From 679802bdc37ab2bf66122b074c38bfb7c9c63f8b Mon Sep 17 00:00:00 2001
From: Benoit Crickboom <pro@sleig.be>
Date: Fri, 23 Aug 2024 10:26:59 +0200
Subject: [PATCH] improved  to handle

---
 orthanc_tools/orthanc_cleaner.py      | 38 ++++++++++++++++++---------
 tests/docker-setup/docker-compose.yml |  1 +
 tests/test_3_orthancs.py              | 30 +++++++++++++++++++++
 3 files changed, 56 insertions(+), 13 deletions(-)

diff --git a/orthanc_tools/orthanc_cleaner.py b/orthanc_tools/orthanc_cleaner.py
index acd291a..a88d853 100644
--- a/orthanc_tools/orthanc_cleaner.py
+++ b/orthanc_tools/orthanc_cleaner.py
@@ -65,9 +65,29 @@ def clean(self):
         for rule in labels_rules:
             logger.info(f"{rule.label_name} - {rule.retention_duration} weeks")
 
-        studies_to_delete = []
+        # get the list of studies to delete
+        studies_to_delete = self.get_studies_to_delete(labels_rules=labels_rules)
+
+        while len(studies_to_delete) > 0:
+            
+            # Delete the found studies
+            for s in studies_to_delete:
+                try:
+                    self._api_client.studies.delete(s.orthanc_id)
+                    logger.info(f"Deleting study {s.dicom_id} from {s.main_dicom_tags.get('StudyDate')}...")
+                except Exception as ex:
+                    logger.error(f"ERROR: {str(ex)}")
+            
+            # Get one more time the list of studies to delete (because we may have been limited to the value of 'LimitFindResults')
+            studies_to_delete = self.get_studies_to_delete(labels_rules=labels_rules)
+
+        logger.info("Clean up done!")
 
-        # Query Orthanc and delete the studies for each label rule
+    def get_studies_to_delete(self, labels_rules: LabelRule) -> List[str]:
+        '''
+        Query Orthanc to get the list of studies to delete (depending on the date and the label)
+        '''
+        studies_to_delete = []
         for label_rule in labels_rules:
 
             # Let's compute the date
@@ -79,21 +99,13 @@ def clean(self):
                 labels=[label_rule.label_name]
                 )
 
-            # Filter the old studies which were recently stored in Orthanc
+            # Filter out the old studies which were recently stored in Orthanc
             for s in studies_to_delete_by_study_date:
                 if limit_date > s.last_update.date():
                     studies_to_delete.append(s)
 
-        # Delete the found studies
-        for s in studies_to_delete:
-            try:
-                self._api_client.studies.delete(s.orthanc_id)
-                logger.info(f"Deleting study {s.dicom_id} from {s.main_dicom_tags.get('StudyDate')}...")
-            except Exception as ex:
-                logger.error(f"ERROR: {str(ex)}")
-
-        logger.info("Clean up done!")
-
+        logger.info(f"Found {len(studies_to_delete)} studies to delete...")
+        return studies_to_delete
 
     def compute_limit_date(self, number_of_weeks) -> datetime.date:
         limit_date = datetime.date.today() - datetime.timedelta(weeks=number_of_weeks)
diff --git a/tests/docker-setup/docker-compose.yml b/tests/docker-setup/docker-compose.yml
index 32868ca..63036e2 100644
--- a/tests/docker-setup/docker-compose.yml
+++ b/tests/docker-setup/docker-compose.yml
@@ -6,6 +6,7 @@ services:
     environment:
       VERBOSE_STARTUP: "true"
       VERBOSE_ENABLED: "true"
+      ORTHANC__LIMIT_FIND_RESULTS: 100
       TRANSFERS_PLUGIN_ENABLED: "true"
       ORTHANC__STABLE_AGE: "1"
       ORTHANC__KEEP_ALIVE_TIMEOUT: "5"
diff --git a/tests/test_3_orthancs.py b/tests/test_3_orthancs.py
index 225cbc9..b622780 100644
--- a/tests/test_3_orthancs.py
+++ b/tests/test_3_orthancs.py
@@ -693,6 +693,36 @@ def test_orthanc_cleaner_with_future_studies(self):
         self.assertEqual(len(self.oa.studies.get_all_ids()), 1)
         self.assertNotEqual(old_study_id, self.oa.studies.get_all_ids()[0])
 
+    def test_orthanc_cleaner_with_more_than_100_studies(self):
+        self.oa.delete_all_content()
+
+        # populate Orthanc with 120 old studies...
+        populator = OrthancTestDbPopulator(
+            api_client=self.oa,
+            studies_count=120,
+            series_count=1,
+            instances_count=1,
+            from_study_date=datetime.date.today() + datetime.timedelta(weeks=3),
+            to_study_date=datetime.date.today() + datetime.timedelta(weeks=4)
+        )
+        populator.execute()
+
+        # apply a label to all the studies.
+        studies_ids = self.oa.studies.get_all_ids()
+        for id in studies_ids:
+            self.oa.studies.add_label(id, "LABEL3")
+
+        # then, remove the label for a single study
+        self.oa.studies.delete_label(studies_ids[0], "LABEL3")
+
+        # execute cleaner
+        cleaner = OrthancCleaner(api_client=self.oa, execution_time=None, labels_file_path=here / "stimuli/labels.csv")
+        cleaner.execute()
+
+        # only one single study should be kept
+        self.assertEqual(len(self.oa.studies.get_all_ids()), 1)
+
+
 if __name__ == '__main__':
     logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
     unittest.main()