uproot 5.5.0__tar.gz → 5.5.2__tar.gz

{uproot-5.5.0 → uproot-5.5.2}/.all-contributorsrc

@@ -605,6 +605,15 @@
       "contributions": [
         "code"
       ]
+    },
+    {
+      "login": "ariostas",
+      "name": "Andres Rios Tascon",
+      "avatar_url": "https://avatars.githubusercontent.com/u/7596837?v=4",
+      "profile": "http://www.ariostas.com",
+      "contributions": [
+        "code"
+      ]
     }
   ],
   "contributorsPerLine": 7,

{uproot-5.5.0 → uproot-5.5.2}/.github/workflows/build-test.yml

@@ -90,7 +90,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
           allow-prereleases: true
 
-      - uses: astral-sh/setup-uv@v3
+      - uses: astral-sh/setup-uv@v5
 
       - name: Pip install the package
         run: uv pip install --system .[test,dev]

{uproot-5.5.0 → uproot-5.5.2}/.github/workflows/deploy.yml

@@ -36,10 +36,10 @@ jobs:
         run: ls -lha dist/
 
       - name: Generate artifact attestation for sdist and wheel
-        uses: actions/attest-build-provenance@1c608d11d69870c2092266b3f9a6f3abbf17002c # v1.4.3
+        uses: actions/attest-build-provenance@520d128f165991a6c774bcb264f323e3d70747f4 # v2.2.0
         with:
           subject-path: dist/uproot-*
 
       - uses: pypa/gh-action-pypi-publish@release/v1
         with:
-          password: ${{ secrets.pypi_password }}
+          print-hash: true

{uproot-5.5.0 → uproot-5.5.2}/.pre-commit-config.yaml

@@ -18,19 +18,19 @@ repos:
       - id: trailing-whitespace
 
   - repo: https://github.com/psf/black-pre-commit-mirror
-    rev: 24.10.0
+    rev: 25.1.0
     hooks:
       - id: black
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.7.2
+    rev: v0.9.6
     hooks:
       - id: ruff
         args: [--fix, --show-fixes]
 
 
   - repo: https://github.com/asottile/pyupgrade
-    rev: v3.19.0
+    rev: v3.19.1
     hooks:
       - id: pyupgrade
         args: [--py38-plus]

{uproot-5.5.0 → uproot-5.5.2}/CITATION.cff

@@ -8,7 +8,7 @@ authors:
   given-names: "Jim"
   affiliation: "Princeton University"
   orcid: "https://orcid.org/0000-0002-6649-343X"
-  email: "pivarski@princeton.edu"
+  email: "jpivarski@gmail.com"
 - family-names: "Schreiner"
   given-names: "Henry"
   affiliation: "Princeton University"

{uproot-5.5.0 → uproot-5.5.2}/PKG-INFO

@@ -1,10 +1,10 @@
-Metadata-Version: 2.3
+Metadata-Version: 2.4
 Name: uproot
-Version: 5.5.0
+Version: 5.5.2
 Summary: ROOT I/O in pure Python and NumPy.
 Project-URL: Download, https://github.com/scikit-hep/uproot5/releases
 Project-URL: Homepage, https://github.com/scikit-hep/uproot5
-Author-email: Jim Pivarski <pivarski@princeton.edu>
+Author-email: Jim Pivarski <jpivarski@gmail.com>
 License-Expression: BSD-3-Clause
 License-File: LICENSE
 Classifier: Development Status :: 5 - Production/Stable
@@ -40,7 +40,7 @@ Requires-Dist: xxhash
 Provides-Extra: dev
 Requires-Dist: awkward-pandas; extra == 'dev'
 Requires-Dist: boost-histogram>=0.13; extra == 'dev'
-Requires-Dist: dask-awkward>=2023.12.1; extra == 'dev'
+Requires-Dist: dask-awkward>=2025.2.0; extra == 'dev'
 Requires-Dist: dask[array,distributed]; extra == 'dev'
 Requires-Dist: hist>=1.2; extra == 'dev'
 Requires-Dist: pandas; extra == 'dev'
@@ -67,7 +67,6 @@ Requires-Dist: pytest-pyodide; extra == 'test-pyodide'
 Requires-Dist: pytest-timeout; extra == 'test-pyodide'
 Requires-Dist: pytest>=6; extra == 'test-pyodide'
 Requires-Dist: scikit-hep-testdata; extra == 'test-pyodide'
-Requires-Dist: selenium<=4.25.0; extra == 'test-pyodide'
 Provides-Extra: xrootd
 Requires-Dist: fsspec-xrootd; extra == 'xrootd'
 Description-Content-Type: text/markdown
@@ -267,6 +266,7 @@ Thanks especially to the gracious help of Uproot contributors (including the [or
     <tr>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/MatousVozak"><img src="https://avatars.githubusercontent.com/u/33348191?v=4?s=100" width="100px;" alt="MatousVozak"/><br /><sub><b>MatousVozak</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=MatousVozak" title="Code">💻</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/rossodisera"><img src="https://avatars.githubusercontent.com/u/46961664?v=4?s=100" width="100px;" alt="Andrea Serafini"/><br /><sub><b>Andrea Serafini</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=rossodisera" title="Code">💻</a></td>
+      <td align="center" valign="top" width="14.28%"><a href="http://www.ariostas.com"><img src="https://avatars.githubusercontent.com/u/7596837?v=4?s=100" width="100px;" alt="Andres Rios Tascon"/><br /><sub><b>Andres Rios Tascon</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=ariostas" title="Code">💻</a></td>
     </tr>
   </tbody>
 </table>

