From 5899ed64ec2c0347baba53281222a35fe6b06d0b Mon Sep 17 00:00:00 2001
From: Soufyan Abdellati <abdellatis149@gmail.com>
Date: Thu, 31 Oct 2024 09:32:32 +0100
Subject: [PATCH] Removed unneeded attributes of normalizers' tasks

---
 boefjes/boefjes/app.py                      |  3 +--
 boefjes/boefjes/clients/scheduler_client.py | 16 ++++++++++++----
 mula/scheduler/models/task.py               |  2 --
 3 files changed, 13 insertions(+), 8 deletions(-)

diff --git a/boefjes/boefjes/app.py b/boefjes/boefjes/app.py
index e4cd9cf7095..b577af4d373 100644
--- a/boefjes/boefjes/app.py
+++ b/boefjes/boefjes/app.py
@@ -272,7 +272,7 @@ def get_runtime_manager(settings: Settings, queue: WorkerManager.Queue, log_leve
     plugin_service = PluginService(create_plugin_storage(session), create_config_storage(session), local_repository)
 
     item_handler: Handler
-    capabilities: list[str]
+    capabilities: list[str] | None = None
     reachable_networks: list[str] | None = None
 
     if queue is WorkerManager.Queue.BOEFJES:
@@ -283,7 +283,6 @@ def get_runtime_manager(settings: Settings, queue: WorkerManager.Queue, log_leve
         item_handler = NormalizerHandler(
             LocalNormalizerJobRunner(local_repository), bytes_api_client, settings.scan_profile_whitelist
         )
-        capabilities = [WorkerManager.Queue.NORMALIZERS.value]
 
     return SchedulerWorkerManager(
         item_handler,
diff --git a/boefjes/boefjes/clients/scheduler_client.py b/boefjes/boefjes/clients/scheduler_client.py
index 4db8e0572f8..f34a4645b33 100644
--- a/boefjes/boefjes/clients/scheduler_client.py
+++ b/boefjes/boefjes/clients/scheduler_client.py
@@ -69,7 +69,9 @@ def push_item(self, p_item: Task) -> None:
 
 
 class SchedulerAPIClient(SchedulerClientInterface):
-    def __init__(self, base_url: str, task_capabilities: list[str] = [], reachable_networks: list[str] | None = None):
+    def __init__(
+        self, base_url: str, task_capabilities: list[str] | None = None, reachable_networks: list[str] | None = None
+    ):
         self._session = Client(base_url=base_url, transport=HTTPTransport(retries=6))
         self._task_capabilities = task_capabilities
         self._reachable_networks = reachable_networks
@@ -85,15 +87,21 @@ def get_queues(self) -> list[Queue]:
         return TypeAdapter(list[Queue]).validate_json(response.content)
 
     def pop_item(self, queue_id: str) -> Task | None:
-        filters: list[Filter] = [
-            Filter(column="data", field="requirements", operator="<@", value=json.dumps(self._task_capabilities))
-        ]
+        filters: list[Filter] = []
 
+        # Client should only pop tasks that lie on a network that the runner is capable of reaching (e.g. the internet)
         if self._reachable_networks:
             filters.append(
                 Filter(column="data", field="network", operator="<@", value=json.dumps(self._reachable_networks))
             )
 
+        # Client should only pop tasks that have requirements that this runner is capable of (e.g. being able
+        # to handle ipv6 requests)
+        if self._task_capabilities:
+            filters.append(
+                Filter(column="data", field="requirements", operator="<@", value=json.dumps(self._task_capabilities))
+            )
+
         response = self._session.post(
             f"/queues/{queue_id}/pop", data=QueuePopRequest(filters=filters).model_dump_json()
         )
diff --git a/mula/scheduler/models/task.py b/mula/scheduler/models/task.py
index 7ee90728bc2..df190a3b4e6 100644
--- a/mula/scheduler/models/task.py
+++ b/mula/scheduler/models/task.py
@@ -99,8 +99,6 @@ class NormalizerTask(BaseModel):
     normalizer: Normalizer
     raw_data: RawData
 
-    requirements: list[str] | None = None
-
     @property
     def hash(self) -> str:
         """Make NormalizerTask hashable, so that we can de-duplicate it when