plotGraph.py fails with ValueError in parser.py due to inhomogeneous array in pd.array(full_stats)

[Androidbeingx]

Description

Running plotGraph.py on any collected metrics directory fails with a ValueError in treecript/parser.py at line 214. The error occurs because columns like processor_ids, core_ids, and cpu_ids in the metrics CSV files contain space-separated strings (e.g. "1 3 5"), making full_stats an inhomogeneous list. pd.array() without an explicit dtype tries to infer a uniform numeric type and fails.

This bug is reproducible on any platform with any pandas version (tested on 1.5.3 and 2.3.3).

Environment

• OS: Ubuntu 22.04 (native Linux)
• Python: 3.10
• Installation: pip install -c constraints-3.10.txt git+https://github.com/inab/treecript.git@exec
• pandas: as pinned by constraints-3.10.txt

Steps to reproduce

# Install treecript
pip install -c constraints-3.10.txt git+https://github.com/inab/treecript.git@exec

# Run plotGraph.py on any collected metrics directory
plotGraph.py path/to/metrics_dir/ path/to/output_dir/

Traceback

Traceback (most recent call last):
  File "/home/andreabsc/miniforge3/envs/Treecript/bin/plotGraph.py", line 24, in 
    main()
  File "/home/andreabsc/miniforge3/envs/Treecript/lib/python3.10/site-packages/treecript/plot_graph.py", line 279, in main
    plot_graphs(
  File "/home/andreabsc/miniforge3/envs/Treecript/lib/python3.10/site-packages/treecript/plot_graph.py", line 239, in plot_graphs
    pids, num_cpu_cores, sampling_period_seconds = metrics_parser(
  File "/home/andreabsc/miniforge3/envs/Treecript/lib/python3.10/site-packages/treecript/parser.py", line 214, in metrics_parser
    pids["full_stats"] = pd.array(full_stats)
  File "/home/andreabsc/miniforge3/envs/Treecript/lib/python3.10/site-packages/pandas/core/construction.py", line 384, in array
    return NumpyExtensionArray._from_sequence(data, dtype=dtype, copy=copy)
  File "/home/andreabsc/miniforge3/envs/Treecript/lib/python3.10/site-packages/pandas/core/arrays/numpy_.py", line 130, in _from_sequence
    result = np.asarray(scalars, dtype=dtype)  # type: ignore[arg-type]
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (11,) + inhomogeneous part.

Root cause

In treecript/parser.py line 214:

# current — fails
pids["full_stats"] = pd.array(full_stats)

full_stats contains rows where some fields (processor_ids, core_ids, cpu_ids) are space-separated strings while others are numeric, resulting in an inhomogeneous structure that pd.array() cannot coerce to a single dtype.

Fix

# fixed — explicitly allow mixed types
pids["full_stats"] = pd.array(full_stats, dtype=object)

This tells pandas not to infer a uniform type and simply store the elements as-is, which is the correct behaviour for a mixed-type column.

[jmfernandez]

Issue was fixed both in 2854744 and bad1bdc commits