801 Remove NoSuchNodeError catch in dst loading

#801

[author: gondiaz]

This PR solves issue #800. The proposed solution removes the
NoSuchNodeError catch in load_dst function.

[reviewer: gonzaponte]

Clear and concise.

invisible_cities/io/dst_io.py

@@ -18,28 +18,34 @@ def _decode_str_columns(df):
     df        = df.apply(to_string)
     return df
 
-def load_dst(filename, group, node, evt_list=None):
+def load_dst(filename, group, node, evt_list=None, ignore_errors=False):
     """load a kdst if filename, group and node correctly found"""
-    try:
+
+    def read_dst_(filename, group, node, evt_list):
         with tb.open_file(filename) as h5in:
-            try:
-                table  = getattr(getattr(h5in.root, group), node)
-                if evt_list is None:
-                    values = table.read()
-                else:
-                    events = table.read(field='event')
-                    values = table.read_coordinates(np.where(np.isin(events, evt_list)))
-                return _decode_str_columns(pd.DataFrame.from_records(values, columns=table.colnames))
-            except NoSuchNodeError:
-                warnings.warn(f' not of kdst type: file= {filename} ', UserWarning)
-    except HDF5ExtError:
-        warnings.warn(f' corrupted: file = {filename} ', UserWarning)
-    except IOError:
-        warnings.warn(f' does not exist: file = {filename} ', UserWarning)
-
-
-def load_dsts(dst_list, group, node, evt_list=None):
-    dsts = [load_dst(filename, group, node, evt_list) for filename in dst_list]
+            table  = getattr(getattr(h5in.root, group), node)
+            if evt_list is None:
+                values = table.read()
+            else:
+                events = table.read(field='event')
+                values = table.read_coordinates(np.where(np.isin(events, evt_list)))
+            return _decode_str_columns(pd.DataFrame.from_records(values, columns=table.colnames))
+
+    if ignore_errors:
+        try:
+            return read_dst_(filename, group, node, evt_list)
+        except NoSuchNodeError:
+            warnings.warn(f' not of kdst type: file= {filename} ', UserWarning)
+        except HDF5ExtError:
+            warnings.warn(f' corrupted: file = {filename} ', UserWarning)
+        except IOError:
+            warnings.warn(f' does not exist: file = {filename} ', UserWarning)
+    else:
+        return read_dst_(filename, group, node, evt_list)
+
+
+def load_dsts(dst_list, group, node, evt_list=None, ignore_errors=False):
+    dsts = [load_dst(filename, group, node, evt_list, ignore_errors) for filename in dst_list]
     return pd.concat(dsts, ignore_index=True)
 
 def _make_tabledef(column_types : np.dtype, str_col_length : int) -> dict:

invisible_cities/io/dst_io_test.py

@@ -112,7 +112,7 @@ def test_load_dsts_warns_not_of_kdst_type(ICDATADIR):
     group = "DST"
     node  = "Events"
     with pytest.warns(UserWarning, match='not of kdst type'):
-        load_dsts([good_file, wrong_file], group, node)
+        load_dsts([good_file, wrong_file], group, node, ignore_errors=True)
 
 
 def test_load_dsts_warns_if_not_existing_file(ICDATADIR):
@@ -124,7 +124,7 @@ def test_load_dsts_warns_if_not_existing_file(ICDATADIR):
     group = "DST"
     node  = "Events"
     with pytest.warns(UserWarning, match='does not exist'):
-        load_dsts([good_file, wrong_file], group, node)
+        load_dsts([good_file, wrong_file], group, node, ignore_errors=True)
 
 
 def test_load_dst_converts_from_bytes(ICDATADIR, fixed_dataframe):

invisible_cities/io/mcinfo_io.py

@@ -132,16 +132,12 @@ def safe_copy_nexus_eventmap(h5out  : tb.file.File,
             'evt_number': the evt_number copied to the output;
             'nexus_evt' : the nexus event_ids copied to the output.
     """
-    # Suppress warning raised when table not
-    # present since case dealt with in try except.
-    with warnings.catch_warnings():
-        warnings.simplefilter("ignore", category=UserWarning)
-        evt_map = load_eventnumbermap(file_in)
     try:
+        evt_map = load_eventnumbermap(file_in)
         evt_mask = evt_map.evt_number.isin(evt_arr)
         df_writer(h5out, evt_map[evt_mask], 'Run', 'eventMap')
         return evt_map[evt_mask].to_dict('list')
-    except AttributeError:
+    except (AttributeError, tb.exceptions.NoSuchNodeError):
         ## Invoked when the eventMap table not found
         ## which occurs for pre-2020 MC files and
         ## at the first processing stage from nexus