PFNano additions for NanoEvents

coffea/nanoevents/__init__.py

@@ -2,11 +2,17 @@
 
 """
 from coffea.nanoevents.factory import NanoEventsFactory
-from coffea.nanoevents.schemas import BaseSchema, NanoAODSchema, TreeMakerSchema
+from coffea.nanoevents.schemas import (
+    BaseSchema,
+    NanoAODSchema,
+    PFNanoAODSchema,
+    TreeMakerSchema,
+)
 
 __all__ = [
     "NanoEventsFactory",
     "BaseSchema",
     "NanoAODSchema",
+    "PFNanoAODSchema",
     "TreeMakerSchema",
 ]

coffea/nanoevents/factory.py

@@ -110,7 +110,9 @@ def from_root(
         )
         uuidpfn = {partition_key[0]: tree.file.file_path}
         mapping = UprootSourceMapping(
-            TrivialUprootOpener(uuidpfn, uproot_options), cache={}, access_log=access_log
+            TrivialUprootOpener(uuidpfn, uproot_options),
+            cache={},
+            access_log=access_log,
         )
         mapping.preload_column_source(partition_key[0], partition_key[1], tree)
 
@@ -208,9 +210,11 @@ def from_parquet(
             TrivialParquetOpener(uuidpfn, parquet_options), access_log=access_log
         )
 
-        format_ = 'parquet'
-        if 'ceph_config_path' in rados_parquet_options:
-            format_ = ds.RadosParquetFileFormat(rados_parquet_options['ceph_config_path'].encode())
+        format_ = "parquet"
+        if "ceph_config_path" in rados_parquet_options:
+            format_ = ds.RadosParquetFileFormat(
+                rados_parquet_options["ceph_config_path"].encode()
+            )
 
         dataset = ds.dataset(file, schema=table_file.schema_arrow, format=format_)
 
@@ -340,13 +344,15 @@ def _from_mapping(
                 Arbitrary metadata to add to the `base.NanoEvents` object
 
         """
-        if not issubclass(schemaclass, BaseSchema):
-            raise RuntimeError("Invalid schema type")
         if persistent_cache is not None:
             mapping = CachedMapping(persistent_cache, mapping)
         if metadata is not None:
             base_form["parameters"]["metadata"] = metadata
+        if not callable(schemaclass):
+            raise ValueError("Invalid schemaclass type")
         schema = schemaclass(base_form)
+        if not isinstance(schema, BaseSchema):
+            raise RuntimeError("Invalid schema type")
         return cls(schema, mapping, tuple_to_key(partition_key), cache=runtime_cache)
 
     def __len__(self):

coffea/nanoevents/methods/base.py

@@ -29,6 +29,10 @@ class NanoCollection:
     and other advanced mixin types.
     """
 
+    def _collection_name(self):
+        """The name of the collection (i.e. the field under events where it is found)"""
+        return self.layout.purelist_parameter("collection_name")
+
     def _getlistarray(self):
         """Do some digging to find the initial listarray"""
 

coffea/nanoevents/methods/nanoaod.py

@@ -298,6 +298,12 @@ def matched_muons(self):
     def matched_gen(self):
         return self._events().GenJet._apply_global_index(self.genJetIdxG)
 
+    @property
+    def constituents(self):
+        if "pFCandsIdxG" not in self.fields:
+            raise RuntimeError("PF candidates are only available for PFNano")
+        return self._events().JetPFCands._apply_global_index(self.pFCandsIdxG)
+
 
 _set_repr_name("Jet")
 
@@ -336,6 +342,12 @@ def subjets(self):
     def matched_gen(self):
         return self._events().GenJetAK8._apply_global_index(self.genJetAK8IdxG)
 
+    @property
+    def constituents(self):
+        if "pFCandsIdxG" not in self.fields:
+            raise RuntimeError("PF candidates are only available for PFNano")
+        return self._events().FatJetPFCands._apply_global_index(self.pFCandsIdxG)
+
 
 _set_repr_name("FatJet")
 
@@ -351,6 +363,99 @@ def r(self):
 
 _set_repr_name("MissingET")
 
+
+@awkward.mixin_class(behavior)
+class Vertex(vector.ThreeVector, base.NanoCollection):
+    """NanoAOD vertex object"""
+
+    pass
+
+
+_set_repr_name("Vertex")
+
+
+@awkward.mixin_class(behavior)
+class SecondaryVertex(Vertex):
+    """NanoAOD secondary vertex object"""
+
+    @property
+    def p4(self):
+        """4-momentum vector of tracks associated to this SV"""
+        return awkward.zip(
+            {
+                "pt": self["pt"],
+                "eta": self["eta"],
+                "phi": self["phi"],
+                "mass": self["mass"],
+            },
+            with_name="PtEtaPhiMLorentzVector",
+        )
+
+
+_set_repr_name("SecondaryVertex")
+
+
+@awkward.mixin_class(behavior)
+class AssociatedPFCand(base.NanoCollection):
+    """PFNano PF candidate to jet association object"""
+
+    collection_map = {
+        "JetPFCands": ("Jet", "PFCands"),
+        "FatJetPFCands": ("FatJet", "PFCands"),
+        "GenJetCands": ("GenJet", "GenCands"),
+        "GenFatJetCands": ("GenJetAK8", "GenCands"),
+    }
+
+    @property
+    def jet(self):
+        collection = self._events()[self.collection_map[self._collection_name()][0]]
+        return collection._apply_global_index(self.jetIdxG)
+
+    @property
+    def pf(self):
+        collection = self._events()[self.collection_map[self._collection_name()][1]]
+        return collection._apply_global_index(self.pFCandsIdxG)
+
+
+_set_repr_name("AssociatedPFCand")
+
+
+@awkward.mixin_class(behavior)
+class AssociatedSV(base.NanoCollection):
+    """PFNano secondary vertex to jet association object"""
+
+    collection_map = {
+        "JetSVs": ("Jet", "SV"),
+        "FatJetSVs": ("FatJet", "SV"),
+        # these two are unclear
+        "GenJetSVs": ("GenJet", "SV"),
+        "GenFatJetSVs": ("GenJetAK8", "SV"),
+    }
+
+    @property
+    def jet(self):
+        collection = self._events()[self.collection_map[self._collection_name()][0]]
+        return collection._apply_global_index(self.jetIdxG)
+
+    @property
+    def sv(self):
+        collection = self._events()[self.collection_map[self._collection_name()][1]]
+        return collection._apply_global_index(self.sVIdxG)
+
+
+_set_repr_name("AssociatedSV")
+
+
+@awkward.mixin_class(behavior)
+class PFCand(candidate.PtEtaPhiMCandidate, base.NanoCollection):
+    """PFNano particle flow candidate object"""
+
+    pass
+
+
+_set_repr_name("PFCand")
+
+
 __all__ = [
     "PtEtaPhiMCollection",
     "GenParticle",
@@ -363,4 +468,9 @@ def r(self):
     "Jet",
     "FatJet",
     "MissingET",
+    "Vertex",
+    "SecondaryVertex",
+    "AssociatedPFCand",
+    "AssociatedSV",
+    "PFCand",
 ]

coffea/nanoevents/schemas/__init__.py

@@ -1,5 +1,5 @@
 from .base import BaseSchema
-from .nanoaod import NanoAODSchema
+from .nanoaod import NanoAODSchema, PFNanoAODSchema
 from .treemaker import TreeMakerSchema
 
-__all__ = ["BaseSchema", "NanoAODSchema", "TreeMakerSchema"]
+__all__ = ["BaseSchema", "NanoAODSchema", "PFNanoAODSchema", "TreeMakerSchema"]