feat: support categorical axes on boost histograms

src/uproot/writing/identify.py

@@ -326,6 +326,7 @@ def to_writable(obj):
                 fXmin=axis.edges[0],
                 fXmax=axis.edges[-1],
                 fXbins=_fXbins_maybe_regular(axis, boost_histogram),
+                fLabels=_fLabels_maybe_categorical(axis, boost_histogram),
             )
             for axis, default_name in zip(obj.axes, ["xaxis", "yaxis", "zaxis"])
         ]
@@ -659,6 +660,31 @@ def _fXbins_maybe_regular(axis, boost_histogram):
             return axis.edges
 
 
+def _fLabels_maybe_categorical(axis, boost_histogram):
+    if boost_histogram is None:
+        return None
+
+    if not isinstance(axis, boost_histogram.axis.IntCategory) and not isinstance(
+        axis, boost_histogram.axis.StrCategory
+    ):
+        return None
+
+    labels = [str(label) for label in axis]
+    if isinstance(axis, boost_histogram.axis.IntCategory):
+        # Check labels are valid integers (this may be redundant)
+        for label in labels:
+            try:
+                int(label)
+            except ValueError:
+                raise ValueError(
+                    f"IntCategory labels must be valid integers. Found {label} on axis {axis}"
+                )
+
+    labels = [to_TObjString(label) for label in labels]
+
+    return to_THashList(labels)
+
+
 def _root_stats_1d(entries, edges):
     centers = (edges[:-1] + edges[1:]) / 2.0
 
@@ -735,12 +761,16 @@ def to_TObjString(string):
     This function is for developers to create TObjString objects that can be
     written to ROOT files, to implement conversion routines.
     """
+    tobject = uproot.models.TObject.Model_TObject.empty()
+    tobject._members["@fUniqueID"] = 0
+    tobject._members["@fBits"] = 0
+
     tobjstring = uproot.models.TObjString.Model_TObjString(str(string))
     tobjstring._deeply_writable = True
     tobjstring._cursor = None
     tobjstring._parent = None
     tobjstring._members = {}
-    tobjstring._bases = (uproot.models.TObject.Model_TObject(),)
+    tobjstring._bases = [tobject]
     tobjstring._num_bytes = len(string) + (1 if len(string) < 255 else 5) + 16
     tobjstring._instance_version = 1
     return tobjstring
@@ -777,6 +807,30 @@ def to_TList(data, name=""):
     return tlist
 
 
+def to_THashList(data, name=""):
+    """
+    Args:
+        data (:doc:`uproot.model.Model`): Python iterable to convert into a THashList.
+        name (str): Name of the list (usually empty: ``""``).
+
+    This function is for developers to create THashList objects that can be
+    written to ROOT files, to implement conversion routines.
+    """
+
+    if not all(isinstance(x, uproot.model.Model) for x in data):
+        raise TypeError(
+            "list to convert to THashList must only contain ROOT objects (uproot.Model)"
+        )
+
+    tlist = to_TList(data, name)
+
+    thashlist = uproot.models.THashList.Model_THashList.empty()
+
+    thashlist._bases.append(tlist)
+
+    return thashlist
+
+
 def to_TArray(data):
     """
     Args:

tests/test_0349-write-TObjString.py

@@ -2,7 +2,6 @@
 
 import os
 
-import numpy as np
 import pytest
 
 import uproot
@@ -78,3 +77,21 @@ def test_update(tmp_path):
         assert f6["subdir/wowie"] == "wowie"
         assert f6["subdir/zowie"] == "zowie"
         assert list(f6.file.streamers) == ["TObjString"]
+
+
+def test_serialization(tmp_path):
+    filename = os.path.join(tmp_path, "whatever.root")
+
+    string = "hey"
+    tobjstring = uproot.writing.identify.to_TObjString(string)
+    assert (
+        tobjstring.tojson()["_typename"] == "TObjString"
+    )  # https://github.com/scikit-hep/uproot5/issues/762
+
+    with uproot.recreate(filename) as f1:
+        f1["first"] = tobjstring
+        f1["second"] = str(tobjstring)  # also checks conversion to "str"
+
+    with uproot.open(filename) as f2:
+        assert f2["first"] == f2["second"]
+        assert str(f2["first"]) == string