mirror of
https://review.haiku-os.org/buildtools
synced 2025-02-12 08:47:41 +01:00
Updated dependencies: * GMP 6.2.1 * ISL 0.24 * MPL 1.2.1 * MPFR 4.1.0 The dependencies were pulled in by running the ./contrib/download_prerequisites script and then manually removing the symbolic links and archives, and renaming the directories (i.e mv isl-0.24 to isl)
803 lines
28 KiB
D
803 lines
28 KiB
D
/**
|
|
* Written in the D programming language.
|
|
* This module provides functions to uniform calculating hash values for different types
|
|
*
|
|
* Copyright: Copyright Igor Stepanov 2013-2013.
|
|
* License: $(HTTP www.boost.org/LICENSE_1_0.txt, Boost License 1.0).
|
|
* Authors: Igor Stepanov
|
|
* Source: $(DRUNTIMESRC core/internal/_hash.d)
|
|
*/
|
|
module core.internal.hash;
|
|
|
|
import core.internal.traits : Unconst;
|
|
|
|
// If true ensure that positive zero and negative zero have the same hash.
|
|
// Historically typeid(float).getHash did this but hashOf(float) did not.
|
|
private enum floatCoalesceZeroes = true;
|
|
// If true ensure that all NaNs of the same floating point type have the same hash.
|
|
// Historically typeid(float).getHash didn't do this but hashOf(float) did.
|
|
private enum floatCoalesceNaNs = true;
|
|
|
|
// If either of the above are true then no struct or array that contains the
|
|
// representation of a floating point number may be hashed with `bytesHash`.
|
|
|
|
@nogc nothrow pure @safe unittest
|
|
{
|
|
static if (floatCoalesceZeroes)
|
|
assert(hashOf(+0.0) == hashOf(-0.0)); // Same hash for +0.0 and -0.0.
|
|
static if (floatCoalesceNaNs)
|
|
assert(hashOf(double.nan) == hashOf(-double.nan)); // Same hash for different NaN.
|
|
}
|
|
|
|
private enum hasCallableToHash(T) = __traits(compiles,
|
|
{
|
|
size_t hash = ((T* x) => (*x).toHash())(null);
|
|
});
|
|
|
|
@nogc nothrow pure @safe unittest
|
|
{
|
|
static struct S { size_t toHash() { return 4; } }
|
|
assert(hasCallableToHash!S);
|
|
assert(!hasCallableToHash!(shared const S));
|
|
}
|
|
|
|
private enum isFinalClassWithAddressBasedHash(T) = __traits(isFinalClass, T)
|
|
// Use __traits(compiles, ...) in case there are multiple overloads of `toHash`.
|
|
&& __traits(compiles, {static assert(&Object.toHash is &T.toHash);});
|
|
|
|
@nogc nothrow pure @safe unittest
|
|
{
|
|
static class C1 {}
|
|
final static class C2 : C1 {}
|
|
final static class C3 : C1 { override size_t toHash() const nothrow { return 1; }}
|
|
static assert(!isFinalClassWithAddressBasedHash!Object);
|
|
static assert(!isFinalClassWithAddressBasedHash!C1);
|
|
static assert(isFinalClassWithAddressBasedHash!C2);
|
|
static assert(!isFinalClassWithAddressBasedHash!C3);
|
|
}
|
|
|
|
private template isCppClassWithoutHash(T)
|
|
{
|
|
static if (!is(T == class) && !is(T == interface))
|
|
enum isCppClassWithoutHash = false;
|
|
else
|
|
enum bool isCppClassWithoutHash = __traits(getLinkage, T) == "C++"
|
|
&& !is(immutable T* : immutable Object*) && !hasCallableToHash!T;
|
|
}
|
|
|
|
/+
|
|
Is it valid to calculate a hash code for T based on the bits of its
|
|
representation? Always false for interfaces, dynamic arrays, and
|
|
associative arrays. False for all classes except final classes that do
|
|
not override `toHash`.
|
|
|
|
Note: according to the spec as of
|
|
https://github.com/dlang/dlang.org/commit/d66eff16491b0664c0fc00ba80a7aa291703f1f2
|
|
the contents of unnamed paddings between fields is undefined. Currently
|
|
this hashing implementation assumes that the padding contents (if any)
|
|
for all instances of `T` are the same. The correctness of this
|
|
assumption is yet to be verified.