{uproot-5.5.0 → uproot-5.5.2}/README.md

@@ -193,6 +193,7 @@ Thanks especially to the gracious help of Uproot contributors (including the [or
     <tr>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/MatousVozak"><img src="https://avatars.githubusercontent.com/u/33348191?v=4?s=100" width="100px;" alt="MatousVozak"/><br /><sub><b>MatousVozak</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=MatousVozak" title="Code">💻</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/rossodisera"><img src="https://avatars.githubusercontent.com/u/46961664?v=4?s=100" width="100px;" alt="Andrea Serafini"/><br /><sub><b>Andrea Serafini</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=rossodisera" title="Code">💻</a></td>
+      <td align="center" valign="top" width="14.28%"><a href="http://www.ariostas.com"><img src="https://avatars.githubusercontent.com/u/7596837?v=4?s=100" width="100px;" alt="Andres Rios Tascon"/><br /><sub><b>Andres Rios Tascon</b></sub></a><br /><a href="https://github.com/scikit-hep/uproot5/commits?author=ariostas" title="Code">💻</a></td>
     </tr>
   </tbody>
 </table>

{uproot-5.5.0 → uproot-5.5.2}/pyproject.toml

@@ -10,7 +10,7 @@ max-complexity = 100
 
 [project]
 authors = [
-  {name = "Jim Pivarski", email = "pivarski@princeton.edu"}
+  {name = "Jim Pivarski", email = "jpivarski@gmail.com"}
 ]
 classifiers = [
   "Development Status :: 5 - Production/Stable",
@@ -57,7 +57,7 @@ requires-python = ">=3.9"
 [project.optional-dependencies]
 dev = [
   "boost_histogram>=0.13",
-  "dask-awkward>=2023.12.1",
+  "dask-awkward>=2025.2.0",
   "dask[array,distributed]",
   "hist>=1.2",
   "pandas",
@@ -82,7 +82,6 @@ test = [
 ]
 test-pyodide = [
   "pytest>=6",
-  "selenium<=4.25.0",  # unpin once >4.26.0 is available
   "pytest-pyodide",
   "pytest-timeout",
   "scikit-hep-testdata"

{uproot-5.5.0 → uproot-5.5.2}/src/uproot/_dask.py

@@ -383,6 +383,8 @@ def _dask_array_from_map(
     **kwargs,
 ):
     dask = uproot.extras.dask()
+    _dask_uses_tasks = hasattr(dask, "_task_spec")
+
     da = uproot.extras.dask_array()
     if not callable(func):
         raise ValueError("`func` argument must be `callable`")
@@ -446,14 +448,22 @@ def _dask_array_from_map(
         produces_tasks=produces_tasks,
     )
 
-    dsk = dask.blockwise.Blockwise(
-        output=name,
-        output_indices="i",
-        dsk={name: (io_func, dask.blockwise.blockwise_token(0))},
-        indices=[(io_arg_map, "i")],
-        numblocks={},
-        annotations=None,
-    )
+    blockwise_kwargs = {
+        "output": name,
+        "output_indices": "i",
+        "indices": [(io_arg_map, "i")],
+        "numblocks": {},
+        "annotations": None,
+    }
+
+    if _dask_uses_tasks:
+        blockwise_kwargs["task"] = dask._task_spec.Task(
+            name, io_func, dask._task_spec.TaskRef(dask.blockwise.blockwise_token(0))
+        )
+    else:
+        blockwise_kwargs["dsk"] = {name: (io_func, dask.blockwise.blockwise_token(0))}
+
+    dsk = dask.blockwise.Blockwise(**blockwise_kwargs)
 
     hlg = dask.highlevelgraph.HighLevelGraph.from_collections(name, dsk)
     return da.core.Array(hlg, name, chunks, dtype=dtype)

uproot-5.5.2/src/uproot/const.py

@@ -0,0 +1,287 @@
+# BSD 3-Clause License; see https://github.com/scikit-hep/uproot5/blob/main/LICENSE
+
+"""
+This module defines integer constants used by serialization and deserialization routines.
+"""
+from __future__ import annotations
+
+from enum import IntEnum
+
+import numpy
+
+# determines when a file is "big"
+kStartBigFile = 2000000000
+
+# used in unmarshaling
+kByteCountMask = numpy.int64(0x40000000)
+kByteCountVMask = numpy.int64(0x4000)
+kClassMask = numpy.int64(0x80000000)
+kNewClassTag = numpy.int64(0xFFFFFFFF)
+
+kIsOnHeap = numpy.uint32(0x01000000)
+kNotDeleted = numpy.uint32(0x02000000)
+kMustCleanup = numpy.uint32(1 << 3)
+kIsReferenced = numpy.uint32(1 << 4)
+
+kMapOffset = 2
+
+# not used?
+kNullTag = 0
+kZombie = numpy.uint32(0x04000000)
+kBitMask = numpy.uint32(0x00FFFFFF)
+kDisplacementMask = numpy.uint32(0xFF000000)
+
+############# core/zip/inc/Compression.h
+
+kZLIB = 1
+kLZMA = 2
+kOldCompressionAlgo = 3
+kLZ4 = 4
+kZSTD = 5
+kUndefinedCompressionAlgorithm = 6
+
+############# constants for streamers
+
+kBase = 0
+kChar = 1
+kShort = 2
+kInt = 3
+kLong = 4
+kFloat = 5
+kCounter = 6
+kCharStar = 7
+kDouble = 8
+kDouble32 = 9
+kLegacyChar = 10
+kUChar = 11
+kUShort = 12
+kUInt = 13
+kULong = 14
+kBits = 15
+kLong64 = 16
+kULong64 = 17
+kBool = 18
+kFloat16 = 19
+kOffsetL = 20
+kOffsetP = 40
+kObject = 61
+kAny = 62
+kObjectp = 63
+kObjectP = 64
+kTString = 65
+kTObject = 66
+kTNamed = 67
+kAnyp = 68
+kAnyP = 69
+kAnyPnoVT = 70
+kSTLp = 71
+
+kSkip = 100
+kSkipL = 120
+kSkipP = 140
+
+kConv = 200
+kConvL = 220
+kConvP = 240
+
+kSTL = 300
+kSTLstring = 365
+
+kStreamer = 500
+kStreamLoop = 501
+
+############# constants from core/foundation/inc/ESTLType.h
+
+kNotSTL = 0
+kSTLvector = 1
+kSTLlist = 2
+kSTLdeque = 3
+kSTLmap = 4
+kSTLmultimap = 5
+kSTLset = 6
+kSTLmultiset = 7
+kSTLbitset = 8
+kSTLforwardlist = 9
+kSTLunorderedset = 10
+kSTLunorderedmultiset = 11
+kSTLunorderedmap = 12
+kSTLunorderedmultimap = 13
+kSTLend = 14
+kSTLany = 300
+
+############# IOFeatures
+
+kGenerateOffsetMap = numpy.uint8(1)
+
+############# other
+
+kStreamedMemberWise = numpy.uint16(1 << 14)
+
+############ RNTuple https://github.com/root-project/root/blob/0b9cdbcfd326ba50ee6c2f202675656129eafbe7/tree/ntuple/v7/doc/BinaryFormatSpecification.md
+rntuple_col_num_to_dtype_dict = {
+    0x00: "bit",
+    0x01: "uint8",  # uninterpreted byte
+    0x02: "uint8",  # char
+    0x03: "int8",
+    0x04: "uint8",
+    0x05: "int16",
+    0x06: "uint16",
+    0x07: "int32",
+    0x08: "uint32",
+    0x09: "int64",
+    0x0A: "uint64",
+    0x0B: "float16",
+    0x0C: "float32",
+    0x0D: "float64",
+    0x0E: "uint32",  # Index32
+    0x0F: "uint64",  # Index64
+    0x10: "switch",  # Switch: (uint64, uint32)
+    0x11: "int16",  # SplitInt16: split + zigzag encoding
+    0x12: "uint16",  # SplitUInt16: split encoding
+    0x13: "int32",  # SplitInt32: split + zigzag encoding
+    0x14: "uint32",  # SplitUInt32: split encoding
+    0x15: "int64",  # SplitInt64: split + zigzag encoding
+    0x16: "uint64",  # SplitUInt64: split encoding
+    0x17: "float16",  # SplitReal16: split encoding
+    0x18: "float32",  # SplitReal32: split encoding
+    0x19: "float64",  # SplitReal64: split encoding
+    0x1A: "uint32",  # SplitIndex32: split + delta encoding
+    0x1B: "uint64",  # SplitIndex64: split + delta encoding
+    0x1C: "real32trunc",  # Real32Trunc: float32 with truncated mantissa
+    0x1D: "real32quant",  # Real32Quant: float32 with quantized integer representation
+}
+rntuple_col_num_to_size_dict = {
+    0x00: 1,
+    0x01: 8,
+    0x02: 8,
+    0x03: 8,
+    0x04: 8,
+    0x05: 16,
+    0x06: 16,
+    0x07: 32,
+    0x08: 32,
+    0x09: 64,
+    0x0A: 64,
+    0x0B: 16,
+    0x0C: 32,
+    0x0D: 64,
+    0x0E: 32,
+    0x0F: 64,
+    0x10: 96,
+    0x11: 16,
+    0x12: 16,
+    0x13: 32,
+    0x14: 32,
+    0x15: 64,
+    0x16: 64,
+    0x17: 16,
+    0x18: 32,
+    0x19: 64,
+    0x1A: 32,
+    0x1B: 64,
+    0x1C: 31,  # variable from 10 to 31
+    0x1D: 32,  # variable from 1 to 32
+}
+rntuple_col_type_to_num_dict = {
+    "bit": 0x00,
+    "byte": 0x01,
+    "char": 0x02,
+    "int8": 0x03,
+    "uint8": 0x04,
+    "int16": 0x05,
+    "uint16": 0x06,
+    "int32": 0x07,
+    "uint32": 0x08,
+    "int64": 0x09,
+    "uint64": 0x0A,
+    "real16": 0x0B,
+    "real32": 0x0C,
+    "real64": 0x0D,
+    "index32": 0x0E,
+    "index64": 0x0F,
+    "switch": 0x10,
+    "splitint16": 0x11,
+    "splituint16": 0x12,
+    "splitint32": 0x13,
+    "splituint32": 0x14,
+    "splitint64": 0x15,
+    "splituint64": 0x16,
+    "splitreal16": 0x17,
+    "splitreal32": 0x18,
+    "splitreal64": 0x19,
+    "splitindex32": 0x1A,
+    "splitindex64": 0x1B,
+    "real32trunc": 0x1C,
+    "real32quant": 0x1D,
+}
+rntuple_index_types = (
+    rntuple_col_type_to_num_dict["index32"],
+    rntuple_col_type_to_num_dict["index64"],
+    rntuple_col_type_to_num_dict["splitindex32"],
+    rntuple_col_type_to_num_dict["splitindex64"],
+)
+rntuple_split_types = (
+    rntuple_col_type_to_num_dict["splitint16"],
+    rntuple_col_type_to_num_dict["splituint16"],
+    rntuple_col_type_to_num_dict["splitint32"],
+    rntuple_col_type_to_num_dict["splituint32"],
+    rntuple_col_type_to_num_dict["splitint64"],
+    rntuple_col_type_to_num_dict["splituint64"],
+    rntuple_col_type_to_num_dict["splitreal16"],
+    rntuple_col_type_to_num_dict["splitreal32"],
+    rntuple_col_type_to_num_dict["splitreal64"],
+    rntuple_col_type_to_num_dict["splitindex32"],
+    rntuple_col_type_to_num_dict["splitindex64"],
+)
+rntuple_zigzag_types = (
+    rntuple_col_type_to_num_dict["splitint16"],
+    rntuple_col_type_to_num_dict["splitint32"],
+    rntuple_col_type_to_num_dict["splitint64"],
+)
+rntuple_delta_types = (
+    rntuple_col_type_to_num_dict["splitindex32"],
+    rntuple_col_type_to_num_dict["splitindex64"],
+)
+rntuple_custom_float_types = (
+    rntuple_col_type_to_num_dict["real32trunc"],
+    rntuple_col_type_to_num_dict["real32quant"],
+)
+
+
+class RNTupleLocatorType(IntEnum):
+    STANDARD = 0x00
+    LARGE = 0x01
+
+
+class RNTupleEnvelopeType(IntEnum):
+    RESERVED = 0x00
+    HEADER = 0x01
+    FOOTER = 0x02
+    PAGELIST = 0x03
+
+
+class RNTupleFieldRole(IntEnum):
+    LEAF = 0x00
+    COLLECTION = 0x01
+    RECORD = 0x02
+    VARIANT = 0x03
+    STREAMER = 0x04
+
+
+class RNTupleFieldFlag(IntEnum):
+    REPETITIVE = 0x01
+    PROJECTED = 0x02
+    CHECKSUM = 0x04
+
+
+class RNTupleColumnFlag(IntEnum):
+    DEFERRED = 0x01
+    RANGE = 0x02
+
+
+class RNTupleExtraTypeIdentifier(IntEnum):
+    ROOT = 0x00
+
+
+class RNTupleClusterFlag(IntEnum):
+    SHARDED = 0x01

