`inspect.getframeinfo()` doesn't handle legitimate frames with `lineno=None`

[yilei]

Bug report

Bug description:

Here is a minimal example:

import inspect
import sys

code = """
try:
    1/0
except* Exception:
    raise
"""

try:
    exec(code)
except:
    tb = sys.exc_info()[2]
    while tb:
        print(f"{tb.tb_frame=}")
        info = inspect.getframeinfo(tb.tb_frame)  # This may raise.
        tb = tb.tb_next

print("DONE")

Tested on Python 3.13.7, the output is:

tb.tb_frame=<frame at 0x7fc9bd40ac80, file '/tmp/demo.py', line 16, code <module>>
tb.tb_frame=<frame at 0x7fc9bd4ae9b0, file '<string>', line -1, code <module>>
  + Exception Group Traceback (most recent call last):
  |   File "/tmp/demo.py", line 12, in <module>
  |     exec(code)
  |     ~~~~^^^^^^
  |   File "<string>", line 4, in <module>
  | ExceptionGroup:  (1 sub-exception)
  +-+---------------- 1 ----------------
    | Traceback (most recent call last):
    |   File "<string>", line 3, in <module>
    | ZeroDivisionError: division by zero
    +------------------------------------

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/tmp/demo.py", line 17, in <module>
    info = inspect.getframeinfo(tb.tb_frame)  # This may raise.
  File "/root/.pyenv/versions/3.13.7/lib/python3.13/inspect.py", line 1689, in getframeinfo
    start = lineno - 1 - context//2
            ~~~~~~~^~~
TypeError: unsupported operand type(s) for -: 'NoneType' and 'int'

This is reproducible in Python 3.12+. It works on Python 3.11 but it might because it doesn't have the second problematic getframeinfo call while traversing the stack:

tb.tb_frame=<frame at 0x7f41ea7f97a0, file '/tmp/demo.py', line 16, code <module>>
DONE

This might be a duplicate of #89726, but we encountered an example using exception groups and it works on Python 3.11 but not 3.12+ so I'm filing a separate bug. Feel free to dedup.

CPython versions tested on:

3.13, 3.12, 3.14

Operating systems tested on:

Linux

Linked PRs

• gh-139531: Fix TypeError in inspect.getframeinfo when lineno is None #139535

[PrinceNaroliya]

Hi @yilei,

I came across this issue (#139531) regarding inspect.getframeinfo() raising a TypeError when lineno is None in exception groups. I am very interested in contributing a fix for this.

I have already been working on Python open-source contributions and am familiar with writing test cases for Python standard library modules. My plan is to:

1. Add a guard in inspect.getframeinfo to handle cases where lineno is None gracefully.
2. Write a test case in Lib/test/test_inspect.py to reproduce the issue and verify that the fix prevents the TypeError.

Before I start, I wanted to confirm with you if this approach aligns with the project’s coding standards. Any guidance or suggestions would be greatly appreciated.

Thank you for your time! Looking forward to contributing.

Best regards,
Prince Naroliya

[picnixz]

I'll consider this one as a duplicate of #89726 although the latter is a wider issue. The latter wonders whether we should support manually crafted code objects, in which case, it's fine to have an error. However, here, we don't construct anything manually, and thus, it's a legitimate issue (either at the interpreter's level or inspect level)

[picnixz]

Actually, I'm sorry but the PR for #89726 also shows an example of where a legitimate frame can have lineno=None, so it's effectively a duplicate. I'll cross-post your reproducer which is much simpler though.