|
|
+/
|
|
private template canBitwiseHash(T)
|
|
{
|
|
static if (is(T EType == enum))
|
|
enum canBitwiseHash = .canBitwiseHash!EType;
|
|
else static if (__traits(isFloating, T))
|
|
enum canBitwiseHash = !(floatCoalesceZeroes || floatCoalesceNaNs);
|
|
else static if (__traits(isScalar, T))
|
|
enum canBitwiseHash = true;
|
|
else static if (is(T == class))
|
|
{
|
|
enum canBitwiseHash = isFinalClassWithAddressBasedHash!T || isCppClassWithoutHash!T;
|
|
}
|
|
else static if (is(T == interface))
|
|
{
|
|
enum canBitwiseHash = isCppClassWithoutHash!T;
|
|
}
|
|
else static if (is(T == struct))
|
|
{
|
|
static if (hasCallableToHash!T || __traits(isNested, T))
|
|
enum canBitwiseHash = false;
|
|
else
|
|
{
|
|
import core.internal.traits : allSatisfy;
|
|
enum canBitwiseHash = allSatisfy!(.canBitwiseHash, typeof(T.tupleof));
|
|
}
|
|
}
|
|
else static if (is(T == union))
|
|
{
|
|
// Right now we always bytewise hash unions that lack callable `toHash`.
|
|
enum canBitwiseHash = !hasCallableToHash!T;
|
|
}
|
|
else static if (is(T E : E[]))
|
|
{
|
|
static if (__traits(isStaticArray, T))
|
|
enum canBitwiseHash = (T.length == 0) || .canBitwiseHash!E;
|
|
else
|
|
enum canBitwiseHash = false;
|
|
}
|
|
else static if (__traits(isAssociativeArray, T))
|
|
{
|
|
enum canBitwiseHash = false;
|
|
}
|
|
else
|
|
{
|
|
static assert(is(T == delegate) || is(T : void) || is(T : typeof(null)),
|
|
"Internal error: unanticipated type "~T.stringof);
|
|
enum canBitwiseHash = true;
|
|
}
|
|
}
|
|
|
|
//enum hash. CTFE depends on base type
|
|
size_t hashOf(T)(auto ref T val, size_t seed = 0)
|
|
if (is(T == enum) && !__traits(isScalar, T))
|
|
{
|
|
static if (is(T EType == enum)) {} //for EType
|
|
return hashOf(cast(EType) val, seed);
|
|
}
|
|
|
|
//CTFE ready (depends on base type).
|
|
size_t hashOf(T)(scope const auto ref T val, size_t seed = 0)
|
|
if (!is(T == enum) && __traits(isStaticArray, T) && canBitwiseHash!T)
|
|
{
|
|
import core.internal.convert : toUbyte;
|
|
// FIXME:
|
|
// We would like to to do this:
|
|
//
|
|
//static if (T.length == 0)
|
|
// return seed;
|
|
//else static if (T.length == 1)
|
|
// return hashOf(val[0], seed);
|
|
//else
|
|
// return bytesHashWithExactSizeAndAlignment!T(toUbyte(val), seed);
|
|
//
|
|
// ... but that's inefficient when using a runtime TypeInfo (introduces a branch)
|
|
// and PR #2243 wants typeid(T).getHash(&val) to produce the same result as
|
|
// hashOf(val).
|
|
static if (T.length == 0)
|
|
{
|
|
return bytesHashAlignedBy!size_t((ubyte[]).init, seed);
|
|
}
|
|
static if (is(typeof(toUbyte(val)) == const(ubyte)[]))
|
|
{
|
|
return bytesHashAlignedBy!T(toUbyte(val), seed);
|
|
}
|
|
else //Other types. CTFE unsupported
|
|
{
|
|
assert(!__ctfe, "unable to compute hash of "~T.stringof~" at compile time");
|
|
return bytesHashAlignedBy!T((cast(const(ubyte)*) &val)[0 .. T.sizeof], seed);
|
|
}
|
|
}
|
|
|
|
//CTFE ready (depends on base type).
|
|
size_t hashOf(T)(auto ref T val, size_t seed = 0)
|
|
if (!is(T == enum) && __traits(isStaticArray, T) && !canBitwiseHash!T)
|
|
{
|
|
// FIXME:
|
|
// We would like to to do this:
|
|
//
|
|
//static if (T.length == 0)
|
|
// return seed;
|
|
//else static if (T.length == 1)
|
|
// return hashOf(val[0], seed);
|
|
//else
|
|
// /+ hash like a dynamic array +/
|
|
//
|
|
// ... but that's inefficient when using a runtime TypeInfo (introduces a branch)
|
|
// and PR #2243 wants typeid(T).getHash(&val) to produce the same result as
|
|
// hashOf(val).
