add xxh64

2026-08-14 03:55:05 +03:00
parent 01208f4143
commit 558450b573
9 changed files with 320 additions and 22 deletions
@@ -0,0 +1,224 @@
+/*
+ * Standalone XXH64 command-line wrapper for avalanche testing.
+ *
+ * XXH64 algorithm derived from xxHash by Yann Collet:
+ * https://github.com/Cyan4973/xxHash
+ *
+ * Copyright (C) 2012-2023 Yann Collet
+ *
+ * BSD 2-Clause License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice,
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES ARE DISCLAIMED. IN NO EVENT SHALL THE
+ * COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DAMAGES ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
+ * DAMAGE.
+ */
+
+#include <inttypes.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define XXH64_SEED UINT64_C(0)
+#define XXH_PRIME64_1 UINT64_C(11400714785074694791)
+#define XXH_PRIME64_2 UINT64_C(14029467366897019727)
+#define XXH_PRIME64_3 UINT64_C(1609587929392839161)
+#define XXH_PRIME64_4 UINT64_C(9650029242287828579)
+#define XXH_PRIME64_5 UINT64_C(2870177450012600261)
+
+static uint64_t rotate_left64(uint64_t value, unsigned int count)
+{
+    return (value << count) | (value >> (64U - count));
+}
+
+static uint32_t read_little_endian32(const unsigned char *data)
+{
+    return (uint32_t)data[0] | ((uint32_t)data[1] << 8U) |
+           ((uint32_t)data[2] << 16U) | ((uint32_t)data[3] << 24U);
+}
+
+static uint64_t read_little_endian64(const unsigned char *data)
+{
+    return (uint64_t)read_little_endian32(data) |
+           ((uint64_t)read_little_endian32(data + 4) << 32U);
+}
+
+static uint64_t xxh64_round(uint64_t accumulator, uint64_t input)
+{
+    accumulator += input * XXH_PRIME64_2;
+    accumulator = rotate_left64(accumulator, 31U);
+    accumulator *= XXH_PRIME64_1;
+    return accumulator;
+}
+
+static uint64_t xxh64_merge_round(uint64_t accumulator, uint64_t value)
+{
+    value = xxh64_round(UINT64_C(0), value);
+    accumulator ^= value;
+    accumulator = accumulator * XXH_PRIME64_1 + XXH_PRIME64_4;
+    return accumulator;
+}
+
+static uint64_t xxh64(const unsigned char *data, size_t length, uint64_t seed)
+{
+    const unsigned char *position = data;
+    const unsigned char *const end = data + length;
+    uint64_t hash;
+
+    if (length >= 32U) {
+        const unsigned char *const block_end = end - 32U;
+        uint64_t accumulator1 = seed + XXH_PRIME64_1 + XXH_PRIME64_2;
+        uint64_t accumulator2 = seed + XXH_PRIME64_2;
+        uint64_t accumulator3 = seed;
+        uint64_t accumulator4 = seed - XXH_PRIME64_1;
+
+        do {
+            accumulator1 = xxh64_round(accumulator1, read_little_endian64(position));
+            position += 8;
+            accumulator2 = xxh64_round(accumulator2, read_little_endian64(position));
+            position += 8;
+            accumulator3 = xxh64_round(accumulator3, read_little_endian64(position));
+            position += 8;
+            accumulator4 = xxh64_round(accumulator4, read_little_endian64(position));
+            position += 8;
+        } while (position <= block_end);
+
+        hash = rotate_left64(accumulator1, 1U) +
+               rotate_left64(accumulator2, 7U) +
+               rotate_left64(accumulator3, 12U) +
+               rotate_left64(accumulator4, 18U);
+        hash = xxh64_merge_round(hash, accumulator1);
+        hash = xxh64_merge_round(hash, accumulator2);
+        hash = xxh64_merge_round(hash, accumulator3);
+        hash = xxh64_merge_round(hash, accumulator4);
+    } else {
+        hash = seed + XXH_PRIME64_5;
+    }
+
+    hash += (uint64_t)length;
+
+    while ((size_t)(end - position) >= 8U) {
+        uint64_t value = xxh64_round(UINT64_C(0), read_little_endian64(position));
+        hash ^= value;
+        hash = rotate_left64(hash, 27U) * XXH_PRIME64_1 + XXH_PRIME64_4;
+        position += 8;
+    }
+
+    if ((size_t)(end - position) >= 4U) {
+        hash ^= (uint64_t)read_little_endian32(position) * XXH_PRIME64_1;
+        hash = rotate_left64(hash, 23U) * XXH_PRIME64_2 + XXH_PRIME64_3;
+        position += 4;
+    }
+
+    while (position < end) {
+        hash ^= (uint64_t)(*position) * XXH_PRIME64_5;
+        hash = rotate_left64(hash, 11U) * XXH_PRIME64_1;
+        ++position;
+    }
+
+    hash ^= hash >> 33U;
+    hash *= XXH_PRIME64_2;
+    hash ^= hash >> 29U;
+    hash *= XXH_PRIME64_3;
+    hash ^= hash >> 32U;
+    return hash;
+}
+
+static int is_ascii_trailing_space(unsigned char character)
+{
+    return character == ' ' || character == '\t' || character == '\n' ||
+           character == '\r' || character == '\v' || character == '\f';
+}
+
+static int read_stdin(unsigned char **data, size_t *length)
+{
+    size_t capacity = 256;
+    unsigned char *buffer = malloc(capacity);
+
+    if (buffer == NULL) {
+        return -1;
+    }
+
+    *length = 0;
+    for (;;) {
+        size_t available = capacity - *length;
+        size_t bytes_read = fread(buffer + *length, 1, available, stdin);
+        *length += bytes_read;
+
+        if (bytes_read < available) {
+            if (ferror(stdin)) {
+                free(buffer);
+                return -1;
+            }
+            break;
+        }
+
+        if (capacity > SIZE_MAX / 2U) {
+            free(buffer);
+            return -1;
+        }
+        capacity *= 2U;
+
+        {
+            unsigned char *larger_buffer = realloc(buffer, capacity);
+            if (larger_buffer == NULL) {
+                free(buffer);
+                return -1;
+            }
+            buffer = larger_buffer;
+        }
+    }
+
+    *data = buffer;
+    return 0;
+}
+
+int main(int argc, char **argv)
+{
+    const unsigned char *word;
+    unsigned char *stdin_buffer = NULL;
+    size_t length;
+
+    if (argc > 2) {
+        fprintf(stderr, "usage: %s [ASCII_WORD]\n", argv[0]);
+        return EXIT_FAILURE;
+    }
+
+    if (argc == 2) {
+        word = (const unsigned char *)argv[1];
+        length = strlen(argv[1]);
+    } else {
+        if (read_stdin(&stdin_buffer, &length) != 0) {
+            fprintf(stderr, "failed to read input\n");
+            return EXIT_FAILURE;
+        }
+        word = stdin_buffer;
+    }
+
+    while (length > 0U && is_ascii_trailing_space(word[length - 1U])) {
+        --length;
+    }
+
+    for (size_t index = 0; index < length; ++index) {
+        if (word[index] > 0x7fU) {
+            fprintf(stderr, "input must contain ASCII characters only\n");
+            free(stdin_buffer);
+            return EXIT_FAILURE;
+        }
+    }
+
+    printf("%016" PRIx64 "\n", xxh64(word, length, XXH64_SEED));
+    free(stdin_buffer);
+    return EXIT_SUCCESS;
+}
@@ -33,6 +33,7 @@ DEFAULT_OUTPUT_DIR = ROOT / "probability_maps"
