auto create data source on project publish

When a project is published, we should always create a datasource
object based on the project storage type because with  the
DataSource  all access will go though it, otherwise the files wont
be accessible.

This commit creates the DataSource type automatically
when project is published.

The type of DataSource created depends on the settings.
For Physionet, the default data storage is direct,
and for HDN its GCP(defined by STORAGE_TYPE).

For auto creation of DataSource, i added a new environment variable
to control type of DataSource.  On second thought , i don't think we need
the new environment variable as we only have two type of storage
when project is published, and the new env variable don't server any new purpose
(i will wait for feedback from reviewers)

Author: superryeti

.env.example

@@ -172,3 +172,9 @@ DEFAULT_NUMBER_OF_APPLICATIONS_TO_REMIND = 5
 
 # minimum number of word needed for research_summary field for Credentialing Model.
 MIN_WORDS_RESEARCH_SUMMARY_CREDENTIALING = 20
+
+
+# Data Source configurations
+# Data Source determines how the data is stored(direct on server, gcs, big query, aws s3) and accessed(Direct, google, aws, Research Environment)
+# OPTIONS: direct, google, aws, research_environment
+DEFAULT_PROJECT_ACCESS_MECHANISM = 'direct'

physionet-django/physionet/settings/base.py

@@ -612,3 +612,5 @@ class StorageTypes:
 
 # minimum number of word needed for research_summary field for Credentialing Model.
 MIN_WORDS_RESEARCH_SUMMARY_CREDENTIALING = config('MIN_WORDS_RESEARCH_SUMMARY_CREDENTIALING', cast=int, default=20)
+
+DEFAULT_PROJECT_ACCESS_MECHANISM = config('DEFAULT_PROJECT_ACCESS_MECHANISM', default='direct')

physionet-django/project/projectfiles/gcs.py

@@ -4,6 +4,7 @@
 from django.shortcuts import redirect
 from google.cloud.exceptions import Conflict, NotFound
 from physionet.gcs import GCSObject, GCSObjectException, create_bucket, delete_bucket
+from project.models import DataSource
 from project.projectfiles.base import BaseProjectFiles
 from project.utility import DirectoryInfo, FileInfo, readable_size
 
@@ -117,6 +118,14 @@ def publish_initial(self, active_project, published_project):
 
     def publish_complete(self, active_project, published_project):
         self.rm_dir(active_project.file_root())
+        if settings.DEFAULT_PROJECT_ACCESS_MECHANISM == 'research_environment':
+            DataSource.objects.create(
+                project=published_project,
+                files_available=True,
+                data_location=DataSource.DataLocation.GOOGLE_CLOUD_STORAGE,
+                access_mechanism=DataSource.AccessMechanism.RESEARCH_ENVIRONMENT,
+                uri=f'gs://{published_project.project_file_root()}/',
+            )
 
     def publish_rollback(self, active_project, published_project):
         delete_bucket(published_project.project_file_root())

physionet-django/project/projectfiles/local.py

@@ -4,6 +4,7 @@
 
 from django.conf import settings
 from physionet.utility import serve_file, sorted_tree_files, zip_dir
+from project.models import DataSource
 from project.projectfiles.base import BaseProjectFiles
 from project.utility import (
     clear_directory,
@@ -120,7 +121,13 @@ def publish_rollback(self, active_project, published_project):
         os.rename(published_project.file_root(), active_project.file_root())
 
     def publish_complete(self, active_project, published_project):
-        pass
+        if settings.DEFAULT_PROJECT_ACCESS_MECHANISM == 'direct':
+            DataSource.objects.create(
+                project=published_project,
+                files_available=True,
+                data_location=DataSource.DataLocation.DIRECT,
+            )
+
 
     def get_project_file_root(self, slug, version, access_policy, klass):
         if access_policy: