Tachyon sample dump/replay crashes

[hauntsaninja]

Bug description:

I'm running python -m profiling.sampling replay prof.bin --heatmap after having captured some profile and it crashes.

When I dropped a breakpoint in I saw the dump somehow has ended up with one instance where stat.filename == '\x00'

Error: Failed to export heatmap: embedded null byte
Traceback (most recent call last):
  File "<frozen runpy>", line 196, in _run_module_as_main
  File "<frozen runpy>", line 87, in _run_code
  File "/.../lib/python3.15/profiling/sampling/__main__.py", line 65, in <module>
    main()
    ~~~~^^
  File "/.../lib/python3.15/profiling/sampling/cli.py", line 788, in main
    _main()
    ~~~~~^^
  File "/.../lib/python3.15/profiling/sampling/cli.py", line 922, in _main
    handler(args)
    ~~~~~~~^^^^^^
  File "/.../lib/python3.15/profiling/sampling/cli.py", line 1206, in _handle_replay
    collector.export(filename)
    ~~~~~~~~~~~~~~~~^^^^^^^^^^
  File "/.../lib/python3.15/profiling/sampling/heatmap_collector.py", line 725, in export
    self._generate_file_reports(output_dir, file_stats)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../lib/python3.15/profiling/sampling/heatmap_collector.py", line 766, in _generate_file_reports
    self._generate_file_html(
    ~~~~~~~~~~~~~~~~~~~~~~~~^
        file_path,
        ^^^^^^^^^^
    ...<3 lines>...
        stat
        ^^^^
    )
    ^
  File "/.../lib/python3.15/profiling/sampling/heatmap_collector.py", line 905, in _generate_file_html
    source_lines = Path(filename).read_text(encoding='utf-8', errors='replace').splitlines()
                   ~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../lib/python3.15/pathlib/__init__.py", line 985, in read_text
    with self.open(mode='r', encoding=encoding, errors=errors, newline=newline) as f:
         ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../lib/python3.15/pathlib/__init__.py", line 969, in open
    return io.open(self, mode, buffering, encoding, errors, newline)
           ~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: embedded null byte

Don't yet have a minimal repro, but maybe the issue is apparent to you

CPython versions tested on:

3.15

Operating systems tested on:

macOS

[StanFromIreland]

Maybe after reading the filename in parse_code_object, we should check the content and not just the length? WDYT @pablogsal

[pablogsal]

Don't yet have a minimal repro, but maybe the issue is apparent to you

Unfortunately not :(

If you could have a repro or you can send over the file that would be amazing!

[pablogsal]

Oh wait this is opening a python file:

  File "/.../lib/python3.15/profiling/sampling/heatmap_collector.py", line 905, in _generate_file_html
    source_lines = Path(filename).read_text(encoding='utf-8', errors='replace').splitlines()
                   ~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../lib/python3.15/pathlib/__init__.py", line 985, in read_text
    with self.open(mode='r', encoding=encoding, errors=errors, newline=newline) as f:
         ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../lib/python3.15/pathlib/__init__.py", line 969, in open
    return io.open(self, mode, buffering, encoding, errors, newline)
           ~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: embedded null byte

It's here:

cpython/Lib/profiling/sampling/heatmap_collector.py

Line 905 in 96e4cd6

source_lines = Path(filename).read_text(encoding='utf-8', errors='replace').splitlines()

Is not the result file is just a python file!