workflows: author submissions not erroring

also some small config fixes for easy testing with docker

* ref: cern-sis/issues-inspire#643

Makefile

@@ -9,6 +9,7 @@ run: services start-inspirehep sleep setup-inspirehep start-backoffice sleep set
 run-inspirehep: services start-inspirehep sleep setup-inspirehep
 run-backoffice: services start-backoffice sleep setup-backoffice
 
+start: services start-inspirehep start-backoffice
 
 start-inspirehep:
 	echo -e "\033[0;32m Starting HEP. \033[0m"

backend/inspirehep/config.py

@@ -31,7 +31,7 @@
 FEATURE_FLAG_ENABLE_ORCID_PUSH = False
 FEATURE_FLAG_ENABLE_REDIRECTION_OF_PIDS = False
 FEATURE_FLAG_ENABLE_ASSIGN_AUTHOR_PAPERS = False
-FEATURE_FLAG_ENABLE_SEND_TO_BACKOFFICE = False
+FEATURE_FLAG_ENABLE_SEND_TO_BACKOFFICE = True
 # Only push to ORCIDs that match this regex.
 # Examples:
 #   any ORCID -> ".*"
@@ -51,8 +51,8 @@
 INSPIRE_NEXT_URL = "http://localhost:5000"
 LEGACY_BASE_URL = "https://old.inspirehep.net"
 LEGACY_RECORD_URL_PATTERN = "http://inspirehep.net/record/{recid}"
-INSPIRE_BACKOFFICE_URL = "https://backoffice.dev.inspirebeta.net"
-AUTHENTICATION_TOKEN_BACKOFFICE = "CHANGE_ME"
+INSPIRE_BACKOFFICE_URL = "http://host.docker.internal:8001"
+AUTHENTICATION_TOKEN_BACKOFFICE = "2e04111a61e8f5ba6ecec52af21bbb9e81732085"
 MAX_API_RESULTS = 10000
 REST_MIMETYPE_QUERY_ARG_NAME = "format"
 

backend/inspirehep/submissions/views.py

@@ -89,6 +89,7 @@ def send_post_request_to_workflows(
         """
 
         LOGGER.info("Sending post request to workflows", url=url, endpoint=endpoint)
+        LOGGER.info("YAHOO", url=url, endpoint=endpoint)
         headers = {
             "content-type": "application/json",
             "Authorization": f"{bearer_keyword} {token}",

workflows/dags/author/author_create/author_create_init.py

@@ -17,17 +17,22 @@
 logger = logging.getLogger(__name__)
 
 
+def task_failure_alert(context):
+    print(f"Task has failed, task_instance_key_str: {context['task_instance_key_str']}")
+
+
 @dag(
     params={
         "workflow_id": Param(type="string", default=""),
         "data": Param(type="object", default={}),
+        "collection": Param(type="string", default=AUTHORS),
     },
     start_date=datetime.datetime(2024, 5, 5),
     schedule=None,
     catchup=False,
     # TODO: what if callback fails? Data in backoffice not up to date!
     on_failure_callback=set_workflow_status_to_error,
-    tags=["authors"],
+    tags=[AUTHORS],
 )
 def author_create_initialization_dag():
     """
@@ -47,6 +52,7 @@ def author_create_initialization_dag():
     @task()
     def set_workflow_status_to_running(**context):
         status_name = "running"
+        print(workflow_management_hook.headers)
         workflow_management_hook.set_workflow_status(
             status_name=status_name, workflow_id=context["params"]["workflow_id"]
         )

workflows/plugins/hooks/backoffice/workflow_management_hook.py

@@ -19,6 +19,8 @@ class WorkflowManagementHook(BackofficeHook):
     def __init__(self, collection):
         super().__init__()
         self.endpoint = f"api/workflows/{collection}"
+        print("this is the endpoint")
+        print(self.endpoint)
 
     def set_workflow_status(self, status_name: str, workflow_id: str) -> Response:
         """
@@ -33,6 +35,8 @@ def set_workflow_status(self, status_name: str, workflow_id: str) -> Response:
         request_data = {
             "status": status_name,
         }
+        print("MY NAME IS JEFF")
+        print("setting up partial update")
         return self.partial_update_workflow(
             workflow_partial_update_data=request_data, workflow_id=workflow_id
         )
@@ -58,9 +62,12 @@ def partial_update_workflow(
         self, workflow_id: str, workflow_partial_update_data: dict
     ) -> Response:
         endpoint = f"{self.endpoint}/{workflow_id}/"
+        print("wow this is great")
+        print(self.headers)
+        print(workflow_partial_update_data)
         return self.run_with_advanced_retry(
             _retry_args=self.tenacity_retry_kwargs,
             method="PATCH",
-            data=workflow_partial_update_data,
+            json=workflow_partial_update_data,
             endpoint=endpoint,
         )

workflows/plugins/hooks/generic_http_hook.py

@@ -51,6 +51,12 @@ def run(
         else:
             url = self.base_url + endpoint
 
+        print("wowawewa")
+        print(json)
+        print(data)
+        print(params)
+        print(headers)
+
         req = requests.Request(
             method, url, json=json, data=data, params=params, headers=headers
         )

workflows/plugins/include/utils/set_workflow_status.py

@@ -1,6 +1,6 @@
 import logging
 
-from hooks.backoffice.workflow_management_hook import AUTHORS, WorkflowManagementHook
+from hooks.backoffice.workflow_management_hook import WorkflowManagementHook
 from requests import Response
 
 logger = logging.getLogger(__name__)
@@ -36,11 +36,15 @@ def set_workflow_status_to_error(context: dict) -> None:
         workflow_id (str): The identifier for the workflow.
     """
     logger.info("Setting workflow status to error")
-    response = WorkflowManagementHook().set_workflow_status(
-        status_name="error",
-        workflow_id=context["params"]["workflow_id"],
-        collection=AUTHORS,
+    print("cossa")
+    print(context)
+    response = WorkflowManagementHook(
+        collection=context["params"]["collection"]
+    ).set_workflow_status(
+        status_name="error", workflow_id=context["params"]["workflow_id"]
     )
+    print("response is")
+    print(response)
     try:
         response.raise_for_status()
     except Exception as e: