Troubles opening a tree with a custom-class branch

[mattiasoldani]

Howdy!

I'm using uproot 4.3.7 to open a ROOT file (of which a sample is attached: TEST230417_LOCAL_SMALL.zip) and access a tree ("Z") that is inside it. One of the branches of this tree, "Cluster[6]", should be an array of 6 custom zCluster objects for each event. Each zCluster object includes several members, mostly float and TVector3 entries. I checked with ROOT and all the branch dimensions are as expected, as seen in the TBrowser screenshots below: the tree contains 116 events, whereas all the Cluster[6] numerical subbranches contain 116*6 = 696 entries.

If I try to open the same file in uproot via

file = uproot.open("TEST230417_LOCAL_SMALL.root")
tree = file["Z"]

I get the correct tree, i.e. tree.num_entries outputs 116 and tree.show() outputs

name                 | typename                 | interpretation                
---------------------+--------------------------+-------------------------------
Event                | zEvent                   | AsGroup(<TBranchElement 'Ev...
Event/TObject        | (group of fUniqueID:u... | AsGroup(<TBranchElement 'TO...
Event/TObject/fUn... | uint32_t                 | AsDtype('>u4')
Event/TObject/fBits  | uint32_t                 | AsDtype('>u4')
Event/iEv            | int32_t                  | AsDtype('>i4')
Event/W              | double                   | AsDtype('>f8')
Event/pK             | TLorentzVector           | AsStridedObjects(Model_TLor...
Event/xK             | TVector3                 | AsStridedObjects(Model_TVec...
Event/iPair          | int32_t                  | AsDtype('>i4')
Event/ig[6]          | int32_t[6]               | AsDtype("('>i4', (6,))")
Event/igClus[6]      | int32_t[6]               | AsDtype("('>i4', (6,))")
Event/pg[6]          | TLorentzVector           | AsStridedObjects(Model_TLor...
Event/Hit[6]         | zHit                     | AsStridedObjects(Model_zHit...
Event/nCluster       | int32_t                  | AsDtype('>i4')
Event/Cluster[6]     | zCluster                 | AsObjects(Model_zCluster)
Event/Vertex         | zVertex                  | AsGroup(<TBranchElement 'Ve...
Event/Vertex/Vert... | int32_t                  | AsDtype('>i4')
Event/Vertex/Vert... | int32_t[2]               | AsDtype("('>i4', (2,))")
Event/Vertex/Vert... | int32_t                  | AsDtype('>i4')
Event/Vertex/Vert... | TVector3                 | AsStridedObjects(Model_TVec...
Event/Vertex/Vert... | TLorentzVector           | AsStridedObjects(Model_TLor...
Event/Vertex/Vert... | TVector3                 | AsStridedObjects(Model_TVec...
Event/Vertex/Vert... | TLorentzVector           | AsStridedObjects(Model_TLor...
Event/Vertex/Vert... | int32_t                  | AsDtype('>i4')

However, tree.arrays(["Event/Cluster[6]"]) raises a CannotBeAwkward: classes that can contain members of the same type cannot be Awkward Arrays because the depth of instances is unbounded error and tree.arrays(["Event/Cluster[6]"], library="np") outputs a dictionary of 116 entries whose values are single zCluster objects, i.e. one single zCluster per event. Accessing any of them via e.g. tree["Event/Cluster[6]"].arrays(library="np")["Cluster[6]"][0].all_members seems to correctly output all the information regarding one single cluster, i.e.

{'Status': 1,
 'Bin': 3,
 'iSubdet': 0,
 'Detected': 0,
 'Primary': 0,
 'Used': 0,
 'Charge': 1,
 'Eff': 1.0,
 'xTrue': <TVector3 (version 3) at 0x7feb5830dc40>,
 'xRec': <TVector3 (version 3) at 0x7feb58314280>,
 'PosRes': 0.015788506169338613,
 'xTrue1': <TVector3 (version 3) at 0x7feb58314190>,
 'xTrue2': <TVector3 (version 3) at 0x7feb58314040>,
 'xPre1': <TVector3 (version 3) at 0x7feb58314490>,
 'xPre2': <TVector3 (version 3) at 0x7feb583145b0>,
 'PreRes': 0.06397004001907128,
 'ETrue': 0.07086701659225284,
 'ERec': 0.03220946465343243,
 'ERes': 0.09040274561889668,
 'Converted': 1,
 'dXdZTrue': 0.13310240197699721,
 'dYdZTrue': -0.34894589067080145,
 'nHits': 1,
 'HitList': array([0, 1, 1, 1, 1, 1], dtype=int32),
 'nConverted': 0}

but where are the other 5 clusters? Am I missing something trivial?

Thank you very much!