User-provided attributes dictionary is mutated by the public API

[stibrew]

Zarr version

v3.2.1

Numcodecs version

0.16.5

Python Version

3.12

Operating System

Linux

Installation

using pip into virtual environment

Description

When creating a Zarr group or array with a user-provided attributes dictionary, the original dictionary gets mutated when the user subsequently modifies attributes via the public API (attrs.setitem, attrs.put, or del attrs[key]).

Steps to reproduce

# /// script
# requires-python = ">=3.12"
# dependencies = [
#   "zarr@git+https://github.com/zarr-developers/zarr-python.git@main",
# ]
# ///
#
# This script automatically imports the development branch of zarr to check for issues

import zarr
init_dict = {"foo": "bar"}
user_attrs = dict(init_dict)
print("User's dict before passing to Zarr:", user_attrs)
g = zarr.group(attributes=user_attrs)
g.attrs["new_key"] = "new_value"
print("User's dict after passing to Zarr:", user_attrs)
print("Expected:   ", init_dict)
assert user_attrs == init_dict, "FAIL: Original dict mutated by"
zarr.print_debug_info()

Additional output

There is a key called cache_attrs under zarr.group() which is not yet implemented. Is there a reason to pursue that over just copying the attributes which user provides?

[d-v-b]

that's an unpleasant bug. We should be copying the user input.