Difference in CTRL+D Behavior Between `code.interact` and Python REPL

[adqm]

Bug report

Bug description:

I'm not sure whether I would call this a bug per se, but I just noticed that code.interact() has different behavior from both the old and the new REPL when working with multiline input. When in multiline mode and hitting CTRL+D on a new, blank line, e.g.:

>>> def foo(x):
... [ctrl+d]

• The old REPL (tested in Python 3.11) seems to run the input up to the EOF, leaving us in the REPL:

  Python 3.11.2 (main, Apr 28 2025, 14:11:48) [GCC 12.2.0] on linux
  Type "help", "copyright", "credits" or "license" for more information.
  >>> def foo(x):
  ... 
    File "<stdin>", line 1
      def foo(x):
                 ^
  IndentationError: expected an indented sensibleblock after function definition on line 1
  

• In new REPL (testing in Python 3.13), it seems to be a NOP (we can still edit from here):

  Python 3.13.1 (main, Feb  2 2025, 23:16:12) [GCC 12.2.0] on linux
  Type "help", "copyright", "credits" or "license" for more information.
  >>> def foo(x):
  ... 
  

• In code.interact(), though, it exits the REPL entirely, without running the code:

  >>> code.interact()
  Python 3.13.1 (main, Feb  2 2025, 23:16:12) [GCC 12.2.0] on linux
  Type "help", "copyright", "credits" or "license" for more information.
  (InteractiveConsole)
  >>> def foo(x):
  ... 
  
  now exiting InteractiveConsole...
  

I could see arguments for any of those three behaviors being the "right" one, but regardless of what the right behavior is, it feels like code.interact and the Python REPL should have the same behavior here.

I'd be happy to have a go at a PR for this but am unsure what the most reasonable change is (or if this is even worth changing).

CPython versions tested on:

3.11, 3.13

Operating systems tested on:

No response

[adqm]

@StanFromIreland , I had indeed missed that other issue during my quick search, but I'm not sure I agree that this is a duplicate since the other issue wasn't about discrepancies with the code module, but rather about the behavior of the REPL being unexpected.

I'm fine if you think it makes sense to keep the conversation on that other thread, though (and/or to tack the necessary changes to code.py on to that other PR instead of opening a separate one).

[StanFromIreland]

I do not think there is a need to change anything in
code.interact, the current behaviour is fine (and what IMO pyrepl should do), if there is a need it should be a separate discussion. I also do not know of any promises made to ensure stdlib CLIs are similar, e.g. pdb is quite different.

[adqm]

Sure, not all CLIs should be the same, but InteractiveConsole's docstring states an explicit goal: "Closely emulate the behavior of the interactive Python interpreter."