Fixes last event not read issue (#64)

Reported by Pim Herbschleb via email.

The HepMC3 C++ has a bug, because under some circumstances it reports
reading an event as success (with read_event()), even if the input
stream did not contain any event data. I am using a workaround for this,
which unfortunately require the use of heuristics that can fail. Pim
reported an input file on which the previous heuristic failed, so I
changed the code once again to pass this and the previous tests. Some
tests needed to be modified. The new code is not able to distinguish
between an EOF error at the end of an event and a reading error in the
middle of an event anymore, but it is more important to read events
correctly which have been correctly written than handling a rare error
cases more elegantly.

Author: HDembinski

README.rst

@@ -34,7 +34,7 @@ pyhepmc is a hand-crafted mapping of C++ code to Python, `see documentation for
 - An alternative Numpy API accelerates event processing up to **70x** compared to the standard API.
 - The public API is fully documented with Python docstrings.
 - Objects are inspectable in Jupyter notebooks (have useful ``repr`` strings).
-- Events render as graphs in Jupyter notebooks (see next item). 
+- Events render as graphs in Jupyter notebooks (see next item).
 
 **pyhepmc supports visualizations powered by graphviz**
 

src/pyhepmc/io.py

@@ -12,10 +12,10 @@
 from __future__ import annotations
 from ._core import (
     GenEvent,
-    ReaderAscii,
-    ReaderAsciiHepMC2,
-    ReaderLHEF,
-    ReaderHEPEVT,
+    ReaderAscii as ReaderAsciiBase,
+    ReaderAsciiHepMC2 as ReaderAsciiHepMC2Base,
+    ReaderLHEF as ReaderLHEFBase,
+    ReaderHEPEVT as ReaderHEPEVTBase,
     WriterAscii,
     WriterAsciiHepMC2,
     WriterHEPEVT,
@@ -38,25 +38,6 @@
 ]
 
 
-class _Iter:
-    def __init__(self, parent: Any):
-        self.parent = parent
-
-    def __next__(self) -> GenEvent:
-        evt = GenEvent()
-        success = self.parent.read_event(evt)
-        if self.parent.failed():
-            if success:  # indicates EOF
-                raise StopIteration
-            raise IOError("error reading event")
-        return evt
-
-    def __iter__(self) -> "_Iter":
-        return self
-
-    next = __next__
-
-
 def _enter(self: Any) -> Any:
     return self
 
@@ -71,49 +52,68 @@ def _exit_flush(self: Any, type: Exception, value: str, tb: Any) -> bool:
     return False
 
 
-def _iter(self: Any) -> _Iter:
-    return _Iter(self)
-
-
-def _read(self: Any) -> Optional[GenEvent]:
-    evt = GenEvent()
-    success = self.read_event(evt)
-    if self.failed():
-        if success:  # indicates EOF
-            return None
-        raise IOError("error reading event")
-    return evt if success else None
-
-
 def _read_event_lhef_patch(self: Any, evt: GenEvent) -> bool:
     failed = self._read_event_unpatched(evt)
     if failed and self.failed():  # probably EOF
         return True
     return not failed
 
 
+class _Iter:
+    def __init__(self, reader: Any):
+        self.reader = reader
+
+    def __next__(self) -> GenEvent:
+        evt = self.reader.read()
+        if evt is None:
+            raise StopIteration
+        return evt
+
+    def __iter__(self) -> "_Iter":
+        return self
+
+    next = __next__
+
+
+class ReaderMixin:
+    def read(self) -> Optional[GenEvent]:
+        assert hasattr(self, "failed")
+        assert hasattr(self, "read_event")
+        if self.failed():
+            # usually this means EOF was reached previously
+            return None
+        evt = GenEvent()
+        success = self.read_event(evt)
+        # workaround for bug in HepMC3, which reports success even if
+        # the next section of the file does not contain any event data
+        if len(evt.particles) == 0:
+            success = False
+        return evt if success else None
+
+    def __iter__(self: Any) -> _Iter:
+        return _Iter(self)
+
+    __enter__ = _enter
+    __exit__ = _exit_close
+
+
 # add contextmanager interface to IO classes
-ReaderAscii.__enter__ = _enter
-ReaderAscii.__exit__ = _exit_close
-ReaderAscii.__iter__ = _iter
-ReaderAscii.read = _read
-
-ReaderAsciiHepMC2.__enter__ = _enter
-ReaderAsciiHepMC2.__exit__ = _exit_close
-ReaderAsciiHepMC2.__iter__ = _iter
-ReaderAsciiHepMC2.read = _read
-
-ReaderLHEF.__enter__ = _enter
-ReaderLHEF.__exit__ = _exit_close
-ReaderLHEF.__iter__ = _iter
-ReaderLHEF._read_event_unpatched = ReaderLHEF.read_event
-ReaderLHEF.read_event = _read_event_lhef_patch
-ReaderLHEF.read = _read
-
-ReaderHEPEVT.__enter__ = _enter
-ReaderHEPEVT.__exit__ = _exit_close
-ReaderHEPEVT.__iter__ = _iter
-ReaderHEPEVT.read = _read
+class ReaderAscii(ReaderAsciiBase, ReaderMixin):  # type:ignore
+    pass
+
+
+class ReaderAsciiHepMC2(ReaderAsciiHepMC2Base, ReaderMixin):  # type:ignore
+    pass
+
+
+class ReaderLHEF(ReaderLHEFBase, ReaderMixin):  # type:ignore
+    _read_event_unpatched = ReaderLHEFBase.read_event
+    read_event = _read_event_lhef_patch
+
+
+class ReaderHEPEVT(ReaderHEPEVTBase, ReaderMixin):  # type:ignore
+    pass
+
 
 WriterAscii.__enter__ = _enter
 WriterAscii.__exit__ = _exit_close
@@ -256,6 +256,7 @@ def open_file() -> Any:
                 self._file = open_file()
             self._ios = pyiostream(self._file)
 
+            Reader: Optional[Any] = None
             if format is None:
                 # auto-detect
                 if not self._file.seekable():
@@ -305,9 +306,7 @@ def open_file() -> Any:
         else:
             raise ValueError(f"mode must be 'r' or 'w', got {mode!r}")
 
-    def __enter__(self) -> HepMCFile:
-        return self
-
+    __enter__ = _enter
     __exit__ = _exit_close
 
     def __iter__(self) -> Any:
@@ -331,7 +330,7 @@ def write(self, event: GenEvent) -> None:
 
     def close(self) -> None:
         if self._reader:
-            self._reader.close()
+            self._reader.close()  # type:ignore
         if self._writer:
             self._writer.close()
         self._ios.flush()