{uproot-5.5.0 → uproot-5.5.2}/src/uproot/interpretation/identify.py

@@ -577,6 +577,8 @@ def _parse_node(tokens, i, typename, file, quote, header, inner_header):
     if tokens[i].group(0) == ",":
         _parse_error(tokens[i].start() + 1, typename, file)
 
+    elif tokens[i].group(0) == "const":
+        return _parse_node(tokens, i + 1, typename, file, quote, header, inner_header)
     elif tokens[i].group(0) == "Bool_t":
         return i + 1, _parse_maybe_quote('numpy.dtype("?")', quote)
     elif tokens[i].group(0) == "bool":
@@ -919,19 +921,6 @@ def _parse_node(tokens, i, typename, file, quote, header, inner_header):
             _parse_maybe_quote(f"uproot.containers.AsString({header})", quote),
         )
 
-    elif (
-        has2
-        and tokens[i].group(0) == "const"
-        and (
-            tokens[i + 1].group(0) == "string"
-            or _simplify_token(tokens[i + 1]) == "std::string"
-        )
-    ):
-        return (
-            i + 2,
-            _parse_maybe_quote(f"uproot.containers.AsString({header})", quote),
-        )
-
     elif tokens[i].group(0) == "TString":
         return (
             i + 1,
@@ -947,18 +936,6 @@ def _parse_node(tokens, i, typename, file, quote, header, inner_header):
                 quote,
             ),
         )
-    elif (
-        has2
-        and tokens[i].group(0) == "const"
-        and _simplify_token(tokens[i + 1]) == "char*"
-    ):
-        return (
-            i + 2,
-            _parse_maybe_quote(
-                "uproot.containers.AsString(False, length_bytes='4', typename='char*')",
-                quote,
-            ),
-        )
 
     elif tokens[i].group(0) == "bitset" or _simplify_token(tokens[i]) == "std::bitset":
         _parse_expect("<", tokens, i + 1, typename, file)

