Merge pull request #1398 from cta-observatory/check_duplicated_runs

Prevent using more than one datacheck file per run

Author: moralejo

lstchain/scripts/lstchain_cherenkov_transparency.py

@@ -94,6 +94,10 @@ def main():
     picture_thresh = []
     boundary_thresh = []
 
+    # Prevent more than one datacheck file per run ID
+    # Dict with key: run_id and value: file
+    run_info = {}
+
     for file in tqdm(files):
         a = tables.open_file(file)
 
@@ -167,10 +171,22 @@ def main():
         corrected_elapsed_time.append(newtime)
 
         subrun_index.append(a.root.dl1datacheck.cosmics.col('subrun_index'))
-        run.append(int(file[file.find('.Run') + 4:file.find('.Run') + 9]))
+        runnumber = int(file[file.find('.Run') + 4:file.find('.Run') + 9])
+        run.append(runnumber)
 
         num_subruns = len(a.root.dl1datacheck.cosmics.col('subrun_index'))
 
+        # check if the run is duplicated
+        prev_file = run_info.get(runnumber)
+        if prev_file is None:
+            run_info[runnumber] = file
+        else:
+            a.close()
+            raise RuntimeError(
+                f"The datacheck for run {runnumber} has been loaded twice: "
+                f"from path {prev_file} and path {file}"
+            )
+
         if 'pedestals' in a.root.dl1datacheck:
             if len(a.root.dl1datacheck.pedestals.col(
                     'subrun_index')) == num_subruns:  # all subruns present

lstchain/scripts/lstchain_longterm_dl1_check.py

@@ -245,6 +245,10 @@ class pixwise_info(tables.IsDescription):
 
     dicts = [cosmics, pedestals, flatfield]
 
+    # Prevent more than one datacheck file per run ID
+    # Dict with key: run_id and value: file
+    run_info = {}
+
     # files are of the type datacheck_dl1_LST-1.RunXXXXX.h5
     for file in files:
 
@@ -257,6 +261,17 @@ class pixwise_info(tables.IsDescription):
         runnumber = int(file.name[file.name.find('.Run') + 4:
                                   file.name.find('.Run') + 9])
 
+        # check if the run is duplicated
+        prev_file = run_info.get(runnumber)
+        if prev_file is None:
+            run_info[runnumber] = file
+        else:
+            a.close()
+            raise RuntimeError(
+                f"The datacheck for run {runnumber} has been loaded twice: "
+                f"from path {prev_file} and path {file}"
+            )
+
         # Lists to keep the datacheck tables for cosmics, pedestals and
         # flatfield. The "_no_stars" list will have nans for pixels which
         # were close to stars during a given subrun