Check file integrity in readers

invisible_cities/io/pmaps_io.py

@@ -4,15 +4,19 @@
 import tables as tb
 import pandas as pd
 
+from .. core.exceptions    import InvalidInputFileStructure
+
 from .. evm .pmaps         import  PMTResponses
 from .. evm .pmaps         import SiPMResponses
 from .. evm .pmaps         import S1
 from .. evm .pmaps         import S2
 from .. evm .pmaps         import PMap
 from .. evm                import nh5     as table_formats
+
 from .. reco.tbl_functions import filters as tbl_filters
 
 
+
 def store_peak(pmt_table, pmti_table, si_table,
                peak, peak_number, event_number):
     pmt_row  =  pmt_table.row
@@ -78,13 +82,23 @@ def _make_tables(hdf5_file, compression):
     return pmp_tables
 
 
+def check_file_integrity(file):
+    events_run      = file.root.Run  .events.read(field="evt_number")
+    events_pmaps_s1 = file.root.PMAPS.S1    .read(field="event")
+    events_pmaps_s2 = file.root.PMAPS.S2    .read(field="event")
+    if set(events_run) != set(events_pmaps_s1.tolist() + events_pmaps_s2.tolist()):
+        raise InvalidInputFileStructure("Inconsistent data: event number mismatch")
+
+
 def load_pmaps_as_df(filename, lazy=False):
     loader = load_pmaps_as_df_lazy if lazy else load_pmaps_as_df_eager
     return loader(filename)
 
 
 def load_pmaps_as_df_eager(filename):
     with tb.open_file(filename, 'r') as h5f:
+        check_file_integrity(h5f)
+
         pmap  = h5f.root.PMAPS
         to_df = pd.DataFrame.from_records
         return (to_df(pmap.S1   .read()),
@@ -102,6 +116,8 @@ def read_event(table, event):
 
     tables = "S1 S2 S2Si S1Pmt S2Pmt".split()
     with tb.open_file(filename, 'r') as h5f:
+        check_file_integrity(h5f)
+
         events = h5f.root.Run.events.read(field="evt_number")
         tables = [getattr(h5f.root.PMAPS, table, None) for table in tables]
         for event in events: