Fixed test

Author: timconrad

tasks/update_software_items.py

@@ -8,10 +8,6 @@
 from tasks.init_db_task import get_connection
 from helper.config import cfg
 
-_mardi_cfg = cfg("mardi_kg")
-SOFTWARE_PROFILE_QID = _mardi_cfg.get("mardi_software_profile_qid")
-MARDI_PROFILE_TYPE_PID = _mardi_cfg.get("mardi_profile_type_pid")
-
 
 def build_query(offset: int, limit: Optional[int]) -> str:
     """
@@ -25,14 +21,17 @@ def build_query(offset: int, limit: Optional[int]) -> str:
         str: SPARQL query string.
     """
     limit_clause = f"\n    LIMIT {limit}" if limit is not None else ""
+    mardi_cfg = cfg("mardi_kg")
+    software_qid = mardi_cfg.get("mardi_software_profile_qid", "")
+    profile_pid = mardi_cfg.get("mardi_profile_type_pid", "")
 
     return f"""
     PREFIX wd: <https://portal.mardi4nfdi.de/entity/>
     PREFIX wdt: <https://portal.mardi4nfdi.de/prop/direct/>
 
     SELECT ?qid
     WHERE {{
-      ?item wdt:{MARDI_PROFILE_TYPE_PID} wd:{SOFTWARE_PROFILE_QID} .
+      ?item wdt:{profile_pid} wd:{software_qid} .
       BIND(REPLACE(STR(?item), "^.*/", "") AS ?qid)
     }}
     ORDER BY ?qid

tests/test_task_update_software_items.py

@@ -9,8 +9,6 @@
 from helper.config import cfg
 from tasks.init_db_task import _init_db, get_connection
 from tasks.update_software_items import (
-    SOFTWARE_PROFILE_QID,
-    MARDI_PROFILE_TYPE_PID,
     build_query,
     run_query,
     update_software_item_index_from_mardi,
@@ -63,8 +61,10 @@ def fake_run_query(endpoint: str, query: str, logger, max_retries: int, timeout:
 
     monkeypatch.setattr("tasks.update_software_items.cfg", fake_cfg)
     monkeypatch.setattr("tasks.update_software_items.run_query", fake_run_query)
-    monkeypatch.setattr("tasks.update_software_items.SOFTWARE_PROFILE_QID", "QSOFT")
-    monkeypatch.setattr("tasks.update_software_items.MARDI_PROFILE_TYPE_PID", "PPID")
+    # Ensure globals are present for backward compatibility in the module
+    mod = __import__("tasks.update_software_items")
+    setattr(mod, "SOFTWARE_PROFILE_QID", "QSOFT")
+    setattr(mod, "MARDI_PROFILE_TYPE_PID", "PPID")
     monkeypatch.setattr("tasks.update_software_items.time.sleep", lambda _: None)
     monkeypatch.setattr("tasks.update_software_items.get_run_logger", lambda: logging.getLogger("test_logger"))
 
@@ -110,7 +110,5 @@ def test_run_query_against_real_endpoint(monkeypatch):
 
     assert len(results) <= limit
     assert all(qid.startswith("Q") for qid in results)
-    if not SOFTWARE_PROFILE_QID or not MARDI_PROFILE_TYPE_PID:
-        pytest.skip("Profile constants not configured; skipping integration assertion")
-    assert SOFTWARE_PROFILE_QID.startswith("Q")
-    assert MARDI_PROFILE_TYPE_PID.startswith("P")
+    assert wb_cfg.get("mardi_software_profile_qid", "").startswith("Q")
+    assert wb_cfg.get("mardi_profile_type_pid", "").startswith("P")

workflow_main.py

@@ -18,8 +18,8 @@
 
 @flow(name="start_update_embedding_workflow")
 def start_update_embedding_workflow(
-        update_embeddings_loop_iterations: int = 1,
-        update_embeddings_embeddings_per_loop: int = 10,
+        update_embeddings_loop_iterations: int = 10,
+        update_embeddings_embeddings_per_loop: int = 100,
 ):
     """
     Orchestrate the end-to-end software documentation embedding sync flow.