uproot-5.5.2/src/uproot/interpretation/known_forth/__init__.py

@@ -0,0 +1,46 @@
+# BSD 3-Clause License; see https://github.com/scikit-hep/uproot5/blob/main/LICENSE
+
+"""
+This module provides known forth code and awkward forms for types where it is known a priori.
+
+See :doc:`uproot.interpretation.known_forth.known_forth_of` for the function
+that provides the lookup of known forth codes and :doc:`uproot.interpretation.known_forth.atlas.VectorVectorElementLink` for an
+implementation used in ATLAS (D)AODs.
+"""
+from __future__ import annotations
+
+import uproot
+from uproot.interpretation.known_forth.atlas import VectorVectorElementLink
+
+KNOWN_FORTH_DICT = {
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::CaloCluster_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::IParticle>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::MuonSegment_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::NeutralParticle_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::TauTrack_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::TrackParticle_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::TruthParticle_v1>>>>": VectorVectorElementLink,
+    "std::vector<std::vector<ElementLink<DataVector<xAOD::Vertex_v1>>>>": VectorVectorElementLink,
+}
+
+
+def known_forth_of(model):
+    """
+    Args:
+        model: The :doc:`uproot.model.Model` to look up known forth for
+
+    Returns an object with attributes `forth_code` and `awkward_form` if a known
+    special case exists, else None
+    """
+    try:
+        typename = model.typename
+    except AttributeError:
+        try:
+            typename = model.classname
+        except AttributeError:
+            typename = uproot.model.classname_decode(model.__name__)
+
+    if typename not in KNOWN_FORTH_DICT:
+        return
+
+    return KNOWN_FORTH_DICT[typename](typename)

uproot-5.5.2/src/uproot/interpretation/known_forth/atlas.py

@@ -0,0 +1,99 @@
+# BSD 3-Clause License; see https://github.com/scikit-hep/uproot5/blob/main/LICENSE
+
+"""
+This module defines ATLAS specific known forth code
+"""
+
+from __future__ import annotations
+
+import re
+
+
+class VectorVectorElementLink:
+    """
+    Known forth and awkward form for ``std::vector<std::vector<ElementLink<T>>`` types in ATLAS (D)AODs
+
+    The forth code was adjusted from what was provided in
+    ``branch._complete_forth_code`` after running ``.array()`` once.
+
+    The binary data of one vector<vector<ElementLink<T>> looks as follows:
+
+    * 6 bytes header for the outer vector
+    * 4 bytes big endian uint for the size of the outer vector (node1)
+    * for each outer vector element:
+    * 4 bytes big endian uint for the size of the inner vector (node2)
+    * for each inner vector element:
+    * 20 bytes header for the ElementLink object
+    * 4 bytes big endian uint for the ``m_persKey`` member (node3)
+    * 4 bytes big endian uint for the ``m_persIndex`` member (node4)
+    """
+
+    forth_code = """
+input stream
+input byteoffsets
+input bytestops
+output node1-offsets int64
+output node2-offsets int64
+output node3-data uint32
+output node4-data uint32
+
+0 node1-offsets <- stack
+0 node2-offsets <- stack
+
+0 do
+    byteoffsets I-> stack
+    stream seek
+    6 stream skip
+    stream !I-> stack
+    dup node1-offsets +<- stack
+    0 do
+        stream !I-> stack
+        dup node2-offsets +<- stack
+        0 do
+            20 stream skip
+            stream !I-> node3-data
+            stream !I-> node4-data
+        loop
+    loop
+loop
+"""
+
+    def __init__(self, typename):
+        self.typename = typename
+        self.inner_typename = re.sub(
+            "std::vector<std::vector<(.*)>>", r"\1", self.typename
+        )
+
+    @property
+    def awkward_form(self):
+        return {
+            "class": "ListOffsetArray",
+            "offsets": "i64",
+            "form_key": "node1",
+            "content": {
+                "class": "ListOffsetArray",
+                "offsets": "i64",
+                "form_key": "node2",
+                "content": {
+                    "class": "RecordArray",
+                    "fields": ["m_persKey", "m_persIndex"],
+                    "contents": [
+                        {
+                            "class": "NumpyArray",
+                            "primitive": "uint32",
+                            "inner_shape": [],
+                            "parameters": {},
+                            "form_key": "node3",
+                        },
+                        {
+                            "class": "NumpyArray",
+                            "primitive": "uint32",
+                            "inner_shape": [],
+                            "parameters": {},
+                            "form_key": "node4",
+                        },
+                    ],
+                    "parameters": {"__record__": f"{self.inner_typename}"},
+                },
+            },
+        }