|
|
return hashOf(val[], seed);
|
|
}
|
|
|
|
//dynamic array hash
|
|
size_t hashOf(T)(scope const T val, size_t seed = 0)
|
|
if (is(T == S[], S) && (__traits(isScalar, S) || canBitwiseHash!S)) // excludes enum types
|
|
{
|
|
import core.internal.convert : toUbyte;
|
|
alias ElementType = typeof(val[0]);
|
|
static if (!canBitwiseHash!ElementType)
|
|
{
|
|
size_t hash = seed;
|
|
foreach (ref o; val)
|
|
{
|
|
hash = hashOf(hashOf(o), hash); // double hashing to match TypeInfo.getHash
|
|
}
|
|
return hash;
|
|
}
|
|
else static if (is(typeof(toUbyte(val)) == const(ubyte)[]))
|
|
//ubyteble array (arithmetic types and structs without toHash) CTFE ready for arithmetic types and structs without reference fields
|
|
{
|
|
return bytesHashAlignedBy!ElementType(toUbyte(val), seed);
|
|
}
|
|
else //Other types. CTFE unsupported
|
|
{
|
|
assert(!__ctfe, "unable to compute hash of "~T.stringof~" at compile time");
|
|
return bytesHashAlignedBy!ElementType((cast(const(ubyte)*) val.ptr)[0 .. ElementType.sizeof*val.length], seed);
|
|
}
|
|
}
|
|
|
|
//dynamic array hash
|
|
size_t hashOf(T)(T val, size_t seed = 0)
|
|
if (is(T == S[], S) && !(__traits(isScalar, S) || canBitwiseHash!S)) // excludes enum types
|
|
{
|
|
size_t hash = seed;
|
|
foreach (ref o; val)
|
|
{
|
|
hash = hashOf(hashOf(o), hash); // double hashing because TypeInfo.getHash doesn't allow to pass seed value
|
|
}
|
|
return hash;
|
|
}
|
|
|
|
// Indicates if F is a built-in complex number type.
|
|
private F coalesceFloat(F)(const F val)
|
|
if (__traits(isFloating, val) && !is(F == __vector) && !is(F : creal))
|
|
{
|
|
static if (floatCoalesceZeroes)
|
|
if (val == cast(F) 0)
|
|
return cast(F) 0;
|
|
static if (floatCoalesceNaNs)
|
|
if (val != val)
|
|
return F.nan;
|
|
return val;
|
|
}
|
|
|
|
//scalar type hash
|
|
@trusted @nogc nothrow pure
|
|
size_t hashOf(T)(scope const T val) if (__traits(isScalar, T) && !is(T == __vector))
|
|
{
|
|
static if (is(T V : V*))
|
|
{
|
|
if (__ctfe)
|
|
{
|
|
if (val is null) return 0;
|
|
assert(0, "Unable to calculate hash of non-null pointer at compile time");
|
|
}
|
|
size_t result = cast(size_t) val;
|
|
return result ^ (result >> 4);
|
|
}
|
|
else static if (is(T EType == enum) && is(typeof(val[0])))
|
|
{
|
|
// Enum type whose base type is vector.
|
|
return hashOf(cast(EType) val);
|
|
}
|
|
else static if (__traits(isIntegral, T))
|
|
{
|
|
static if (T.sizeof <= size_t.sizeof)
|
|
return val;
|
|
else
|
|
return cast(size_t) (val ^ (val >>> (size_t.sizeof * 8)));
|
|
}
|
|
else static if (is(T : creal))
|
|
{
|
|
return hashOf(coalesceFloat(val.re), hashOf(coalesceFloat(val.im)));
|
|
}
|
|
else
|
|
{
|
|
static assert(__traits(isFloating, T));
|
|
auto data = coalesceFloat(val);
|
|
static if (T.sizeof == float.sizeof && T.mant_dig == float.mant_dig)
|
|
return *cast(const uint*) &data;
|
|
else static if (T.sizeof == double.sizeof && T.mant_dig == double.mant_dig)
|
|
return hashOf(*cast(const ulong*) &data);
|
|
else
|
|
{
|
|
import core.internal.convert : floatSize, toUbyte;
|
|
return bytesHashWithExactSizeAndAlignment!T(toUbyte(data)[0 .. floatSize!T], 0);
|
|
}
|
|
}
|
|
}
|
|
|
|
//scalar type hash
|
|
@trusted @nogc nothrow pure
|
|
size_t hashOf(T)(scope const T val, size_t seed) if (__traits(isScalar, T) && !is(T == __vector))
|
|
{
|
|
static if (is(T V : V*))
|
|
{
|
|
if (__ctfe)
|
|
{
|
|
if (val is null) return hashOf(size_t(0), seed);
|
|
assert(0, "Unable to calculate hash of non-null pointer at compile time");
|
|
}
|
|
return hashOf(cast(size_t) val, seed);
|
|
}
|
|
else static if (is(T EType == enum) && is(typeof(val[0])))
|
|
{
|
|
// Enum type whose base type is vector.
|
|
return hashOf(cast(EType) val, seed);
|
|
}
|
|
else static if (__traits(isIntegral, val) && T.sizeof <= size_t.sizeof)
|
|
{
|
|
static if (size_t.sizeof < ulong.sizeof)
|
|
{
|
|
//MurmurHash3 32-bit single round
|
|
enum uint c1 = 0xcc9e2d51;
|
|
enum uint c2 = 0x1b873593;
|
|
enum uint c3 = 0xe6546b64;
|
|
enum uint r1 = 15;
|
|
enum uint r2 = 13;
|
|
}
|
|
else
|
|
{
|
|
//Half of MurmurHash3 64-bit single round
|
|
//(omits second interleaved update)
|
|
enum ulong c1 = 0x87c37b91114253d5;
|
|
enum ulong c2 = 0x4cf5ad432745937f;
|
|
enum ulong c3 = 0x52dce729;
|
|
enum uint r1 = 31;
|
|
enum uint r2 = 27;
|
|
}
|
|
size_t h = c1 * val;
|
|
h = (h << r1) | (h >>> (size_t.sizeof * 8 - r1));
|
|
h = (h * c2) ^ seed;
|
|
h = (h << r2) | (h >>> (size_t.sizeof * 8 - r2));
|
|
return h * 5 + c3;
|
|
}
|
|
else static if (__traits(isIntegral, val) && T.sizeof > size_t.sizeof)
|
|
{
|
|
static foreach (i; 0 .. T.sizeof / size_t.sizeof)
|
|
seed = hashOf(cast(size_t) (val >>> (size_t.sizeof * 8 * i)), seed);
|
|
return seed;
|
|
}
|
|
else static if (is(T : creal))
|
|
{
|
|
return hashOf(val.re, hashOf(val.im, seed));
|
|
}
|
|
else static if (__traits(isFloating, T))
|
|
{
|
|
auto data = coalesceFloat(val);
|
|
static if (T.sizeof == float.sizeof && T.mant_dig == float.mant_dig)
|
|
return hashOf(*cast(const uint*) &data, seed);
|
|
else static if (T.sizeof == double.sizeof && T.mant_dig == double.mant_dig)
|
|
return hashOf(*cast(const ulong*) &data, seed);
|
|
else
|
|
{
|
|
import core.internal.convert : floatSize, toUbyte;
|
|
return bytesHashWithExactSizeAndAlignment!T(toUbyte(data)[0 .. floatSize!T], seed);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
static assert(0);
|
|
}
|
|
}
|
|
|
|
size_t hashOf(T)(scope const T val, size_t seed = 0) @safe @nogc nothrow pure
|
|
if (is(T == __vector)) // excludes enum types
|
|
{
|
|
static if (__traits(isFloating, T) && (floatCoalesceZeroes || floatCoalesceNaNs))
|
|
{
|
|
if (__ctfe)
|
|
{
|
|
// Workaround for CTFE bug.
|
|
static if (is(immutable typeof(val[0]) == immutable E, E)) {} // Get E.
|
|
E[T.sizeof / E.sizeof] array;
|
|
foreach (i; 0 .. T.sizeof / E.sizeof)
|
|
array[i] = val[i];
|
|
return hashOf(array, seed);
|
|
}
|
|
return hashOf(val.array, seed);
|
|
}
|
|
else
|
|
{
|
|
import core.internal.convert : toUbyte;
|
|
return bytesHashAlignedBy!T(toUbyte(val), seed);
|
|
}
|
|
}
|
|
|
|
//typeof(null) hash. CTFE supported
|
|
@trusted @nogc nothrow pure
|
|
size_t hashOf(T)(scope const T val) if (!is(T == enum) && is(T : typeof(null)))
|
|
{
|
|
return 0;
|
|
}
|
|
|
|
//typeof(null) hash. CTFE supported
|
|
@trusted @nogc nothrow pure
|
|
size_t hashOf(T)(scope const T val, size_t seed) if (!is(T == enum) && is(T : typeof(null)))
|
|
{
|
|
return hashOf(size_t(0), seed);
|
|
}
|
|
|
|
private enum _hashOfStruct =
|
|
q{
|
|
enum bool isChained = is(typeof(seed) : size_t);
|
|
static if (!isChained) enum size_t seed = 0;
|
|
static if (hasCallableToHash!(typeof(val))) //CTFE depends on toHash()
|
|
{
|
|
static if (!__traits(isSame, typeof(val), __traits(parent, val.toHash))
|
|
&& is(typeof(val is null)))
|
|
{
|
|
static if (isChained)
|
|
return hashOf(__traits(getMember, val, __traits(getAliasThis, typeof(val))), seed);
|
|
else
|
|
return hashOf(__traits(getMember, val, __traits(getAliasThis, typeof(val))));
|
|
}
|
|
else
|
|
{
|
|
static if (isChained)
|
|
return hashOf(cast(size_t) val.toHash(), seed);
|
|
else
|
|
return val.toHash();
|
|
}
|
|
}
|
|
else
|
|
{
|
|
import core.internal.convert : toUbyte;
|
|
static if (__traits(hasMember, T, "toHash") && is(typeof(T.toHash) == function))
|
|
{
|
|
// TODO: in the future maybe this should be changed to a static
|
|
// assert(0), because if there's a `toHash` the programmer probably
|
|
// expected it to be called and a compilation failure here will
|
|
// expose a bug in his code.
|
|
// In the future we also might want to disallow non-const toHash
|
|
// altogether.
|
|
pragma(msg, "Warning: struct "~__traits(identifier, T)
|
|
~" has method toHash, however it cannot be called with "
|
|
~typeof(val).stringof~" this.");
|
|
static if (__traits(compiles, __traits(getLocation, T.toHash)))
|
|
{
|
|
enum file = __traits(getLocation, T.toHash)[0];
|
|
enum line = __traits(getLocation, T.toHash)[1].stringof;
|
|
pragma(msg, " ",__traits(identifier, T),".toHash defined here: ",file,"(",line,")");
|
|
}
|
|
}
|
|
|
|
static if (T.tupleof.length == 0)
|
|
{
|
|
return seed;
|
|
}
|
|
else static if ((is(T == struct) && !canBitwiseHash!T) || T.tupleof.length == 1)
|
|
{
|
|
static if (isChained) size_t h = seed;
|
|
static foreach (i, F; typeof(val.tupleof))
|
|
{
|
|
static if (__traits(isStaticArray, F))
|
|
{
|
|
static if (i == 0 && !isChained) size_t h = 0;
|
|
static if (F.sizeof > 0 && canBitwiseHash!F)
|
|
// May use smallBytesHash instead of bytesHash.
|
|
h = bytesHashWithExactSizeAndAlignment!F(toUbyte(val.tupleof[i]), h);
|
|
else
|
|
// We can avoid the "double hashing" the top-level version uses
|
|
// for consistency with TypeInfo.getHash.
|
|
foreach (ref e; val.tupleof[i])
|
|
h = hashOf(e, h);
|
|
}
|
|
else static if (is(F == struct) || is(F == union))
|
|
{
|
|
static if (hasCallableToHash!F)
|
|
{
|
|
static if (!__traits(isSame, F, __traits(parent, val.tupleof[i].toHash))
|
|
&& is(typeof(val.tupleof[i] is null)))
|
|
{
|
|
static if (i == 0 && !isChained)
|
|
size_t h = hashOf(__traits(getMember, val.tupleof[i], __traits(getAliasThis, F)));
|
|
else
|
|
h = hashOf(__traits(getMember, val.tupleof[i], __traits(getAliasThis, F)), h);
|
|
}
|
|
else
|
|
{
|
|
static if (i == 0 && !isChained)
|
|
size_t h = val.tupleof[i].toHash();
|
|
else
|
|
h = hashOf(cast(size_t) val.tupleof[i].toHash(), h);
|
|
}
|
|
}
|
|
else static if (F.tupleof.length == 1)
|
|
{
|
|
// Handle the single member case separately to avoid unnecessarily using bytesHash.
|
|
static if (i == 0 && !isChained)
|
|
size_t h = hashOf(val.tupleof[i].tupleof[0]);
|
|
else
|
|
h = hashOf(val.tupleof[i].tupleof[0], h);
|
|
}
|
|
else static if (canBitwiseHash!F)
|
|
{
|
|
// May use smallBytesHash instead of bytesHash.
|
|
static if (i == 0 && !isChained) size_t h = 0;
|
|
h = bytesHashWithExactSizeAndAlignment!F(toUbyte(val.tupleof[i]), h);
|
|
}
|
|
else
|
|
{
|
|
// Nothing special happening.
|
|
static if (i == 0 && !isChained)
|
|
size_t h = hashOf(val.tupleof[i]);
|
|
else
|
|
h = hashOf(val.tupleof[i], h);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
// Nothing special happening.
|
|
static if (i == 0 && !isChained)
|
|
size_t h = hashOf(val.tupleof[i]);
|
|
else
|
|
h = hashOf(val.tupleof[i], h);
|
|
}
|
|
}
|
|
return h;
|
|
}
|
|
else static if (is(typeof(toUbyte(val)) == const(ubyte)[]))//CTFE ready for structs without reference fields
|
|
{
|
|
// Not using bytesHashWithExactSizeAndAlignment here because
|
|
// the result may differ from typeid(T).hashOf(&val).
|
|
return bytesHashAlignedBy!T(toUbyte(val), seed);
|
|
}
|
|
else // CTFE unsupported
|
|
{
|
|
assert(!__ctfe, "unable to compute hash of "~T.stringof~" at compile time");
|
|
const(ubyte)[] bytes = (() @trusted => (cast(const(ubyte)*)&val)[0 .. T.sizeof])();
|
|
// Not using bytesHashWithExactSizeAndAlignment here because
|
|
// the result may differ from typeid(T).hashOf(&val).
|
|
return bytesHashAlignedBy!T(bytes, seed);
|
|
}
|
|
}
|
|
};
|
|
|
|
//struct or union hash
|
|
size_t hashOf(T)(scope const auto ref T val, size_t seed = 0)
|
|
if (!is(T == enum) && (is(T == struct) || is(T == union))
|
|
&& !is(T == const) && !is(T == immutable)
|
|
&& canBitwiseHash!T)
|
|
{
|
|
mixin(_hashOfStruct);
|
|
}
|
|
|
|
//struct or union hash
|
|
size_t hashOf(T)(auto ref T val)
|
|
if (!is(T == enum) && (is(T == struct) || is(T == union))
|
|
&& !canBitwiseHash!T)
|
|
{
|
|
mixin(_hashOfStruct);
|
|
}
|
|
|
|
//struct or union hash
|
|
size_t hashOf(T)(auto ref T val, size_t seed)
|
|
if (!is(T == enum) && (is(T == struct) || is(T == union))
|
|
&& !canBitwiseHash!T)
|
|
{
|
|
mixin(_hashOfStruct);
|
|
}
|
|
|
|
//struct or union hash - https://issues.dlang.org/show_bug.cgi?id=19332 (support might be removed in future)
|
|
size_t hashOf(T)(scope auto ref T val, size_t seed = 0)
|
|
if (!is(T == enum) && (is(T == struct) || is(T == union))
|
|
&& (is(T == const) || is(T == immutable))
|
|
&& canBitwiseHash!T && !canBitwiseHash!(Unconst!T))
|
|
{
|
|
mixin(_hashOfStruct);
|
|
}
|
|
|
|
//delegate hash. CTFE only if null.
|
|
@trusted @nogc nothrow pure
|
|
size_t hashOf(T)(scope const T val, size_t seed = 0) if (!is(T == enum) && is(T == delegate))
|
|
{
|
|
if (__ctfe)
|
|
{
|
|
if (val is null) return hashOf(size_t(0), hashOf(size_t(0), seed));
|
|
assert(0, "unable to compute hash of "~T.stringof~" at compile time");
|
|
}
|
|
return hashOf(val.ptr, hashOf(cast(void*) val.funcptr, seed));
|
|
}
|
|
|
|
//address-based class hash. CTFE only if null.
|
|
@nogc nothrow pure @trusted
|
|
size_t hashOf(T)(scope const T val)
|
|
if (!is(T == enum) && (is(T == interface) || is(T == class))
|
|
&& canBitwiseHash!T)
|
|
{
|
|
if (__ctfe) if (val is null) return 0;
|
|
return hashOf(cast(const void*) val);
|
|
}
|
|
|
|
//address-based class hash. CTFE only if null.
|
|
@nogc nothrow pure @trusted
|
|
size_t hashOf(T)(scope const T val, size_t seed)
|
|
if (!is(T == enum) && (is(T == interface) || is(T == class))
|
|
&& canBitwiseHash!T)
|
|
{
|
|
if (__ctfe) if (val is null) return hashOf(size_t(0), seed);
|
|
return hashOf(cast(const void*) val, seed);
|
|
}
|
|
|
|
//class or interface hash. CTFE depends on toHash
|
|
size_t hashOf(T)(T val)
|
|
if (!is(T == enum) && (is(T == interface) || is(T == class))
|
|
&& !canBitwiseHash!T)
|
|
{
|
|
static if (__traits(compiles, {size_t h = val.toHash();}))
|
|
{
|
|
static if (is(__traits(parent, val.toHash) P) && !is(immutable T* : immutable P*)
|
|
&& is(typeof((ref P p) => p is null)))
|
|
return val ? hashOf(__traits(getMember, val, __traits(getAliasThis, T))) : 0;
|
|
else
|
|
return val ? val.toHash() : 0;
|
|
}
|
|
else
|
|
return val ? (cast(Object)val).toHash() : 0;
|
|
}
|
|
|
|
//class or interface hash. CTFE depends on toHash
|
|
size_t hashOf(T)(T val, size_t seed)
|
|
if (!is(T == enum) && (is(T == interface) || is(T == class))
|
|
&& !canBitwiseHash!T)
|
|
{
|
|
static if (__traits(compiles, {size_t h = val.toHash();}))
|
|
{
|
|
static if (is(__traits(parent, val.toHash) P) && !is(immutable T* : immutable P*)
|
|
&& is(typeof((ref P p) => p is null)))
|
|
return hashOf(val ? hashOf(__traits(getMember, val, __traits(getAliasThis, T)))
|
|
: size_t(0), seed);
|
|
else
|
|
return hashOf(val ? cast(size_t) val.toHash() : size_t(0), seed);
|
|
}
|
|
else
|
|
return hashOf(val ? (cast(Object)val).toHash() : 0, seed);
|
|
}
|
|
|
|
//associative array hash. CTFE depends on base types
|
|
size_t hashOf(T)(T aa) if (!is(T == enum) && __traits(isAssociativeArray, T))
|
|
{
|
|
static if (is(typeof(aa) : V[K], K, V)) {} // Put K & V in scope.
|
|
static if (__traits(compiles, (ref K k, ref V v) nothrow => .hashOf(k) + .hashOf(v)))
|
|
scope (failure) assert(0); // Allow compiler to infer nothrow.
|
|
|
|
if (!aa.length) return 0;
|
|
size_t h = 0;
|
|
|
|
// The computed hash is independent of the foreach traversal order.
|
|
foreach (ref key, ref val; aa)
|
|
h += hashOf(hashOf(val), hashOf(key));
|
|
return h;
|
|
}
|
|
|
|
//associative array hash. CTFE depends on base types
|
|
size_t hashOf(T)(T aa, size_t seed) if (!is(T == enum) && __traits(isAssociativeArray, T))
|
|
{
|
|
return hashOf(hashOf(aa), seed);
|
|
}
|
|
|
|
// MurmurHash3 was written by Austin Appleby, and is placed in the public
|
|
// domain. The author hereby disclaims copyright to this source code.
|
|
|
|
// This overload is for backwards compatibility.
|
|
@system pure nothrow @nogc
|
|
size_t bytesHash()(scope const(void)* buf, size_t len, size_t seed)
|
|
{
|
|
return bytesHashAlignedBy!ubyte((cast(const(ubyte)*) buf)[0 .. len], seed);
|
|
}
|
|
|
|
private template bytesHashAlignedBy(AlignType)
|
|
{
|
|
alias bytesHashAlignedBy = bytesHash!(AlignType.alignof >= uint.alignof);
|
|
}
|
|
|
|
private template bytesHashWithExactSizeAndAlignment(SizeAndAlignType)
|
|
{
|
|
static if (SizeAndAlignType.alignof < uint.alignof
|
|
? SizeAndAlignType.sizeof <= 12
|
|
: SizeAndAlignType.sizeof <= 10)
|
|
alias bytesHashWithExactSizeAndAlignment = smallBytesHash;
|
|
else
|
|
alias bytesHashWithExactSizeAndAlignment = bytesHashAlignedBy!SizeAndAlignType;
|
|
}
|
|
|
|
// Fowler/Noll/Vo hash. http://www.isthe.com/chongo/tech/comp/fnv/
|
|
private size_t fnv()(scope const(ubyte)[] bytes, size_t seed) @nogc nothrow pure @safe
|
|
{
|
|
static if (size_t.max <= uint.max)
|
|
enum prime = (1U << 24) + (1U << 8) + 0x93U;
|
|
else static if (size_t.max <= ulong.max)
|
|
enum prime = (1UL << 40) + (1UL << 8) + 0xb3UL;
|
|
else
|
|
enum prime = (size_t(1) << 88) + (size_t(1) << 8) + size_t(0x3b);
|
|
foreach (b; bytes)
|
|
seed = (seed ^ b) * prime;
|
|
return seed;
|
|
}
|
|
private alias smallBytesHash = fnv;
|
|
|
|
//-----------------------------------------------------------------------------
|
|
// Block read - if your platform needs to do endian-swapping or can only
|
|
// handle aligned reads, do the conversion here
|
|
private uint get32bits()(scope const(ubyte)* x) @nogc nothrow pure @system
|
|
{
|
|
version (BigEndian)
|
|
{
|
|
return ((cast(uint) x[0]) << 24) | ((cast(uint) x[1]) << 16) | ((cast(uint) x[2]) << 8) | (cast(uint) x[3]);
|
|
}
|
|
else
|
|
{
|
|
return ((cast(uint) x[3]) << 24) | ((cast(uint) x[2]) << 16) | ((cast(uint) x[1]) << 8) | (cast(uint) x[0]);
|
|
}
|
|
}
|
|
|
|
/+
|
|
Params:
|
|
dataKnownToBeAligned = whether the data is known at compile time to be uint-aligned.
