Attacker-controlled CSV samples can trigger super-linear regular-expression work during dialect sniffing and consume significant CPU when applications pass unbounded input to csv.Sniffer.sniff().
Weakness
The product performs CPU computations using algorithms that are not as efficient as they could be for the needs of the developer, i.e., the computations can be optimized further.
References
- https://github.com/python/cpython/commit/063d4555c94ef412c731527dbf30193327f2ee82
- https://github.com/python/cpython/commit/476fb09cdb0d73e645849d98c610e7e5697ce7c9
- https://github.com/python/cpython/commit/89f29c760c02774b099ddd6863268eb13fa3946a
- https://github.com/python/cpython/commit/b09a67a20c464f8288c9f9a6b9c7082a74560024
- https://github.com/python/cpython/commit/b30c7fa9edd921a118f286e9f90f560777fa693b
- https://github.com/python/cpython/commit/fd78b565d7c326f96ae903ab945b47f35d829cf4
- https://github.com/python/cpython/issues/98820
- https://github.com/python/cpython/pull/153694
- https://mail.python.org/archives/list/security-announce@python.org/thread/KQ7NBMCPAZJHRROQXJQE4GMXGLD5KHBS/