| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | |
| 4 | from __future__ import annotations |
| 5 | |
| 6 | import hashlib |
| 7 | import pickle |
| 8 | from _hashlib import HASH, UnsupportedDigestmodError |
| 9 | from collections.abc import Callable |
| 10 | from typing import Any |
| 11 | |
| 12 | import cbor2 |
| 13 | |
| 14 | try: |
| 15 | # It is important that this remains an optional dependency. |
| 16 | # It would not be allowed in environments with strict security controls, |
| 17 | # so it's best not to have it installed when not in use. |
| 18 | import xxhash as _xxhash |
| 19 | |
| 20 | if not hasattr(_xxhash, "xxh3_128_digest"): |
| 21 | _xxhash = None |
| 22 | except ImportError: # pragma: no cover |
| 23 | _xxhash = None |
| 24 | |
| 25 | |
| 26 | def sha256(input: Any) -> bytes: |
| 27 | """Hash any picklable Python object using SHA-256. |
| 28 | |
| 29 | The input is serialized using pickle before hashing, which allows |
| 30 | arbitrary Python objects to be used. Note that this function does |
| 31 | not use a hash seed—if you need one, prepend it explicitly to the input. |
| 32 | |
| 33 | Args: |
| 34 | input: Any picklable Python object. |
| 35 | |
| 36 | Returns: |
| 37 | Bytes representing the SHA-256 hash of the serialized input. |
| 38 | |
| 39 | """ |
| 40 | input_bytes = pickle.dumps(input, protocol=pickle.HIGHEST_PROTOCOL) |
| 41 | return hashlib.sha256(input_bytes).digest() |
| 42 | |
| 43 | |
| 44 | def sha256_cbor(input: Any) -> bytes: |
| 45 | """Hash objects using CBOR serialization and SHA-256. |
| 46 | |
| 47 | This option is useful for non-Python-dependent serialization and hashing. |
| 48 | |
| 49 | Args: |
| 50 | input: Object to be serialized and hashed. Supported types include |
| 51 | basic Python types and complex structures like lists, tuples, and |
| 52 | dictionaries. |
| 53 | Custom classes must implement CBOR serialization methods. |
| 54 | |
| 55 | Returns: |
| 56 | Bytes representing the SHA-256 hash of the CBOR serialized input. |
| 57 | |
| 58 | """ |
| 59 | input_bytes = cbor2.dumps(input, canonical=True) |
| 60 | return hashlib.sha256(input_bytes).digest() |
| 61 | |
| 62 | |
| 63 | def _xxhash_digest(input_bytes: bytes) -> bytes: |
| 64 | if _xxhash is None: |
| 65 | raise ModuleNotFoundError( |
| 66 | "xxhash is required for the 'xxhash' prefix caching hash algorithms. " |
| 67 | "Install it via `pip install xxhash`." |
| 68 | ) |
| 69 | return _xxhash.xxh3_128_digest(input_bytes) |
| 70 | |
| 71 | |
| 72 | def xxhash(input: Any) -> bytes: |
| 73 | """Hash picklable objects using xxHash.""" |
| 74 | input_bytes = pickle.dumps(input, protocol=pickle.HIGHEST_PROTOCOL) |
| 75 | return _xxhash_digest(input_bytes) |
| 76 | |
| 77 | |
| 78 | def xxhash_cbor(input: Any) -> bytes: |
| 79 | """Hash objects serialized with CBOR using xxHash.""" |
| 80 | input_bytes = cbor2.dumps(input, canonical=True) |
| 81 | return _xxhash_digest(input_bytes) |
| 82 | |
| 83 | |
| 84 | def get_hash_fn_by_name(hash_fn_name: str) -> Callable[[Any], bytes]: |
| 85 | """Get a hash function by name, or raise an error if the function is not found. |
| 86 | |
| 87 | Args: |
| 88 | hash_fn_name: Name of the hash function. |
| 89 | |
| 90 | Returns: |
| 91 | A hash function. |
| 92 | |
| 93 | """ |
| 94 | if hash_fn_name == "sha256": |
| 95 | return sha256 |
| 96 | if hash_fn_name == "sha256_cbor": |
| 97 | return sha256_cbor |
| 98 | if hash_fn_name == "xxhash": |
| 99 | return xxhash |
| 100 | if hash_fn_name == "xxhash_cbor": |
| 101 | return xxhash_cbor |
| 102 | |
| 103 | raise ValueError(f"Unsupported hash function: {hash_fn_name}") |
| 104 | |
| 105 | |
| 106 | def safe_hash(data: bytes, usedforsecurity: bool = True) -> HASH: |
| 107 | """Hash for configs, defaulting to md5 but falling back to sha256 |
| 108 | in FIPS constrained environments. |
| 109 | |
| 110 | Args: |
| 111 | data: bytes |
| 112 | usedforsecurity: Whether the hash is used for security purposes |
| 113 | |
| 114 | Returns: |
| 115 | Hash object |
| 116 | |
| 117 | """ |
| 118 | try: |
| 119 | return hashlib.md5(data, usedforsecurity=usedforsecurity) |
| 120 | except (UnsupportedDigestmodError, ValueError): |
| 121 | return hashlib.sha256(data) |