123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335 |
- /*
- * Copyright 2017 The WebRTC Project Authors. All rights reserved.
- *
- * Use of this source code is governed by a BSD-style license
- * that can be found in the LICENSE file in the root of the source
- * tree. An additional intellectual property rights grant can be found
- * in the file PATENTS. All contributing project authors may
- * be found in the AUTHORS file in the root of the source tree.
- */
- // Minimum and maximum
- // ===================
- //
- // rtc::SafeMin(x, y)
- // rtc::SafeMax(x, y)
- //
- // (These are both constexpr.)
- //
- // Accept two arguments of either any two integral or any two floating-point
- // types, and return the smaller and larger value, respectively, with no
- // truncation or wrap-around. If only one of the input types is statically
- // guaranteed to be able to represent the result, the return type is that type;
- // if either one would do, the result type is the smaller type. (One of these
- // two cases always applies.)
- //
- // * The case with one floating-point and one integral type is not allowed,
- // because the floating-point type will have greater range, but may not
- // have sufficient precision to represent the integer value exactly.)
- //
- // Clamp (a.k.a. constrain to a given interval)
- // ============================================
- //
- // rtc::SafeClamp(x, a, b)
- //
- // Accepts three arguments of any mix of integral types or any mix of
- // floating-point types, and returns the value in the closed interval [a, b]
- // that is closest to x (that is, if x < a it returns a; if x > b it returns b;
- // and if a <= x <= b it returns x). As for SafeMin() and SafeMax(), there is
- // no truncation or wrap-around. The result type
- //
- // 1. is statically guaranteed to be able to represent the result;
- //
- // 2. is no larger than the largest of the three argument types; and
- //
- // 3. has the same signedness as the type of the first argument, if this is
- // possible without violating the First or Second Law.
- //
- // There is always at least one type that meets criteria 1 and 2. If more than
- // one type meets these criteria equally well, the result type is one of the
- // types that is smallest. Note that unlike SafeMin() and SafeMax(),
- // SafeClamp() will sometimes pick a return type that isn't the type of any of
- // its arguments.
- //
- // * In this context, a type A is smaller than a type B if it has a smaller
- // range; that is, if A::max() - A::min() < B::max() - B::min(). For
- // example, int8_t < int16_t == uint16_t < int32_t, and all integral types
- // are smaller than all floating-point types.)
- //
- // * As for SafeMin and SafeMax, mixing integer and floating-point arguments
- // is not allowed, because floating-point types have greater range than
- // integer types, but do not have sufficient precision to represent the
- // values of most integer types exactly.
- //
- // Requesting a specific return type
- // =================================
- //
- // All three functions allow callers to explicitly specify the return type as a
- // template parameter, overriding the default return type. E.g.
- //
- // rtc::SafeMin<int>(x, y) // returns an int
- //
- // If the requested type is statically guaranteed to be able to represent the
- // result, then everything's fine, and the return type is as requested. But if
- // the requested type is too small, a static_assert is triggered.
- #ifndef RTC_BASE_NUMERICS_SAFE_MINMAX_H_
- #define RTC_BASE_NUMERICS_SAFE_MINMAX_H_
- #include <limits>
- #include <type_traits>
- #include "rtc_base/checks.h"
- #include "rtc_base/numerics/safe_compare.h"
- #include "rtc_base/type_traits.h"
- namespace rtc {
- namespace safe_minmax_impl {
- // Make the range of a type available via something other than a constexpr
- // function, to work around MSVC limitations. See
- // https://blogs.msdn.microsoft.com/vcblog/2015/12/02/partial-support-for-expression-sfinae-in-vs-2015-update-1/
- template <typename T>
- struct Limits {
- static constexpr T lowest = std::numeric_limits<T>::lowest();
- static constexpr T max = std::numeric_limits<T>::max();
- };
- template <typename T, bool is_enum = std::is_enum<T>::value>
- struct UnderlyingType;
- template <typename T>
- struct UnderlyingType<T, false> {
- using type = T;
- };
- template <typename T>
- struct UnderlyingType<T, true> {
- using type = typename std::underlying_type<T>::type;
- };
- // Given two types T1 and T2, find types that can hold the smallest (in
- // ::min_t) and the largest (in ::max_t) of the two values.
- template <typename T1,
- typename T2,
- bool int1 = IsIntlike<T1>::value,
- bool int2 = IsIntlike<T2>::value>
- struct MType {
- static_assert(int1 == int2,
- "You may not mix integral and floating-point arguments");
- };
- // Specialization for when neither type is integral (and therefore presumably
- // floating-point).