 # Add directory names from hash_funcs here to include more implementations.
 HASHES = [
    "jenkinsOAAT",
+    "xxh64",
 ]

 HEX_HASH = re.compile(r"(?:0[xX])?([0-9a-fA-F]+)")
@@ -84,9 +85,7 @@ def prepare_hash(hash_name: str, hash_funcs_dir: Path = HASH_FUNCS_DIR) -> Path:
        result = subprocess.run(command, text=True, capture_output=True)
        if result.returncode != 0:
            details = result.stderr.strip() or result.stdout.strip()
-            raise HashToolError(
-                f"не удалось скомпилировать {source}: {details}"
-            )
+            raise HashToolError(f"не удалось скомпилировать {source}: {details}")
        return binary

    python_source = hash_directory / "bin_hash.py"
@@ -98,10 +97,7 @@ def prepare_hash(hash_name: str, hash_funcs_dir: Path = HASH_FUNCS_DIR) -> Path:
                encoding="utf-8",
            )
        python_source.chmod(
-            python_source.stat().st_mode
-            | stat.S_IXUSR
-            | stat.S_IXGRP
-            | stat.S_IXOTH
+            python_source.stat().st_mode | stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH
        )
        return python_source

@@ -131,9 +127,7 @@ def run_hash(executable: Path, word: str) -> tuple[int, int]:
    output = result.stdout.strip()
    match = HEX_HASH.fullmatch(output)
    if match is None:
-        raise HashToolError(
-            f"{executable} вернул не шестнадцатеричный хэш: {output!r}"
-        )
+        raise HashToolError(f"{executable} вернул не шестнадцатеричный хэш: {output!r}")

    digits = match.group(1)
    return int(digits, 16), len(digits) * 4
@@ -169,9 +163,7 @@ def write_csv_table(stream: TextIO, rows: Mapping[str, ProbabilityRow]) -> None:
    bits = widths.pop()

    writer = csv.writer(stream, lineterminator="\n")
-    writer.writerow(
-        ["operation", "pairs", *(f"bit_{bit}" for bit in range(bits))]
-    )
+    writer.writerow(["operation", "pairs", *(f"bit_{bit}" for bit in range(bits))])
    for operation, (pair_count, probabilities) in rows.items():
        writer.writerow(
            [
@@ -306,10 +298,7 @@ def build_parser() -> argparse.ArgumentParser:
        "--output",
        type=Path,
        default=DEFAULT_OUTPUT_DIR,
-        help=(
-            "каталог для CSV-таблиц "
-            f"(по умолчанию: {DEFAULT_OUTPUT_DIR})"
-        ),
+        help=("каталог для CSV-таблиц " f"(по умолчанию: {DEFAULT_OUTPUT_DIR})"),
    )
    parser.add_argument(
        "--hash",
@@ -1,5 +0,0 @@
-operation,pairs,bit_0,bit_1,bit_2,bit_3,bit_4,bit_5,bit_6,bit_7,bit_8,bit_9,bit_10,bit_11,bit_12,bit_13,bit_14,bit_15,bit_16,bit_17,bit_18,bit_19,bit_20,bit_21,bit_22,bit_23,bit_24,bit_25,bit_26,bit_27,bit_28,bit_29,bit_30,bit_31
-replace,1922,0.486472,0.498959,0.483351,0.501561,0.503122,0.495317,0.523413,0.466701,0.520812,0.496878,0.513007,0.503122,0.490114,0.510926,0.482310,0.494277,0.511446,0.505723,0.498959,0.501561,0.512487,0.509365,0.489594,0.514048,0.495317,0.519771,0.505203,0.515609,0.517690,0.489074,0.498959,0.508845
-add,2111,0.506869,0.505448,0.502132,0.502132,0.496921,0.502132,0.505921,0.505921,0.494552,0.507342,0.493605,0.502605,0.492658,0.501658,0.495026,0.508764,0.496921,0.486973,0.507816,0.518238,0.513501,0.506395,0.513974,0.488394,0.504500,0.482236,0.510185,0.520133,0.500711,0.504027,0.520133,0.497868
-first,186,0.451613,0.435484,0.440860,0.462366,0.521505,0.494624,0.510753,0.435484,0.462366,0.483871,0.505376,0.521505,0.451613,0.532258,0.462366,0.505376,0.473118,0.505376,0.446237,0.473118,0.489247,0.537634,0.440860,0.521505,0.537634,0.500000,0.548387,0.575269,0.575269,0.494624,0.462366,0.505376
-last,186,0.516129,0.500000,0.483871,0.510753,0.510753,0.483871,0.478495,0.521505,0.569892,0.500000,0.478495,0.494624,0.462366,0.521505,0.510753,0.500000,0.569892,0.494624,0.505376,0.510753,0.521505,0.478495,0.478495,0.543011,0.500000,0.548387,0.532258,0.451613,0.500000,0.500000,0.516129,0.521505
@@ -0,0 +1,90 @@
+#!/usr/bin/env python3
+"""Behavior tests for the standalone XXH64 hash CLI."""
+
+from __future__ import annotations
+
+import re
+import subprocess
+import tempfile
+import unittest
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+SOURCE = ROOT / "hash_funcs" / "xxh64" / "bin_hash.c"
+
+
+class Xxh64BinHashTests(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls) -> None:
+        cls.temporary_directory = tempfile.TemporaryDirectory()
+        cls.binary = Path(cls.temporary_directory.name) / "bin_hash"
+        subprocess.run(
+            [
+                "cc",
+                "-std=c11",
+                "-O2",
+                "-Wall",
+                "-Wextra",
+                "-Wpedantic",
+                "-Werror",
+                str(SOURCE),
+                "-o",
+                str(cls.binary),
+            ],
+            check=True,
+            text=True,
+            capture_output=True,
+        )
+
+    @classmethod
+    def tearDownClass(cls) -> None:
+        cls.temporary_directory.cleanup()
+
+    def run_hash(self, word: str | None = None, stdin: bytes | None = None) -> subprocess.CompletedProcess[bytes]:
+        command = [str(self.binary)]
+        if word is not None:
+            command.append(word)
+        return subprocess.run(command, input=stdin, capture_output=True, check=False)
+
+    def test_matches_official_xxh64_seed_zero_vectors(self) -> None:
+        vectors = {
+            "": b"ef46db3751d8e999\n",
+            "hello": b"26c7827d889f6da3\n",
+            "HashWord": b"3e26fc2935163fbe\n",
+        }
+
+        for word, expected in vectors.items():
+            with self.subTest(word=word):
+                result = self.run_hash(word)
+                self.assertEqual(result.returncode, 0)
+                self.assertEqual(result.stdout, expected)
+                self.assertRegex(result.stdout.decode(), r"^[0-9a-f]{16}\n$")
+
+    def test_argv_and_stdin_are_equivalent_and_strip_trailing_ascii_space(self) -> None:
+        argv = self.run_hash("hello")
+        stdin = self.run_hash(stdin=b"hello \t\r\n")
+
+        self.assertEqual(stdin.returncode, 0)
+        self.assertEqual(stdin.stdout, argv.stdout)
+
+    def test_handles_long_ascii_input(self) -> None:
+        payload = b"a" * 100_000
+        result = self.run_hash(stdin=payload)
+        reference = subprocess.run(
+            ["xxhsum", "-H64"], input=payload, capture_output=True, check=True
+        ).stdout.split()[0]
+
+        self.assertEqual(result.returncode, 0)
+        self.assertEqual(result.stdout.strip(), reference)
+
+    def test_rejects_non_ascii_input(self) -> None:
+        result = self.run_hash(stdin="ёж".encode())
+
+        self.assertNotEqual(result.returncode, 0)
+        self.assertIn(b"ASCII", result.stderr)
+        self.assertEqual(result.stdout, b"")
+
+
+if __name__ == "__main__":
+    unittest.main()