = self.last_modified
-        self.project.save()
+        update_project_last_modified(self.project, self.last_modified)
 
     def __str__(self):
         return f'{self.start_entity} - {self.relation} - {self.end_entity}'
@@ -402,8 +409,7 @@ class Meta:
 
     def save(self, *args, **kwargs):
         super().save(*args, **kwargs)
-        self.project.last_modified = self.last_modified
-        self.project.save()
+        update_project_last_modified(self.project, self.last_modified)
 
     def __str__(self):
         return str(self.entity)
@@ -551,8 +557,7 @@ class MetaAnnotation(models.Model):
 
     def save(self, *args, **kwargs):
         super().save(*args, **kwargs)
-        self.annotated_entity.last_modified = self.last_modified
-        self.annotated_entity.save()
+        update_project_last_modified(self.annotated_entity.project, self.last_modified)
 
     def __str__(self):
         return str(self.annotated_entity)

medcat-trainer/webapp/api/api/utils.py

@@ -3,6 +3,7 @@
 import os
 from typing import List
 import dill
+from functools import lru_cache
 
 import requests
 from background_task import background
@@ -482,15 +483,21 @@ def prep_docs(project_id: List[int], doc_ids: List[int], user_id: int):
                 project.id, project.prepared_documents)
 
 
+@lru_cache(maxsize=8)
+def _load_cuis_from_file(path):
+    with open(path) as f:
+        return json.load(f)
+
+
 @receiver(post_save, sender=ProjectAnnotateEntities)
 def save_project_anno(sender, instance, **kwargs):
     if instance.cuis_file:
-        post_save.disconnect(save_project_anno, sender=ProjectAnnotateEntities)
-        cuis_from_file = json.load(open(instance.cuis_file.path))
+        cuis_from_file = _load_cuis_from_file(instance.cuis_file.path)
         cui_list = [c.strip() for c in instance.cuis.split(',')]
-        instance.cuis = ','.join(set(cui_list) - set(cuis_from_file))
-        instance.save()
-        post_save.connect(save_project_anno, sender=ProjectAnnotateEntities)
+        new_cuis = ','.join(set(cui_list) - set(cuis_from_file))
+        if new_cuis != instance.cuis:
+            instance.cuis = new_cuis
+            ProjectAnnotateEntities.objects.filter(pk=instance.pk).update(cuis=new_cuis)
 
 
 def env_str_to_bool(var: str, default: bool):