/*
    SuperCollider real time audio synthesis system
    Copyright (c) 2002 James McCartney. All rights reserved.
    http://www.audiosynth.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
*/


#include "SC_PlugIn.h"
#include "SIMD_Unit.hpp"
#include <limits.h>
#include <cstdio>
#include "function_attributes.h"

#include <boost/align/is_aligned.hpp>

static InterfaceTable* ft;

struct Vibrato : public Unit {
    double mPhase, m_attackSlope, m_attackLevel;
    float mFreqMul, m_scaleA, m_scaleB, mFreq;
    int m_delay, m_attack;
    float trig;
};

struct LFPulse : public Unit {
    double mPhase;
    float mFreqMul, mDuty;
};

struct LFSaw : public Unit {
    double mPhase;
    float mFreqMul;
};

struct LFPar : public Unit {
    double mPhase;
    float mFreqMul;
};

struct LFCub : public Unit {
    double mPhase;
    float mFreqMul;
};

struct LFTri : public Unit {
    double mPhase;
    float mFreqMul;
};

struct LFGauss : public Unit {
    double mPhase;
    float mDurMul;
};

struct Impulse : public Unit {
    double mPhase, mPhaseOffset, mPhaseIncrement;
    float mFreqMul;
};

struct VarSaw : public Unit {
    double mPhase;
    float mFreqMul, mDuty, mInvDuty, mInv1Duty;
};

struct SyncSaw : public Unit {
    double mPhase1, mPhase2;
    float mFreqMul;
};

struct Line : public Unit {
    double mLevel, mSlope;
    float mEndLevel;
    int mCounter;
};

struct XLine : public Unit {
    double mLevel, mGrowth;
    float mEndLevel;
    int mCounter;
};

struct Cutoff : public Unit {
    double mLevel, mSlope;
    int mWaitCounter;
};

struct LinExp : public Unit {
    float m_dstratio, m_rsrcrange, m_rrminuslo, m_dstlo;
};

struct Clip : public Unit {
    float m_lo, m_hi;
};

struct Wrap : public Unit {
    float m_lo, m_hi;
};

struct Fold : public Unit {
    float m_lo, m_hi, m_range;
};

struct Unwrap : public Unit {
    float m_range, m_half, m_offset, m_prev;
};

struct ModDif : public Unit {
    float m_dif, m_mod;
};

struct AmpComp : public Unit {
    float m_rootmul, m_exponent;
};

struct AmpCompA : public Unit {
    double m_scale, m_offset;
};

struct InRange : public Unit {
    // nothing
};

struct InRect : public Unit {
    // nothing
};

// struct Trapezoid : public Unit
//{
//  float m_leftScale, m_rightScale, m_a, m_b, m_c, m_d;
//};

struct A2K : public Unit {};

struct T2K : public Unit {};

struct T2A : public Unit {
    float mLevel;
};

struct EnvGen : public Unit {
    double m_a1, m_a2, m_b1, m_y1, m_y2, m_grow, m_level, m_endLevel;
    int m_counter, m_stage, m_shape, m_releaseNode;
    float m_prevGate;
    bool m_released;
};

struct Linen : public Unit {
    float m_endLevel;
    double m_slope, m_level;
    int m_counter, m_stage;
    float m_prevGate;
};


//////////////////////////////////////////////////////////////////////////////////////////////////


extern "C" {
void Vibrato_next(Vibrato* unit, int inNumSamples);
void Vibrato_Ctor(Vibrato* unit);

void LFPulse_next_a(LFPulse* unit, int inNumSamples);
void LFPulse_next_k(LFPulse* unit, int inNumSamples);
void LFPulse_Ctor(LFPulse* unit);

void LFSaw_next_a(LFSaw* unit, int inNumSamples);
void LFSaw_next_k(LFSaw* unit, int inNumSamples);
void LFSaw_Ctor(LFSaw* unit);

void LFTri_next_a(LFTri* unit, int inNumSamples);
void LFTri_next_k(LFTri* unit, int inNumSamples);
void LFTri_Ctor(LFTri* unit);

void LFPar_next_a(LFPar* unit, int inNumSamples);
void LFPar_next_k(LFPar* unit, int inNumSamples);
void LFPar_Ctor(LFPar* unit);

void LFCub_next_a(LFCub* unit, int inNumSamples);
void LFCub_next_k(LFCub* unit, int inNumSamples);
void LFCub_Ctor(LFCub* unit);

void LFGauss_next_a(LFGauss* unit, int inNumSamples);
void LFGauss_next_k(LFGauss* unit, int inNumSamples);
void LFGauss_next_aa(LFGauss* unit, int inNumSamples);
void LFGauss_Ctor(LFGauss* unit);

void VarSaw_next_a(VarSaw* unit, int inNumSamples);
void VarSaw_next_k(VarSaw* unit, int inNumSamples);
void VarSaw_Ctor(VarSaw* unit);

void Impulse_next_aa(Impulse* unit, int inNumSamples);
void Impulse_next_ak(Impulse* unit, int inNumSamples);
void Impulse_next_ai(Impulse* unit, int inNumSamples);
void Impulse_next_kk(Impulse* unit, int inNumSamples);
void Impulse_next_ki(Impulse* unit, int inNumSamples);
void Impulse_next_ik(Impulse* unit, int inNumSamples);
void Impulse_next_ii(Impulse* unit, int inNumSamples);
void Impulse_Ctor(Impulse* unit);

void SyncSaw_next_aa(SyncSaw* unit, int inNumSamples);
void SyncSaw_next_ak(SyncSaw* unit, int inNumSamples);
void SyncSaw_next_ka(SyncSaw* unit, int inNumSamples);
void SyncSaw_next_kk(SyncSaw* unit, int inNumSamples);
void SyncSaw_Ctor(SyncSaw* unit);

void A2K_next(A2K* unit, int inNumSamples);
void A2K_Ctor(A2K* unit);

void T2K_next(T2K* unit, int inNumSamples);
void T2K_Ctor(T2K* unit);

void T2A_next(T2A* unit, int inNumSamples);
void T2A_Ctor(T2A* unit);

void Line_next(Line* unit, int inNumSamples);
void Line_Ctor(Line* unit);

void XLine_next(XLine* unit, int inNumSamples);
void XLine_Ctor(XLine* unit);

void Wrap_next_kk(Wrap* unit, int inNumSamples);
void Wrap_next_ak(Wrap* unit, int inNumSamples);
void Wrap_next_ka(Wrap* unit, int inNumSamples);
void Wrap_next_aa(Wrap* unit, int inNumSamples);
void Wrap_Ctor(Wrap* unit);

void Fold_next_kk(Fold* unit, int inNumSamples);
void Fold_next_ak(Fold* unit, int inNumSamples);
void Fold_next_ka(Fold* unit, int inNumSamples);
void Fold_next_aa(Fold* unit, int inNumSamples);
void Fold_Ctor(Fold* unit);

void Clip_next_kk(Clip* unit, int inNumSamples);
void Clip_next_ka(Clip* unit, int inNumSamples);
void Clip_next_ak(Clip* unit, int inNumSamples);
void Clip_next_aa(Clip* unit, int inNumSamples);
void Clip_Ctor(Clip* unit);

void Unwrap_next(Unwrap* unit, int inNumSamples);
void Unwrap_Ctor(Unwrap* unit);

void ModDif_next_kk(ModDif* unit, int inNumSamples);
void ModDif_next_ak(ModDif* unit, int inNumSamples);
void ModDif_next_ka(ModDif* unit, int inNumSamples);
void ModDif_next_aa(ModDif* unit, int inNumSamples);
void ModDif_Ctor(ModDif* unit);

