Paul Davis
940d8844e3
::snap_to() was intended to round a Beats value to the nearest multiple of another Beats value. It did not do that, but instead rounded down. Worse, it used Beats::operator/ which in turn uses int_div_round(), which is incorrect for a situation where we need integer truncation. The changes fix the actual arithmetic and add 2 variant functions so that the API includes round down, round up and round to nearest.
380 lines
10 KiB
C++
380 lines
10 KiB
C++
/*
|
|
* Copyright (C) 2017-2018 Paul Davis <paul@linuxaudiosystems.com>
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License along
|
|
* with this program; if not, write to the Free Software Foundation, Inc.,
|
|
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
*/
|
|
|
|
#ifndef TEMPORAL_BEATS_HPP
|
|
#define TEMPORAL_BEATS_HPP
|
|
|
|
#include <cassert>
|
|
#include <float.h>
|
|
#include <math.h>
|
|
#include <stdint.h>
|
|
#include <stdlib.h>
|
|
|
|
#include <iostream>
|
|
#include <limits>
|
|
#include <sstream>
|
|
|
|
#include "pbd/failed_constructor.h"
|
|
#include "pbd/integer_division.h"
|
|
#include "pbd/string_convert.h"
|
|
|
|
|
|
#include "temporal/visibility.h"
|
|
#include "temporal/types.h"
|
|
|
|
namespace ARDOUR {
|
|
class Variant; /* Can stay since LV2 has no way to exchange beats as anything except double */
|
|
/* these all need fixing to not use ::to_double() */
|
|
class Track;
|
|
class MidiStretch;
|
|
class MidiModel;
|
|
class AutomationList;
|
|
class MidiSource;
|
|
class MidiRegion;
|
|
class Quantize;
|
|
}
|
|
|
|
namespace Evoral {
|
|
template<typename T> class Sequence;
|
|
}
|
|
|
|
/* XXX hack friends for ::do_double() access ... remove */
|
|
|
|
class QuantizeDialog;
|
|
class NoteDrag;
|
|
class NoteCreateDrag;
|
|
|
|
namespace Temporal {
|
|
|
|
/** Musical time in beats. */
|
|
class /*LIBTEMPORAL_API*/ Beats {
|
|
public:
|
|
LIBTEMPORAL_API static const int32_t PPQN = Temporal::ticks_per_beat;
|
|
|
|
Beats() : _beats(0), _ticks(0) {}
|
|
Beats(const Beats& other) : _beats(other._beats), _ticks(other._ticks) {}
|
|
|
|
/** Normalize so ticks is within PPQN. */
|
|
void normalize() {
|
|
// First, fix negative ticks with positive beats
|
|
while (_beats > 0 && _ticks < 0) {
|
|
--_beats;
|
|
_ticks += PPQN;
|
|
}
|
|
|
|
// Now fix positive ticks with negative beats
|
|
while (_beats < 0 && _ticks > 0) {
|
|
++_beats;
|
|
_ticks -= PPQN;
|
|
}
|
|
|
|
assert ((_beats < 0 && _ticks <= 0) || (_beats > 0 && _ticks >= 0) || _beats == 0);
|
|
|
|
// Work with positive beats and ticks to normalize
|
|
const int32_t sign = _beats < 0 ? -1 : _ticks < 0 ? -1 : 1;
|
|
int32_t beats = ::abs(_beats);
|
|
int32_t ticks = ::abs(_ticks);
|
|
|
|
// Fix ticks greater than 1 beat
|
|
while (ticks >= PPQN) {
|
|
++beats;
|
|
ticks -= PPQN;
|
|
}
|
|
|
|
// Set fields with appropriate sign
|
|
_beats = sign * beats;
|
|
_ticks = sign * ticks;
|
|
}
|
|
|
|
/** Create from a precise beats:ticks pair. */
|
|
explicit Beats(int32_t b, int32_t t) : _beats(b), _ticks(t) {
|
|
normalize();
|
|
}
|
|
|
|
/** Create from a real number of beats. */
|
|
static Beats from_double (double beats) {
|
|
double whole;
|
|
const double frac = modf (beats, &whole);
|
|
return Beats (whole, (int32_t) rint (frac * PPQN));
|
|
}
|
|
|
|
/** Create from an integer number of beats. */
|
|
static Beats beats(int32_t beats) {
|
|
return Beats(beats, 0);
|
|
}
|
|
|
|
/** Create from ticks at the standard PPQN. */
|
|
static Beats ticks(int64_t ticks) {
|
|
assert (ticks/PPQN < std::numeric_limits<int32_t>::max());
|
|
return Beats (ticks / PPQN, ticks % PPQN);
|
|
}
|
|
|
|
/** Create from ticks at a given rate.
|
|
*
|
|
* Note this can also be used to create from frames by setting ppqn to the
|
|
* number of samples per beat. Note the resulting Beats will, like all
|
|
* others, have the default PPQN, so this is a potentially lossy
|
|
* conversion.
|
|
*/
|
|
static Beats ticks_at_rate(int64_t ticks, uint32_t ppqn) {
|
|
return Beats(ticks / ppqn, (ticks % ppqn) * PPQN / ppqn);
|
|
}
|
|
|
|
static int64_t make_ticks (Beats const & b) { return b.get_beats() * ticks_per_beat + b.get_ticks(); }
|
|
|
|
int64_t to_ticks() const { return (int64_t)_beats * PPQN + _ticks; }
|
|
int64_t to_ticks(uint32_t ppqn) const { return (int64_t)_beats * ppqn + (_ticks * ppqn / PPQN); }
|
|
|
|
int32_t get_beats() const { return _beats; }
|
|
int32_t get_ticks() const { return _ticks; }
|
|
|
|
Beats& operator=(double time) {
|
|
double whole;
|
|
const double frac = modf(time, &whole);
|
|
|
|
_beats = whole;
|
|
_ticks = frac * PPQN;
|
|
return *this;
|
|
}
|
|
|
|
Beats& operator=(const Beats& other) {
|
|
_beats = other._beats;
|
|
_ticks = other._ticks;
|
|
return *this;
|
|
}
|
|
|
|
public:
|
|
Beats round_up_to_multiple (Beats const & multiple) const {
|
|
return ticks (((to_ticks() + (multiple.to_ticks() - 1)) / multiple.to_ticks()) * multiple.to_ticks());
|
|
}
|
|
Beats round_to_multiple (Beats const & multiple) const {
|
|
return ticks (((to_ticks() + (int_div_round (multiple.to_ticks(), (int64_t) 2))) / multiple.to_ticks()) * multiple.to_ticks());
|
|
}
|
|
Beats round_down_to_multiple (Beats const & multiple) const {
|
|
return ticks ((to_ticks() / multiple.to_ticks()) * multiple.to_ticks());
|
|
}
|
|
|
|
Beats round_to_beat() const {
|
|
return (_ticks >= (PPQN/2)) ? Beats (_beats + 1, 0) : Beats (_beats, 0);
|
|
}
|
|
|
|
Beats round_up_to_beat() const {
|
|
return (_ticks == 0) ? *this : Beats(_beats + 1, 0);
|
|
}
|
|
|
|
Beats round_down_to_beat() const {
|
|
return Beats(_beats, 0);
|
|
}
|
|
|
|
|
|
Beats prev_beat() const {
|
|
/* always moves backwards even if currently on beat */
|
|
return Beats (_beats-1, 0);
|
|
}
|
|
|
|
Beats next_beat() const {
|
|
/* always moves forwards even if currently on beat */
|
|
return Beats (_beats+1, 0);
|
|
}
|
|
|
|
LIBTEMPORAL_API Beats round_to_subdivision (int subdivision, RoundMode dir) const;
|
|
|
|
Beats abs () const {
|
|
return Beats (::abs (_beats), ::abs (_ticks));
|
|
}
|
|
|
|
Beats diff (Beats const & other) const {
|
|
if (other > *this) {
|
|
return other - *this;
|
|
}
|
|
return *this - other;
|
|
}
|
|
|
|
inline bool operator==(const Beats& b) const {
|
|
return _beats == b._beats && _ticks == b._ticks;
|
|
}
|
|
|
|
inline bool operator==(int beats) const {
|
|
return _beats == beats;
|
|
}
|
|
|
|
inline bool operator!=(const Beats& b) const {
|
|
return !operator==(b);
|
|
}
|
|
|
|
inline bool operator<(const Beats& b) const {
|
|
return _beats < b._beats || (_beats == b._beats && _ticks < b._ticks);
|
|
}
|
|
|
|
inline bool operator<=(const Beats& b) const {
|
|
return _beats < b._beats || (_beats == b._beats && _ticks <= b._ticks);
|
|
}
|
|
|
|
inline bool operator>(const Beats& b) const {
|
|
return _beats > b._beats || (_beats == b._beats && _ticks > b._ticks);
|
|
}
|
|
|
|
inline bool operator>=(const Beats& b) const {
|
|
return _beats > b._beats || (_beats == b._beats && _ticks >= b._ticks);
|
|
}
|
|
|
|
Beats operator+(const Beats& b) const {
|
|
return Beats(_beats + b._beats, _ticks + b._ticks);
|
|
}
|
|
|
|
Beats operator-(const Beats& b) const {
|
|
return Beats(_beats - b._beats, _ticks - b._ticks);
|
|
}
|
|
|
|
Beats operator-() const {
|
|
/* must avoid normalization here, which will convert a negative
|
|
value into a valid beat position before zero, which is not
|
|
we want here.
|
|
*/
|
|
Beats b (_beats, _ticks);
|
|
b._beats = -b._beats;
|
|
b._ticks = -b._ticks;
|
|
return b;
|
|
}
|
|
|
|
Beats operator*(int32_t factor) const {return ticks (to_ticks() * factor); }
|
|
Beats operator/(int32_t factor) const { return ticks (to_ticks() / factor);}
|
|
Beats operator*(ratio_t const & factor) const {return ticks (int_div_round (to_ticks() * factor.numerator(), factor.denominator())); }
|
|
Beats operator/(ratio_t const & factor) const {return ticks (int_div_round (to_ticks() * factor.denominator(), factor.numerator())); }
|
|
|
|
Beats operator% (Beats const & b) { return Beats::ticks (to_ticks() % b.to_ticks());}
|
|
|
|
Beats operator%= (Beats const & b) {
|
|
const Beats B (Beats::ticks (to_ticks() % b.to_ticks()));
|
|
_beats = B._beats;
|
|
_ticks = B._ticks;
|
|
return *this;
|
|
}
|
|
|
|
Beats operator/ (Beats const & other) const {
|
|
return Beats::ticks (int_div_round (to_ticks(), other.to_ticks()));
|
|
}
|
|
|
|
Beats operator* (Beats const & other) const {
|
|
return Beats::ticks (to_ticks () * other.to_ticks());
|
|
}
|
|
|
|
Beats& operator+=(const Beats& b) {
|
|
_beats += b._beats;
|
|
_ticks += b._ticks;
|
|
normalize();
|
|
return *this;
|
|
}
|
|
|
|
Beats& operator-=(const Beats& b) {
|
|
_beats -= b._beats;
|
|
_ticks -= b._ticks;
|
|
normalize();
|
|
return *this;
|
|
}
|
|
|
|
bool operator!() const { return _beats == 0 && _ticks == 0; }
|
|
explicit operator bool () const { return _beats != 0 || _ticks != 0; }
|
|
|
|
static Beats one_tick() { return Beats(0, 1); }
|
|
|
|
protected:
|
|
int32_t _beats;
|
|
int32_t _ticks;
|
|
|
|
};
|
|
|
|
/* Only contexts that really, absolutely need a floating point representation
|
|
* of a Beats value should ever use this.
|
|
*/
|
|
|
|
class DoubleableBeats : public Beats
|
|
{
|
|
public:
|
|
DoubleableBeats (Beats const & b) : Beats (b) {}
|
|
double to_double() const { return (double)_beats + (_ticks / (double)PPQN); }
|
|
};
|
|
|
|
|
|
/*
|
|
TIL, several horrible hours later, that sometimes the compiler looks in the
|
|
namespace of a type (Temporal::Beats in this case) for an operator, and
|
|
does *NOT* look in the global namespace.
|
|
|
|
C++ is proof that hell exists and we are living in it. In any case, move
|
|
these to the global namespace and PBD::Property's loopy
|
|
virtual-method-in-a-template will bite you.
|
|
*/
|
|
|
|
LIBTEMPORAL_API std::ostream& operator<<(std::ostream& ostream, const Temporal::Beats& t);
|
|
LIBTEMPORAL_API std::istream& operator>>(std::istream& istream, Temporal::Beats& b);
|
|
|
|
} // namespace Temporal
|
|
|
|
namespace std {
|
|
template<>
|
|
struct numeric_limits<Temporal::Beats> {
|
|
static Temporal::Beats lowest() {
|
|
return Temporal::Beats(std::numeric_limits<int32_t>::min(), std::numeric_limits<int32_t>::min());
|
|
}
|
|
|
|
/* We don't define min() since this has different behaviour for
|
|
integral and floating point types, but Beats is used as both
|
|
an integral and "fractional" value, so the semantics of
|
|
min() would be unclear.
|
|
|
|
Better to avoid providing a min at all than a confusing one.
|
|
*/
|
|
|
|
/* We must make the number of beats be 1 less than INT32_MAX,
|
|
* because otherwise adding the PPQN-1 ticks would cause
|
|
* overflow (the value would be INT32_MAX+((PPQN-1)/PPQN) which
|
|
* exceeds INT32_MAX.
|
|
*/
|
|
|
|
static Temporal::Beats max() {
|
|
return Temporal::Beats(std::numeric_limits<int32_t>::max() - 1, Temporal::Beats::PPQN - 1);
|
|
}
|
|
};
|
|
|
|
}
|
|
|
|
namespace PBD {
|
|
|
|
template<>
|
|
inline bool to_string (Temporal::Beats val, std::string & str)
|
|
{
|
|
std::ostringstream ostr;
|
|
ostr << val;
|
|
str = ostr.str();
|
|
return true;
|
|
}
|
|
|
|
template<>
|
|
inline bool string_to (std::string const & str, Temporal::Beats & val)
|
|
{
|
|
std::istringstream istr (str);
|
|
istr >> val;
|
|
return (bool) istr;
|
|
}
|
|
|
|
} /* end namsepace PBD */
|
|
|
|
|
|
#endif // TEMPORAL_BEATS_HPP
|