Refactor SimTelFile to allow iterating over all events based

examples/iter_mc_events.py

@@ -16,8 +16,10 @@
 energies = []
 corex = []
 corey = []
-with SimTelFile(args.inputfile) as f:
-    for i, event in enumerate(f.iter_mc_events()):
+
+# to get all simulated mc events, we set skip_non_triggered to False
+with SimTelFile(args.inputfile, skip_non_triggered=False) as f:
+    for i, event in enumerate(f):
         energies.append(event['mc_shower']['energy'])
         corex.append(event['mc_event']['xcore'])
         corey.append(event['mc_event']['ycore'])
@@ -31,19 +33,24 @@
     51
 )
 
-plt.figure()
-plt.hist(energies, bins=bins)
-plt.xscale('log')
-plt.xlabel(r'$E \,/\, \mathrm{TeV}$')
-plt.tight_layout()
+fig, (ax1, ax2) = plt.subplots(1, 2, layout="constrained")
+
+ax1.hist(energies, bins=bins)
+ax1.set(
+    xscale='log',
+    yscale='log',
+    xlabel=r'$E \,/\, \mathrm{TeV}$',
+)
 
 
-fig, ax = plt.subplots()
-ax.set_aspect('equal')
-ax.hist2d(
+ax2.hist2d(
     corex, corey,
     bins=100
 )
+ax2.set(
+    aspect=1,
+    xlabel=r'$x \,/\, \mathrm{m}$',
+    ylabel=r'$y \,/\, \mathrm{m}$',
+)
 
-fig.tight_layout()
 plt.show()

src/eventio/base.py

@@ -1,3 +1,5 @@
+from io import BufferedIOBase
+import warnings
 import weakref
 import struct
 import gzip
@@ -64,7 +66,8 @@ def __init__(self, path, zcat=True):
         self.read_process = None
         self.zstd = False
         self.next = None
-        self._filehandle = None
+        self.peek_error = None
+        self._filehandle: BufferedIOBase | None = None
 
         if not is_eventio(path):
             raise ValueError('File {} is not an eventio file'.format(path))
@@ -94,7 +97,7 @@ def __init__(self, path, zcat=True):
 
         elif is_zstd(path):
             log.info('Found zstd compressed file')
-            self._filehandle = zstd.ZstdDecompressor().stream_reader(open(path, 'rb'))
+            self._filehandle = zstd.ZstdDecompressor().stream_reader(open(path, 'rb'), read_size=1024**2)
             self.zstd = True
 
         else:
@@ -115,14 +118,16 @@ def __iter__(self):
 
     def __next__(self):
         if self.next is not None:
-            o = self.next
-            self.next = None
+            o, self.next = self.next, None
             return o
 
-        self.seek(self._next_header_pos)
-        read_sync_marker(self)
+        return self._read_next_object()
+
+    def _read_next_object(self):
+        self._filehandle.seek(self._next_header_pos)
+        read_sync_marker(self._filehandle)
         header = read_header(
-            self,
+            self._filehandle,
             toplevel=True,
             offset=self._next_header_pos,
         )
@@ -134,8 +139,16 @@ def __next__(self):
         )
 
     def peek(self):
-        if self.next is None:
-            self.next = next(self)
+        if self.next is None and self.peek_error is None:
+            try:
+                self.next = self._read_next_object()
+            except StopIteration:
+                self.next = None
+            except (EOFError, IOError) as e:
+                warnings.warn(f"Error in _read_next_object during peek: {e}")
+                self.next = None
+                self.peek_error = e
+
         return self.next
 
     def seek(self, position, whence=0):

src/eventio/simtel/objects.py

@@ -1,4 +1,4 @@
-''' Implementations of the simtel_array EventIO object types '''
+'''Implementations of the simtel_array EventIO object types.'''
 import os
 import sys
 import numpy as np
@@ -1891,7 +1891,7 @@ class CalibrationPhotoelectrons(EventIOObject):
     eventio_type = 2034
 
     def __str__(self):
-        return f'{self.__class__.__name__}(type={self.header.id})'
+        return f'{self.__class__.__name__}[{self.header.type}](type={self.header.id})'