infinite loop+print in student code can cause enormous json output => `check50.Missing` is not truncated

[Jelleas]

Ran into the following issue, a student's submission contained an infinite loop printing some values each iteration. The check uses .stdout() to check for some expected output. However, the expected output is never produced, leading to a timeout here:

check50/check50/_api.py

Lines 287 to 291 in 483e432

	except TIMEOUT:
	if show_timeout:
	raise Missing(str_output, self.process.before,
	help=_("check50 waited {} seconds for the output of the program").format(timeout))
	raise Missing(str_output, self.process.before)

The entirety of self.process.before is used as collection for Missing here:

check50/check50/_api.py

Lines 407 to 435 in 483e432

	class Missing(Failure):
	"""
	Exception signifying check failure due to an item missing from a collection.
	This is typically a specific substring in a longer string, for instance the contents of stdout.
	:param item: the expected item / substring
	:param collection: the collection / string
	:param help: optional help message to be displayed
	:type help: str
	Example usage::
	actual = check50.run("./fibonacci 5").stdout()
	if "5" not in actual and "3" in actual:
	help = "Be sure to start the sequence at 1"
	raise check50.Missing("5", actual, help=help)
	"""
	def __init__(self, missing_item, collection, help=None):
	if isinstance(collection, list):
	collection = _process_list(collection, _raw)
	super().__init__(rationale=_("Did not find {} in {}").format(_raw(missing_item), _raw(collection)), help=help)
	if missing_item == EOF:
	missing_item = "EOF"
	self.payload.update({"missing_item": str(missing_item), "collection": str(collection)})

check50 3.4.0's _raw will truncate the output, but not truncate the log. That means with -o json the json output is thousands of lines long.

check50 4.0.0's new _raw does not truncate so there is no truncating for Missing and check50 ends up producing thousands of lines both in its output and its log.

Steps to reproduce:

#include <stdio.h>

int main(void)
{
     while (1)
     {
          printf("0\n");
     }
}

check50.c.compile("foo.c")
check50.run("./foo").stdout("hello world\n");

check50 --dev foo/bar/baz -o json

[ivanharvard]

I think Mismatch already had a solution for this, by use of its _safe_truncate() function:

check50/check50/_api.py

Lines 460 to 473 in 483e432

	def __init__(self, expected, actual, help=None):
	def _safe_truncate(x, y):
	return _truncate(x, y) if x not in (EOF, TIMEOUT) else x
	expected, actual = _safe_truncate(expected, actual), _safe_truncate(actual, expected)
	rationale = _("expected: {}\n actual: {}").format(
	_raw(expected),
	_raw(actual)
	)
	super().__init__(rationale=rationale, help=help)
	self.payload.update({"expected": _raw(expected), "actual": _raw(actual)})

We could just implement that function again for Missing. Then just truncate the collection (should we truncate the missing item as well?). Do you believe that truncating the JSON/HTML should be done as well, or just the ANSI (which would keep it close to v3.4.0)?