task/WP-164 Implement Workspace Search

server/portal/apps/projects/views.py

@@ -17,7 +17,9 @@
         list_projects, get_project, create_shared_workspace,\
         update_project, get_workspace_role, change_user_role, add_user_to_workspace,\
         remove_user, transfer_ownership
-
+from portal.apps.search.tasks import tapis_project_listing_indexer
+from portal.libs.elasticsearch.indexes import IndexedProject
+from elasticsearch_dsl import Q
 
 LOGGER = logging.getLogger(__name__)
 
@@ -63,12 +65,35 @@ def get(self, request):
         }
         ```
         """
-        # TODOv3: Support Elasticsearch queries for V3 projects https://jira.tacc.utexas.edu/browse/TV3-160
-        # query_string = request.GET.get('query_string')
-        # offset = int(request.GET.get('offset', 0))
-        # limit = int(request.GET.get('limit', 100))
-        client = request.user.tapis_oauth.client
-        listing = list_projects(client)
+
+        query_string = request.GET.get('query_string')
+        offset = int(request.GET.get('offset', 0))
+        limit = int(request.GET.get('limit', 100))
+
+        listing = []
+
+        if query_string:
+            search = IndexedProject.search()
+
+            ngram_query = Q("query_string", query=query_string,
+                            fields=["title"],
+                            minimum_should_match='100%',
+                            default_operator='or')
+
+            wildcard_query = Q("wildcard", title=f'*{query_string}*')
+
+            search = search.query(ngram_query | wildcard_query)
+            search = search.extra(from_=int(offset), size=int(limit))
+
+            res = search.execute()
+            hits = [hit.to_dict() for hit in res]
+            listing = hits
+        else:
+            client = request.user.tapis_oauth.client
+            listing = list_projects(client)
+
+        tapis_project_listing_indexer.delay(listing)
+
         return JsonResponse({"status": 200, "response": listing})
 
     def post(self, request):  # pylint: disable=no-self-use

server/portal/apps/search/tasks.py

@@ -3,7 +3,7 @@
 from django.conf import settings
 from celery import shared_task
 from portal.libs.agave.utils import user_account, service_account
-from portal.libs.elasticsearch.utils import index_listing
+from portal.libs.elasticsearch.utils import index_listing, index_project_listing
 from portal.apps.users.utils import get_tas_allocations
 from portal.apps.projects.models.metadata import ProjectMetadata
 from portal.libs.elasticsearch.docs.base import (IndexedAllocation,
@@ -78,3 +78,8 @@ def index_project(self, project_id):
     project_doc = IndexedProject(**project_dict)
     project_doc.meta.id = project_id
     project_doc.save()
+
+
+@shared_task(bind=True, max_retries=3, queue='default')
+def tapis_project_listing_indexer(self, projects):
+    index_project_listing(projects)

server/portal/libs/elasticsearch/docs/base.py

@@ -17,37 +17,21 @@
 
 
 class IndexedProject(Document):
+    id = Keyword()
     title = Text(fields={'_exact': Keyword()})
     description = Text()
-    created = Date()
-    lastModified = Date()
-    projectId = Keyword()
+    path = Text()
+    name = Text()
+    host = Text()
     owner = Object(
-        properties={
-            'username': Keyword(),
-            'fullName': Text()
-        }
-    )
-    pi = Object(
-        properties={
-            'username': Keyword(),
-            'fullName': Text()
-        }
-    )
-    coPIs = Object(
-        multi=True,
-        properties={
-            'username': Keyword(),
-            'fullName': Text()
-        }
-    )
-    teamMembers = Object(
-        multi=True,
-        properties={
-            'username': Keyword(),
-            'fullName': Text()
-        }
+            properties={
+                'username': Keyword(),
+                'firstName': Text(),
+                'lastName': Text(),
+                'email': Text()
+            }
     )
+    updated = Date()
 
     @classmethod
     def from_id(cls, projectId):

server/portal/libs/elasticsearch/utils.py

@@ -207,6 +207,8 @@ def index_listing(files):
         file_dict = dict(_file)
         if file_dict['name'][0] == '.':
             continue
+        if not file_dict['path'].startswith('/'):
+            file_dict['path'] = '/' + file_dict['path']
         file_dict['lastUpdated'] = current_time()
         file_dict['basePath'] = os.path.dirname(file_dict['path'])
         file_uuid = file_uuid_sha256(file_dict['system'], file_dict['path'])
@@ -219,3 +221,25 @@ def index_listing(files):
             })
 
     bulk(client, ops)
+
+
+def index_project_listing(projects):
+    from portal.libs.elasticsearch.docs.base import IndexedProject
+
+    idx = IndexedProject.Index.name
+    client = get_connection('default')
+    ops = []
+
+    for _project in projects:
+        project_dict = dict(_project)
+        project_dict['updated'] = current_time()
+        project_uuid = get_sha256_hash(project_dict['id'])
+        ops.append({
+            '_index': idx,
+            '_id': project_uuid,
+            'doc': project_dict,
+            '_op_type': 'update',
+            'doc_as_upsert': True
+        })
+
+    bulk(client, ops)