Resolve "uproot4 integration"

Closes #58 (closed)

km3io/rootio.py

@@ -87,7 +87,7 @@ class EventReader:
 
     def _initialise_keys(self):
         skip_keys = set(self.skip_keys)
-        all_keys  = set(self._fobj[self.event_path].keys())
+        all_keys = set(self._fobj[self.event_path].keys())
         toplevel_keys = set(k.split("/")[0] for k in all_keys)
         valid_aliases = {}
         for fromkey, tokey in self.aliases.items():
@@ -169,15 +169,46 @@ class EventReader:
                 if from_field in branch[key].keys():
                     fields.append(to_field)
             log.debug(fields)
-            return Branch(branch[key], fields, self.nested_branches[key], self._index_chain)
+            return Branch(
+                branch[key], fields, self.nested_branches[key], self._index_chain
+            )
         else:
-            return unfold_indices(branch[self.aliases.get(key, key)].array(), self._index_chain)
+            return unfold_indices(
+                branch[self.aliases.get(key, key)].array(), self._index_chain
+            )
 
-    def __iter__(self):
-        self._events = self._event_generator()
+    def __iter__(self, chunkwise=False):
+        self._events = self._event_generator(chunkwise=chunkwise)
         return self
 
-    def _event_generator(self):
+    def _get_iterator_limits(self):
+        """Determines start and stop, used for event iteration"""
+        if len(self._index_chain) > 1:
+            raise NotImplementedError(
+                "iteration is currently not supported with nested slices"
+            )
+        if self._index_chain:
+            s = self._index_chain[0]
+            if not isinstance(s, slice):
+                raise NotImplementedError("iteration is only supported with slices")
+            if s.step is None or s.step == 1:
+                start = s.start
+                stop = s.stop
+            else:
+                raise NotImplementedError(
+                    "iteration is only supported with single steps"
+                )
+        else:
+            start = None
+            stop = None
+        return start, stop
+
+    def _event_generator(self, chunkwise=False):
+        start, stop = self._get_iterator_limits()
+
+        if chunkwise:
+            raise NotImplementedError("iterating over chunks is not implemented yet")
+
         events = self._fobj[self.event_path]
         group_count_keys = set(
             k for k in self.keys() if k.startswith("n_")
@@ -195,7 +226,11 @@ class EventReader:
         log.debug("keys: %s", keys)
         log.debug("aliases: %s", self.aliases)
         events_it = events.iterate(
-            keys, aliases=self.aliases, step_size=self._step_size
+            keys,
+            aliases=self.aliases,
+            step_size=self._step_size,
+            entry_start=start,
+            entry_stop=stop,
         )
         nested = []
         nested_keys = (
@@ -208,6 +243,8 @@ class EventReader:
                     self.nested_branches[key].keys(),
                     aliases=self.nested_branches[key],
                     step_size=self._step_size,
+                    entry_start=start,
+                    entry_stop=stop,
                 )
             )
         group_counts = {}
@@ -276,6 +313,7 @@ class EventReader:
 
 class Branch:
     """Helper class for nested branches likes tracks/hits"""
+
     def __init__(self, branch, fields, aliases, index_chain):
         self._branch = branch
         self.fields = fields
@@ -284,7 +322,9 @@ class Branch:
 
     def __getattr__(self, attr):
         if attr not in self._aliases:
-            raise AttributeError(f"No field named {attr}. Available fields: {self.fields}")
+            raise AttributeError(
+                f"No field named {attr}. Available fields: {self.fields}"
+            )
         key = self._aliases[attr]
 
         if self._index_chain:
@@ -293,7 +333,9 @@ class Branch:
                 # optimise single-element and slice lookups
                 start = idx0
                 stop = idx0 + 1
-                arr = ak.flatten(self._branch[key].array(entry_start=start, entry_stop=stop))
+                arr = ak.flatten(
+                    self._branch[key].array(entry_start=start, entry_stop=stop)
+                )
                 return unfold_indices(arr, self._index_chain[1:])
             if isinstance(idx0, slice):
                 if idx0.step is None or idx0.step == 1:
@@ -305,7 +347,9 @@ class Branch:
         return unfold_indices(self._branch[key].array(), self._index_chain)
 
     def __getitem__(self, key):
-        return self.__class__(self._branch, self.fields, self._aliases, self._index_chain + [key])
+        return self.__class__(
+            self._branch, self.fields, self._aliases, self._index_chain + [key]
+        )
 
     def __len__(self):
         if not self._index_chain: