python / cpython

The Python programming language
https://www.python.org
Other
62.15k stars 29.86k forks source link

Exception raised in traceback.StackSummary._should_show_carets exits interpreter #122145

Open devdanzin opened 1 month ago

devdanzin commented 1 month ago

Bug report

Bug description:

Running the code sample from #122071 in 3.13.0b4 or main exits the interpreter due to traceback.StackSummary._should_show_carets raising an exception:

>>> exec(compile("tuple()[0]", "s", "exec"))
Traceback (most recent call last):
Exception ignored in the internal traceback machinery:
Traceback (most recent call last):
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 139, in _print_exception_bltin
    return print_exception(exc, limit=BUILTIN_EXCEPTION_LIMIT, file=file, colorize=colorize)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 130, in print_exception
    te.print(file=file, chain=chain, colorize=colorize)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 1448, in print
    for line in self.format(chain=chain, colorize=colorize):
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 1384, in format
    yield from _ctx.emit(exc.stack.format(colorize=colorize))
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 747, in format
    formatted_frame = self.format_frame_summary(frame_summary, colorize=colorize)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 583, in format_frame_summary
    show_carets = self._should_show_carets(start_offset, end_offset, all_lines, anchors)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 701, in _should_show_carets
    statement = tree.body[0]
IndexError: list index out of range
Traceback (most recent call last):
  File "~\PycharmProjects\cpython\Lib\code.py", line 91, in runcode
    exec(code, self.locals)
  File "<python-input-2>", line 1, in <module>
  File "s", line 1, in <module>
IndexError: tuple index out of range

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "~\PycharmProjects\cpython\Lib\runpy.py", line 198, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "~\PycharmProjects\cpython\Lib\runpy.py", line 88, in _run_code
    exec(code, run_globals)
  File "~\PycharmProjects\cpython\Lib\_pyrepl\__main__.py", line 6, in <module>
    __pyrepl_interactive_console()
  File "~\PycharmProjects\cpython\Lib\_pyrepl\main.py", line 59, in interactive_console
    run_multiline_interactive_console(console)
  File "~\PycharmProjects\cpython\Lib\_pyrepl\simple_interact.py", line 156, in run_multiline_interactive_console
    more = console.push(_strip_final_indent(statement), filename=input_name, _symbol="single")  # type: ignore[call-arg]
  File "~\PycharmProjects\cpython\Lib\code.py", line 303, in push
    more = self.runsource(source, filename, symbol=_symbol)
  File "~\PycharmProjects\cpython\Lib\_pyrepl\console.py", line 200, in runsource
    self.runcode(code)
  File "~\PycharmProjects\cpython\Lib\code.py", line 95, in runcode
    self.showtraceback()
  File "~\PycharmProjects\cpython\Lib\_pyrepl\console.py", line 168, in showtraceback
    super().showtraceback(colorize=self.can_colorize)
  File "~\PycharmProjects\cpython\Lib\code.py", line 147, in showtraceback
    lines = traceback.format_exception(ei[0], ei[1], last_tb.tb_next, colorize=colorize)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 155, in format_exception
    return list(te.format(chain=chain, colorize=colorize))
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 1384, in format
    yield from _ctx.emit(exc.stack.format(colorize=colorize))
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 747, in format
    formatted_frame = self.format_frame_summary(frame_summary, colorize=colorize)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 583, in format_frame_summary
    show_carets = self._should_show_carets(start_offset, end_offset, all_lines, anchors)
  File "~\PycharmProjects\cpython\Lib\traceback.py", line 701, in _should_show_carets
    statement = tree.body[0]
IndexError: list index out of range
[Thread 31424.0x8ec0 exited with code 1]
[Thread 31424.0x15dc exited with code 1]
[Thread 31424.0x8b8c exited with code 1]
[Inferior 1 (process 31424) exited with code 01]

We can either protect the _should_show_carets call with suppress(Exception) here: https://github.com/python/cpython/blob/2762c6cc5e4c1c0d630568db5fbba7a3a71a507c/Lib/traceback.py#L581-L583

Or guard against tree.body being empty here: https://github.com/python/cpython/blob/2762c6cc5e4c1c0d630568db5fbba7a3a71a507c/Lib/traceback.py#L700-L701

(or both?)

Should I submit a PR with one of the fixes above? Any preferences?

CPython versions tested on:

3.13, CPython main branch

Operating systems tested on:

Windows, Linux

Linked PRs

gaogaotiantian commented 1 month ago

In my opinion, we should figure out that issue first. This is probably something that should not happen - it's kind of equivalent to a comment only file generates an exception, that's just impossible.

devdanzin commented 1 month ago

In my opinion, we should figure out that issue first. This is probably something that should not happen - it's kind of equivalent to a comment only file generates an exception, that's just impossible.

Indeed, ISTM that due to the linecache bug we try to parse "lines" consisting of only a comment then access the tree's body first node. So fixing that issue will fix this specific failure.

Maybe it would still be interesting to make the code more robust against exceptions in _should_show_carets?

sobolevn commented 1 month ago

Looks like there are two bugs:

  1. With incorrect module detection, exception is not in Lib/_pyrepl/__main__.py (this file is used for sources)
  2. But, tree.body can be empty in other cases as well, so why not guard it?
diff --git Lib/traceback.py Lib/traceback.py
index 6ee1a50ca68..eba6f03a3c6 100644
--- Lib/traceback.py
+++ Lib/traceback.py
@@ -698,6 +698,8 @@ def _should_show_carets(self, start_offset, end_offset, all_lines, anchors):
         with suppress(SyntaxError, ImportError):
             import ast
             tree = ast.parse('\n'.join(all_lines))
+            if not tree.body:
+                return False
             statement = tree.body[0]
             value = None
             def _spawns_full_line(value):

After:

» ./python.exe
Python 3.14.0a0 (heads/main-dirty:3de092b82f5, Jul 20 2024, 09:23:52) [Clang 15.0.0 (clang-1500.3.9.4)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> exec(compile("tuple()[0]", "s", "exec"))
Traceback (most recent call last):
  File "<python-input-0>", line 1, in <module>
    exec(compile("tuple()[0]", "s", "exec"))
    ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "s", line 1, in <module>
    # Important: don't add things to this module, as they will end up in the REPL's
IndexError: tuple index out of range
picnixz commented 1 month ago

I've fixed the second issue and added a test for that one. I can move to fixing the first issue as well.

devdanzin commented 1 month ago
2. But, `tree.body` can be empty in other cases as well, so why not guard it?

Exactly. Another way to get an empty tree.body is to run:

exec(compile("tuple()[0]", "x.py", "exec"))

Where you have a x.py file that consists of:

# Just a comment.
devdanzin commented 1 month ago

Yet another way to get an empty tree.body:

exec(compile("print(2**100000)", "s", "exec"))

EDIT: never mind, I had reverted the fix.

picnixz commented 1 month ago

Errr... ok. I'll amend the NEWS entry in the PR since it only talks about comments.

EDIT: Actually, I think you can only have an empty AST body if you are either an empty string or if you are a comment.

devdanzin commented 1 month ago

In StackSummary.format_frame_summary, StackSummary._should_show_carets is called without error suppression (like is done for _extract_caret_anchors_from_line_segment). This causes any errors in that method to exit the interpreter. I suggest we also ignore errors from _should_show_carets here:

https://github.com/python/cpython/blob/d27a53fc02a87e76066fc4e15ff1fff3922a482d/Lib/traceback.py#L581-L583

This time, an error in parsing a too deeply nested structure exits the interpreter:

