From a363b843720cd2c1c5b84992bcc2cb8eafc1393d Mon Sep 17 00:00:00 2001
From: Marina Gourtovaia <mg8@sanger.ac.uk>
Date: Fri, 24 Apr 2026 13:29:55 +0100
Subject: [PATCH] Added a function for validating runfolder name

---
 CHANGELOG                                   | 12 +++++++++--
 src/npgtracking/db/retrieval.py             | 22 +++++++++++++++++++++
 tests/{test_schema.py => test_retrieval.py} | 19 +++++++++++++++++-
 3 files changed, 50 insertions(+), 3 deletions(-)
 rename tests/{test_schema.py => test_retrieval.py} (83%)

diff --git a/CHANGELOG b/CHANGELOG
index f325039..18709fd 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -5,22 +5,30 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
 
+## [1.0.0] - 2026-04-24
+
 ### Changed
 
 * Upgraded MySQL version for GitHub CI to 8.4
 * Updated the ORM to reflect [changes](https://github.com/wtsi-npg/npg_tracking/pull/969)
   to the production database.
 * Regenerated the ORM with the current (4.0.3) version of `sqlacodegen`.
+* Renamed `tests/test_schema.py` to `tests/test_retrieval.py` to reflect the
+  module name that is being tested.
+
+### Added
+
+* `validate_runfolder` function to npgtracking.db.retrieval 
 
 ## [0.2.0] - 2025-12-12
 
 ### Added
 
-* get_run_by_id function added to retrieval module
+* `get_run_by_id` function added to retrieval module
 
 ### Fixed
 
-* run.batch_id is now a varchar
+* `run.batch_id` is now a varchar
 
 ## [0.1.1] - 2025-12-11
 
diff --git a/src/npgtracking/db/retrieval.py b/src/npgtracking/db/retrieval.py
index e424122..0ab8cdb 100644
--- a/src/npgtracking/db/retrieval.py
+++ b/src/npgtracking/db/retrieval.py
@@ -107,3 +107,25 @@ def get_run_by_id(
         raise ValueError("Can't get one run without an argument")
     result = session.execute(statement).scalar_one_or_none()
     return result
+
+
+def validate_runfolder(session: Session, id_run: int, runfolder_name: str) -> bool:
+    """Validates the runfolder name against the run ID.
+
+    Errors if run with the given ID does not exist.
+
+    Args:
+      session :
+        Database session
+      id_run :
+        Tracking run ID
+      runfolder_name :
+        Runfolder name
+    Returns:
+      `True` if the runfolder name and run id belong to the same run, `False`
+      otherwise.
+    """
+    run = session.execute(select(Run).where(Run.id_run == id_run)).scalar()
+    if not run:
+        raise ValueError(f"Run with ID {id_run} does not exist.")
+    return run.folder_name == runfolder_name
diff --git a/tests/test_schema.py b/tests/test_retrieval.py
similarity index 83%
rename from tests/test_schema.py
rename to tests/test_retrieval.py
index 3ed450e..52d01f6 100644
--- a/tests/test_schema.py
+++ b/tests/test_retrieval.py
@@ -21,10 +21,11 @@
 from npgtracking.db.retrieval import (
     get_run_by_id,
     get_runs_by_currentstatus,
+    validate_runfolder,
 )
 
 
-@m.describe("SchemaModel")
+@m.describe("Data retrieval from the tracking database")
 class TestSchemaModel(object):
     @m.context("When retrieving run records from tracking DB")
     @m.context("When there are no runs having a current status and manufacturer name")
@@ -102,3 +103,19 @@ def test_run_by_id(self, tracking_session):
 
         run = get_run_by_id(session=tracking_session, id_run=12345)
         assert run is None
+
+    @m.context("When runfolder name and run ID are correct")
+    @m.it("Validator returns True, otherwise False")
+    def test_validate_runfolder(self, tracking_session):
+        assert (
+            validate_runfolder(tracking_session, 51533, "430591-20251204_1628") is True
+        )
+        assert (
+            validate_runfolder(tracking_session, 51533, "430591-20251204_XXXX") is False
+        )
+
+    @m.context("When run ID is invalid")
+    @m.it("An error is raised")
+    def test_validate_runfolder_error(self, tracking_session):
+        with raises(ValueError, match="Run with ID 1 does not exist"):
+            validate_runfolder(tracking_session, 1, "430591-20251204_1628")