bpo-27580: Add support of null characters in csv

[serhiy-storchaka]

https://bugs.python.org/issue27580

[serhiy-storchaka]

@smontanaro, please take a look. For unknown reasons I cannot add you to reviewers.

[smontanaro]

For unknown reasons I cannot add you to reviewers.

Maybe that's by design. :-)

LGTM. I assume Doc/library/csv.rst will be updated at some point to make it clear that ASCII NUL is now a valid.

[ammaraskar]

You don't have to necessarily do it here, but just a note to myself that we should update the csv fuzzer:

cpython/Modules/_xxtestfuzz/fuzzer.c

Lines 356 to 360 in 0219017

	/* Ignore non null-terminated strings since _csv can't handle
	embedded nulls */
	if (memchr(data, '\0', size) == NULL) {
	return 0;
	}

to generate inputs with nulls, such as in the json fuzzer here:

cpython/Modules/_xxtestfuzz/fuzzer.c

Lines 166 to 169 in 0219017

	PyObject* input_bytes = PyBytes_FromStringAndSize(data, size);
	if (input_bytes == NULL) {
	return 0;
	}

in this or a follow up PR.

[corona10]

lgtm

[serhiy-storchaka]

I assume Doc/library/csv.rst will be updated at some point to make it clear that ASCII NUL is now a valid.

I have not found a good place for this. It was not documented that ASCII NUL is special. Perhaps a NEWS entry will be enough.

[serhiy-storchaka]

@ammaraskar, I have no experience with fuzzers. Feel free to add a new test if it is useful.

[merwok]

#97503 is requesting a backport to 3.10