Hawkynt

TPy (DarkCrypt)

Tweaked-IV-setup variant of Py by Biham and Seberry (2007), fixing the equivalent-IV weakness of the original Py while keeping its key setup and round function unchanged. As implemented in the DarkCrypt Total Commander plugin, which hardcodes a 512-bit key and 512-bit IV.

Properties

Property Value
Category Stream Ciphers
Sub-category Rolling-Array Stream Cipher
Security status ❌ Broken
Complexity Advanced
Inventor Eli Biham, Jennifer Seberry
Year 2007
Origin 🇮🇱 Israel
Source algorithms/stream/darkcrypt-tpy.js

Parameters

Parameter Supported values
Key sizes 64 bytes (512 bits)
Nonce sizes 64 bytes (512 bits)

Security

Status: ❌ Broken

Known vulnerabilities

Issue Description Mitigation
Distinguishing attacks Crowley and Paul/Preneel/Sekar published distinguishing attacks on TPy independent of the key schedule fix. Use a vetted modern cipher.

Documentation

References

Test vectors

2 vectors ship with this algorithm and run in the test suite. Byte values are hexadecimal.

Vector 1 — DarkCrypt TPy keystream

Field Value
iv 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000
key 000102030405060708090a0b0c0d0e0f 101112131415161718191a1b1c1d1e1f 202122232425262728292a2b2c2d2e2f 303132333435363738393a3b3c3d3e3f
input 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000 00000000000000000000000000000000
expected 9757dfc4e1f7caf9c04013c7269f05a5 5886a19930dd0cd81ca74a2f6eea4ec2 dffb3ed25d10deb24387a064e90b6aeb 5e95f9017273d9653fa9cc28a72f08e7 83f2647815bccad5caa41423f79cfdec 6c1f8f7461dc21b01c80d8152a6d09b6 5fd1f0ed2f942987dab01aacee09124f 5593ecebdce0d166eb8d33d19ac9c09c

Vector 2 — DarkCrypt TPy — non-zero key and IV, 48-byte message (verified against the DarkCrypt implementation)

Field Value
key 0b30557a9fc4e90e33587da2c7ec1136 5b80a5caef14395e83a8cdf2173c6186 abd0f51a3f6489aed3f81d42678cb1d6 fb20456a8fb4d9fe23486d92b7dc0126
iv 073c71a6db10457aafe4194e83b8ed22 578cc1f62b6095caff34699ed3083d72 a7dc11467bb0e51a4f84b9ee23588dc2 f72c6196cb00356a9fd4093e73a8dd12
input 03203d5a7794b1ceeb0825425f7c99b6 d3f00d2a4764819ebbd8f5122f4c6986 a3c0ddfa1734516e8ba8c5e2ff1c3956
expected bf67c4b29fe7b70e0cc9349a01f5d10c 9f74bd60a03c6f62b117362453259152 c2aab066604515b33b6e1fc28c3698b3

← All algorithms