- template <typename T1, typename T2>
- struct MType<T1, T2, false, false> {
- using min_t = typename std::common_type<T1, T2>::type;
- static_assert(std::is_same<min_t, T1>::value ||
- std::is_same<min_t, T2>::value,
- "");
- using max_t = typename std::common_type<T1, T2>::type;
- static_assert(std::is_same<max_t, T1>::value ||
- std::is_same<max_t, T2>::value,
- "");
- };
- // Specialization for when both types are integral.
- template <typename T1, typename T2>
- struct MType<T1, T2, true, true> {
- // The type with the lowest minimum value. In case of a tie, the type with
- // the lowest maximum value. In case that too is a tie, the types have the
- // same range, and we arbitrarily pick T1.
- using min_t = typename std::conditional<
- SafeLt(Limits<T1>::lowest, Limits<T2>::lowest),
- T1,
- typename std::conditional<
- SafeGt(Limits<T1>::lowest, Limits<T2>::lowest),
- T2,
- typename std::conditional<SafeLe(Limits<T1>::max, Limits<T2>::max),
- T1,
- T2>::type>::type>::type;
- static_assert(std::is_same<min_t, T1>::value ||
- std::is_same<min_t, T2>::value,
- "");
- // The type with the highest maximum value. In case of a tie, the types have
- // the same range (because in C++, integer types with the same maximum also
- // have the same minimum).
- static_assert(SafeNe(Limits<T1>::max, Limits<T2>::max) ||
- SafeEq(Limits<T1>::lowest, Limits<T2>::lowest),
- "integer types with the same max should have the same min");
- using max_t = typename std::
- conditional<SafeGe(Limits<T1>::max, Limits<T2>::max), T1, T2>::type;
- static_assert(std::is_same<max_t, T1>::value ||
- std::is_same<max_t, T2>::value,
- "");
- };
- // A dummy type that we pass around at compile time but never actually use.
- // Declared but not defined.
- struct DefaultType;
- // ::type is A, except we fall back to B if A is DefaultType. We static_assert
- // that the chosen type can hold all values that B can hold.
- template <typename A, typename B>
- struct TypeOr {
- using type = typename std::
- conditional<std::is_same<A, DefaultType>::value, B, A>::type;
- static_assert(SafeLe(Limits<type>::lowest, Limits<B>::lowest) &&
- SafeGe(Limits<type>::max, Limits<B>::max),
- "The specified type isn't large enough");
- static_assert(IsIntlike<type>::value == IsIntlike<B>::value &&
- std::is_floating_point<type>::value ==
- std::is_floating_point<type>::value,
- "float<->int conversions not allowed");
- };
- } // namespace safe_minmax_impl
- template <
- typename R = safe_minmax_impl::DefaultType,
- typename T1 = safe_minmax_impl::DefaultType,
- typename T2 = safe_minmax_impl::DefaultType,
- typename R2 = typename safe_minmax_impl::TypeOr<
- R,
- typename safe_minmax_impl::MType<
- typename safe_minmax_impl::UnderlyingType<T1>::type,
- typename safe_minmax_impl::UnderlyingType<T2>::type>::min_t>::type>
- constexpr R2 SafeMin(T1 a, T2 b) {
- static_assert(IsIntlike<T1>::value || std::is_floating_point<T1>::value,
- "The first argument must be integral or floating-point");
- static_assert(IsIntlike<T2>::value || std::is_floating_point<T2>::value,
- "The second argument must be integral or floating-point");
- return SafeLt(a, b) ? static_cast<R2>(a) : static_cast<R2>(b);
- }
- template <
- typename R = safe_minmax_impl::DefaultType,
- typename T1 = safe_minmax_impl::DefaultType,
- typename T2 = safe_minmax_impl::DefaultType,
- typename R2 = typename safe_minmax_impl::TypeOr<
- R,
- typename safe_minmax_impl::MType<
- typename safe_minmax_impl::UnderlyingType<T1>::type,
- typename safe_minmax_impl::UnderlyingType<T2>::type>::max_t>::type>
- constexpr R2 SafeMax(T1 a, T2 b) {
- static_assert(IsIntlike<T1>::value || std::is_floating_point<T1>::value,
- "The first argument must be integral or floating-point");
- static_assert(IsIntlike<T2>::value || std::is_floating_point<T2>::value,
- "The second argument must be integral or floating-point");
- return SafeGt(a, b) ? static_cast<R2>(a) : static_cast<R2>(b);
- }
- namespace safe_minmax_impl {
- // Given three types T, L, and H, let ::type be a suitable return value for
- // SafeClamp(T, L, H). See the docs at the top of this file for details.
- template <typename T,
- typename L,
- typename H,
- bool int1 = IsIntlike<T>::value,
- bool int2 = IsIntlike<L>::value,
- bool int3 = IsIntlike<H>::value>
- struct ClampType {
- static_assert(int1 == int2 && int1 == int3,
- "You may not mix integral and floating-point arguments");
- };
- // Specialization for when all three types are floating-point.
- template <typename T, typename L, typename H>
- struct ClampType<T, L, H, false, false, false> {
- using type = typename std::common_type<T, L, H>::type;
- };
- // Specialization for when all three types are integral.
- template <typename T, typename L, typename H>
- struct ClampType<T, L, H, true, true, true> {
- private:
- // Range of the return value. The return type must be able to represent this
- // full range.
- static constexpr auto r_min =
- SafeMax(Limits<L>::lowest, SafeMin(Limits<H>::lowest, Limits<T>::lowest));
- static constexpr auto r_max =
- SafeMin(Limits<H>::max, SafeMax(Limits<L>::max, Limits<T>::max));
- // Is the given type an acceptable return type? (That is, can it represent
- // all possible return values, and is it no larger than the largest of the
- // input types?)
- template <typename A>
- struct AcceptableType {
- private:
- static constexpr bool not_too_large = sizeof(A) <= sizeof(L) ||
- sizeof(A) <= sizeof(H) ||
- sizeof(A) <= sizeof(T);
- static constexpr bool range_contained =
- SafeLe(Limits<A>::lowest, r_min) && SafeLe(r_max, Limits<A>::max);
- public:
- static constexpr bool value = not_too_large && range_contained;
- };
- using best_signed_type = typename std::conditional<
- AcceptableType<int8_t>::value,
- int8_t,
- typename std::conditional<
- AcceptableType<int16_t>::value,
- int16_t,
- typename std::conditional<AcceptableType<int32_t>::value,
- int32_t,
- int64_t>::type>::type>::type;
- using best_unsigned_type = typename std::conditional<
- AcceptableType<uint8_t>::value,
- uint8_t,
- typename std::conditional<
- AcceptableType<uint16_t>::value,
- uint16_t,
- typename std::conditional<AcceptableType<uint32_t>::value,
- uint32_t,
- uint64_t>::type>::type>::type;
- public:
- // Pick the best type, preferring the same signedness as T but falling back
- // to the other one if necessary.
- using type = typename std::conditional<
- std::is_signed<T>::value,
- typename std::conditional<AcceptableType<best_signed_type>::value,
- best_signed_type,
- best_unsigned_type>::type,
- typename std::conditional<AcceptableType<best_unsigned_type>::value,
- best_unsigned_type,
- best_signed_type>::type>::type;
- static_assert(AcceptableType<type>::value, "");
- };
- } // namespace safe_minmax_impl
- template <
- typename R = safe_minmax_impl::DefaultType,
- typename T = safe_minmax_impl::DefaultType,
- typename L = safe_minmax_impl::DefaultType,
- typename H = safe_minmax_impl::DefaultType,
- typename R2 = typename safe_minmax_impl::TypeOr<
- R,
- typename safe_minmax_impl::ClampType<
- typename safe_minmax_impl::UnderlyingType<T>::type,
- typename safe_minmax_impl::UnderlyingType<L>::type,
- typename safe_minmax_impl::UnderlyingType<H>::type>::type>::type>
- R2 SafeClamp(T x, L min, H max) {
- static_assert(IsIntlike<H>::value || std::is_floating_point<H>::value,
- "The first argument must be integral or floating-point");
- static_assert(IsIntlike<T>::value || std::is_floating_point<T>::value,
- "The second argument must be integral or floating-point");
- static_assert(IsIntlike<L>::value || std::is_floating_point<L>::value,
- "The third argument must be integral or floating-point");
- RTC_DCHECK_LE(min, max);
- return SafeLe(x, min)
- ? static_cast<R2>(min)
- : SafeGe(x, max) ? static_cast<R2>(max) : static_cast<R2>(x);
- }
- } // namespace rtc
- #endif // RTC_BASE_NUMERICS_SAFE_MINMAX_H_
|