void AmpComp_next(AmpComp* unit, int inNumSamples);
void AmpComp_Ctor(AmpComp* unit);

void AmpCompA_next(AmpCompA* unit, int inNumSamples);
void AmpCompA_Ctor(AmpCompA* unit);

void InRange_next(InRange* unit, int inNumSamples);
void InRange_Ctor(InRange* unit);

void InRect_next(InRect* unit, int inNumSamples);
void InRect_Ctor(InRect* unit);

void LinExp_next(LinExp* unit, int inNumSamples);
void LinExp_next_kk(LinExp* unit, int inNumSamples);
void LinExp_next_ak(LinExp* unit, int inNumSamples);
void LinExp_next_ka(LinExp* unit, int inNumSamples);
void LinExp_Ctor(LinExp* unit);

void EnvGen_next_k(EnvGen* unit, int inNumSamples);
void EnvGen_next_aa(EnvGen* unit, int inNumSamples);
void EnvGen_next_ak(EnvGen* unit, int inNumSamples);
void EnvGen_Ctor(EnvGen* unit);

void Linen_next_k(Linen* unit, int inNumSamples);
void Linen_Ctor(Linen* unit);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

// in, rate, depth, rateVariation, depthVariation
// 0   1     2      3              4

void Vibrato_next(Vibrato* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);

    float curtrig = ZIN0(8);
    if (unit->trig <= 0.f && curtrig > 0.f) {
        unit->mFreqMul = 4.0 * SAMPLEDUR;
        unit->mPhase = 4.0 * sc_wrap(ZIN0(7), 0.f, 1.f) - 1.0;

        RGen& rgen = *unit->mParent->mRGen;
        float rate = ZIN0(1) * unit->mFreqMul;
        float depth = ZIN0(2);
        float rateVariation = ZIN0(5);
        float depthVariation = ZIN0(6);
        unit->mFreq = rate * (1.f + rateVariation * rgen.frand2());
        unit->m_scaleA = depth * (1.f + depthVariation * rgen.frand2());
        unit->m_scaleB = depth * (1.f + depthVariation * rgen.frand2());
        unit->m_delay = (int)(ZIN0(3) * SAMPLERATE);
        unit->m_attack = (int)(ZIN0(4) * SAMPLERATE);
        unit->m_attackSlope = 1. / (double)(1 + unit->m_attack);
        unit->m_attackLevel = unit->m_attackSlope;
    }

    unit->trig = curtrig;

    double ffreq = unit->mFreq;
    double phase = unit->mPhase;
    float scaleA = unit->m_scaleA;
    float scaleB = unit->m_scaleB;
    if (unit->m_delay > 0) {
        int remain = sc_min(inNumSamples, unit->m_delay);
        unit->m_delay -= remain;
        inNumSamples -= remain;
        LOOP(remain, ZXP(out) = ZXP(in););
        if (unit->m_delay <= 0 && inNumSamples > 0) {
            if (unit->m_attack > 0)
                goto doAttack;
            else
                goto doNormal;
        }
    } else if (unit->m_attack) {
    doAttack:
        int remain = sc_min(inNumSamples, unit->m_attack);
        unit->m_attack -= remain;
        inNumSamples -= remain;
        double attackSlope = unit->m_attackSlope;
        double attackLevel = unit->m_attackLevel;
        LOOP(
            remain,
            if (phase < 1.f) {
                float z = phase;
                ZXP(out) = ZXP(in) * (1.f + (float)attackLevel * scaleA * (1.f - z * z));
            } else if (phase < 3.f) {
                float z = phase - 2.f;
                ZXP(out) = ZXP(in) * (1.f + (float)attackLevel * scaleB * (z * z - 1.f));
            } else {
                phase -= 4.f;
                float z = phase;

                float depth = ZIN0(2);
                float rateVariation = ZIN0(5);
                float depthVariation = ZIN0(6);

                float rate = ZIN0(1) * unit->mFreqMul;
                RGen& rgen = *unit->mParent->mRGen;
                ffreq = rate * (1.f + rateVariation * rgen.frand2());
                scaleA = depth * (1.f + depthVariation * rgen.frand2());
                scaleB = depth * (1.f + depthVariation * rgen.frand2());

                ZXP(out) = ZXP(in) * (1.f + (float)attackLevel * scaleA * (1.f - z * z));
            } phase += ffreq;
            attackLevel += attackSlope;);
        unit->m_attackLevel = attackLevel;
        if (unit->m_attack <= 0 && inNumSamples > 0)
            goto doNormal;
    } else {
    doNormal:
        LOOP1(
            inNumSamples,
            if (phase < 1.f) {
                float z = phase;
                ZXP(out) = ZXP(in) * (1.f + scaleA * (1.f - z * z));
            } else if (phase < 3.f) {
                float z = phase - 2.f;
                ZXP(out) = ZXP(in) * (1.f + scaleB * (z * z - 1.f));
            } else {
                phase -= 4.f;
                float z = phase;

                float depth = ZIN0(2);
                float rateVariation = ZIN0(5);
                float depthVariation = ZIN0(6);

                float rate = ZIN0(1) * unit->mFreqMul;
                RGen& rgen = *unit->mParent->mRGen;
                ffreq = rate * (1.f + rateVariation * rgen.frand2());
                scaleA = depth * (1.f + depthVariation * rgen.frand2());
                scaleB = depth * (1.f + depthVariation * rgen.frand2());

                ZXP(out) = ZXP(in) * (1.f + scaleA * (1.f - z * z));
            } phase += ffreq;);
    }
    unit->mPhase = phase;
    unit->mFreq = ffreq;
    unit->m_scaleA = scaleA;
    unit->m_scaleB = scaleB;
}

void Vibrato_Ctor(Vibrato* unit) {
    unit->mFreqMul = 4.0 * SAMPLEDUR;
    unit->mPhase = 4.0 * sc_wrap(ZIN0(7), 0.f, 1.f) - 1.0;

    RGen& rgen = *unit->mParent->mRGen;
    float rate = ZIN0(1) * unit->mFreqMul;
    float depth = ZIN0(2);
    float rateVariation = ZIN0(5);
    float depthVariation = ZIN0(6);
    unit->mFreq = rate * (1.f + rateVariation * rgen.frand2());
    unit->m_scaleA = depth * (1.f + depthVariation * rgen.frand2());
    unit->m_scaleB = depth * (1.f + depthVariation * rgen.frand2());
    unit->m_delay = (int)(ZIN0(3) * SAMPLERATE);
    unit->m_attack = (int)(ZIN0(4) * SAMPLERATE);
    unit->m_attackSlope = 1. / (double)(1 + unit->m_attack);
    unit->m_attackLevel = unit->m_attackSlope;
    unit->trig = 0.0f;
    SETCALC(Vibrato_next);
    Vibrato_next(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////


void LFPulse_next_a(LFPulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);
    float nextDuty = ZIN0(2);
    float duty = unit->mDuty;

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;
    LOOP1(
        inNumSamples, float z; if (phase >= 1.f) {
            phase -= 1.f;
            duty = unit->mDuty = nextDuty;
            // output at least one sample from the opposite polarity
            z = duty <= 0.5f ? 1.f : 0.f;
        } else { z = phase < duty ? 1.f : 0.f; } phase += ZXP(freq) * freqmul;
        ZXP(out) = z;);

    unit->mPhase = phase;
}

void LFPulse_next_k(LFPulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;
    float nextDuty = ZIN0(2);
    float duty = unit->mDuty;

    double phase = unit->mPhase;
    LOOP1(
        inNumSamples, float z; if (phase >= 1.f) {
            phase -= 1.f;
            duty = unit->mDuty = nextDuty;
            // output at least one sample from the opposite polarity
            z = duty <= 0.5f ? 1.f : 0.f;
        } else { z = phase < duty ? 1.f : 0.f; } phase += freq;
        ZXP(out) = z;);

    unit->mPhase = phase;
}

void LFPulse_Ctor(LFPulse* unit) {
    if (INRATE(0) == calc_FullRate) {
        SETCALC(LFPulse_next_a);
    } else {
        SETCALC(LFPulse_next_k);
    }

    unit->mFreqMul = unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1);
    unit->mDuty = ZIN0(2);