>>> with open("lambda.txt", "w") as n:
...     n.write("lambda: lambda: " * 1000 + "None")
...
16004
>>> exec(compile("str(2**100000)", "lambda.txt", "exec"))
Traceback (most recent call last):
Exception ignored in the internal traceback machinery:
Traceback (most recent call last):
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 139, in _print_exception_bltin
    return print_exception(exc, limit=BUILTIN_EXCEPTION_LIMIT, file=file, colorize=colorize)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 130, in print_exception
    te.print(file=file, chain=chain, colorize=colorize)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 1449, in print
    for line in self.format(chain=chain, colorize=colorize):
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 1385, in format
    yield from _ctx.emit(exc.stack.format(colorize=colorize))
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 748, in format
    formatted_frame = self.format_frame_summary(frame_summary, colorize=colorize)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 583, in format_frame_summary
    show_carets = self._should_show_carets(start_offset, end_offset, all_lines, anchors)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 701, in _should_show_carets
    tree = ast.parse('\n'.join(all_lines))
  File "/home/danzin/projects/cpython/Lib/ast.py", line 53, in parse
    return compile(source, filename, mode, flags,
RecursionError: maximum recursion depth exceeded during compilation
Traceback (most recent call last):
  File "/home/danzin/projects/cpython/Lib/code.py", line 91, in runcode
    exec(code, self.locals)
  File "<python-input-1>", line 1, in <module>
  File "lambda.txt", line 1, in <module>
    lambda: lambda: lambda: lambda: lambda: [...] lambda: None
ValueError: Exceeds the limit (4300 digits) for integer string conversion; use sys.set_int_max_str_digits() to increase the limit

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/danzin/projects/cpython/Lib/runpy.py", line 198, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/home/danzin/projects/cpython/Lib/runpy.py", line 88, in _run_code
    exec(code, run_globals)
  File "/home/danzin/projects/cpython/Lib/_pyrepl/__main__.py", line 6, in <module>
    __pyrepl_interactive_console()
  File "/home/danzin/projects/cpython/Lib/_pyrepl/main.py", line 59, in interactive_console
    run_multiline_interactive_console(console)
  File "/home/danzin/projects/cpython/Lib/_pyrepl/simple_interact.py", line 156, in run_multiline_interactive_console
    more = console.push(_strip_final_indent(statement), filename=input_name, _symbol="single")  # type: ignore[call-arg]
  File "/home/danzin/projects/cpython/Lib/code.py", line 303, in push
    more = self.runsource(source, filename, symbol=_symbol)
  File "/home/danzin/projects/cpython/Lib/_pyrepl/console.py", line 200, in runsource
    self.runcode(code)
  File "/home/danzin/projects/cpython/Lib/code.py", line 95, in runcode
    self.showtraceback()
  File "/home/danzin/projects/cpython/Lib/_pyrepl/console.py", line 168, in showtraceback
    super().showtraceback(colorize=self.can_colorize)
  File "/home/danzin/projects/cpython/Lib/code.py", line 147, in showtraceback
    lines = traceback.format_exception(ei[0], ei[1], last_tb.tb_next, colorize=colorize)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 155, in format_exception
    return list(te.format(chain=chain, colorize=colorize))
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 1385, in format
    yield from _ctx.emit(exc.stack.format(colorize=colorize))
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 748, in format
    formatted_frame = self.format_frame_summary(frame_summary, colorize=colorize)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 583, in format_frame_summary
    show_carets = self._should_show_carets(start_offset, end_offset, all_lines, anchors)
  File "/home/danzin/projects/cpython/Lib/traceback.py", line 701, in _should_show_carets
    tree = ast.parse('\n'.join(all_lines))
  File "/home/danzin/projects/cpython/Lib/ast.py", line 53, in parse
    return compile(source, filename, mode, flags,
RecursionError: maximum recursion depth exceeded during compilation

If an even deeper nested structure is used, MemoryError is raised, but somehow doesn't exit the interpreter.

Doesn't seem to affect 3.13. Should a new issue be created for this one?