|
|
+/
|
|
@nogc nothrow pure @trusted
|
|
private size_t bytesHash(bool dataKnownToBeAligned)(scope const(ubyte)[] bytes, size_t seed)
|
|
{
|
|
auto len = bytes.length;
|
|
auto data = bytes.ptr;
|
|
auto nblocks = len / 4;
|
|
|
|
uint h1 = cast(uint)seed;
|
|
|
|
enum uint c1 = 0xcc9e2d51;
|
|
enum uint c2 = 0x1b873593;
|
|
enum uint c3 = 0xe6546b64;
|
|
|
|
//----------
|
|
// body
|
|
auto end_data = data+nblocks*uint.sizeof;
|
|
for (; data!=end_data; data += uint.sizeof)
|
|
{
|
|
static if (dataKnownToBeAligned)
|
|
uint k1 = __ctfe ? get32bits(data) : *(cast(const uint*) data);
|
|
else
|
|
uint k1 = get32bits(data);
|
|
k1 *= c1;
|
|
k1 = (k1 << 15) | (k1 >> (32 - 15));
|
|
k1 *= c2;
|
|
|
|
h1 ^= k1;
|
|
h1 = (h1 << 13) | (h1 >> (32 - 13));
|
|
h1 = h1*5+c3;
|
|
}
|
|
|
|
//----------
|
|
// tail
|
|
uint k1 = 0;
|
|
|
|
switch (len & 3)
|
|
{
|
|
case 3: k1 ^= data[2] << 16; goto case;
|
|
case 2: k1 ^= data[1] << 8; goto case;
|
|
case 1: k1 ^= data[0];
|
|
k1 *= c1; k1 = (k1 << 15) | (k1 >> (32 - 15)); k1 *= c2; h1 ^= k1;
|
|
goto default;
|
|
default:
|
|
}
|
|
|
|
//----------
|
|
// finalization
|
|
h1 ^= len;
|
|
// Force all bits of the hash block to avalanche.
|
|
h1 = (h1 ^ (h1 >> 16)) * 0x85ebca6b;
|
|
h1 = (h1 ^ (h1 >> 13)) * 0xc2b2ae35;
|
|
h1 ^= h1 >> 16;
|
|
return h1;
|
|
}
|
|
|
|
// Check that bytesHash works with CTFE
|
|
pure nothrow @system @nogc unittest
|
|
{
|
|
size_t ctfeHash(string x)
|
|
{
|
|
return bytesHash(x.ptr, x.length, 0);
|
|
}
|
|
|
|
enum test_str = "Sample string";
|
|
enum size_t hashVal = ctfeHash(test_str);
|
|
assert(hashVal == bytesHash(&test_str[0], test_str.length, 0));
|
|
|
|
// Detect unintended changes to bytesHash on unaligned and aligned inputs.
|
|
version (BigEndian)
|
|
{
|
|
const ubyte[7] a = [99, 4, 3, 2, 1, 5, 88];
|
|
const uint[2] b = [0x04_03_02_01, 0x05_ff_ff_ff];
|
|
}
|
|
else
|
|
{
|
|
const ubyte[7] a = [99, 1, 2, 3, 4, 5, 88];
|
|
const uint[2] b = [0x04_03_02_01, 0xff_ff_ff_05];
|
|
}
|
|
// It is okay to change the below values if you make a change
|
|
// that you expect to change the result of bytesHash.
|
|
assert(bytesHash(&a[1], a.length - 2, 0) == 2727459272);
|
|
assert(bytesHash(&b, 5, 0) == 2727459272);
|
|
assert(bytesHashAlignedBy!uint((cast(const ubyte*) &b)[0 .. 5], 0) == 2727459272);
|
|
}
|