    LFPulse_next_k(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

void LFSaw_next_a(LFSaw* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;
    LOOP1(inNumSamples,
          float z = phase; // out must be written last for in place operation
          phase += ZXP(freq) * freqmul; if (phase >= 1.f) phase -= 2.f; else if (phase <= -1.f) phase += 2.f;
          ZXP(out) = z;);

    unit->mPhase = phase;
}

void LFSaw_next_k(LFSaw* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;

    double phase = unit->mPhase;
    if (freq >= 0.f) {
        LOOP1(inNumSamples, ZXP(out) = phase; phase += freq; if (phase >= 1.f) phase -= 2.f;);
    } else {
        LOOP1(inNumSamples, ZXP(out) = phase; phase += freq; if (phase <= -1.f) phase += 2.f;);
    }

    unit->mPhase = phase;
}

void LFSaw_Ctor(LFSaw* unit) {
    if (INRATE(0) == calc_FullRate)
        SETCALC(LFSaw_next_a);
    else
        SETCALC(LFSaw_next_k);

    unit->mFreqMul = 2.0 * unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1);

    LFSaw_next_k(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

void LFPar_next_a(LFPar* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;
    float z, y;
    LOOP1(
        inNumSamples,
        if (phase < 1.f) {
            z = phase;
            y = 1.f - z * z;
        } else if (phase < 3.f) {
            z = phase - 2.f;
            y = z * z - 1.f;
        } else {
            phase -= 4.f;
            z = phase;
            y = 1.f - z * z;
        }
        // Note: the following two lines were originally one, but seems to compile wrong on mac
        float phaseadd = ZXP(freq);
        phase += phaseadd * freqmul; ZXP(out) = y;);

    unit->mPhase = phase;
}

void LFPar_next_k(LFPar* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;

    double phase = unit->mPhase;
    LOOP1(
        inNumSamples,
        if (phase < 1.f) {
            float z = phase;
            ZXP(out) = 1.f - z * z;
        } else if (phase < 3.f) {
            float z = phase - 2.f;
            ZXP(out) = z * z - 1.f;
        } else {
            phase -= 4.f;
            float z = phase;
            ZXP(out) = 1.f - z * z;
        } phase += freq;);

    unit->mPhase = phase;
}

void LFPar_Ctor(LFPar* unit) {
    if (INRATE(0) == calc_FullRate)
        SETCALC(LFPar_next_a);
    else
        SETCALC(LFPar_next_k);

    unit->mFreqMul = 4.0 * unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1);

    LFPar_next_k(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

void LFCub_next_a(LFCub* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;
    LOOP1(
        inNumSamples, float z; if (phase < 1.f) { z = phase; } else if (phase < 2.f) { z = 2.f - phase; } else {
            phase -= 2.f;
            z = phase;
        } float phaseadd = ZXP(freq);
        phase += phaseadd * freqmul; ZXP(out) = z * z * (6.f - 4.f * z) - 1.f;);

    unit->mPhase = phase;
}

void LFCub_next_k(LFCub* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;

    double phase = unit->mPhase;
    LOOP1(
        inNumSamples, float z; if (phase < 1.f) { z = phase; } else if (phase < 2.f) { z = 2.f - phase; } else {
            phase -= 2.f;
            z = phase;
        } ZXP(out) = z * z * (6.f - 4.f * z) - 1.f;
        phase += freq;);

    unit->mPhase = phase;
}

void LFCub_Ctor(LFCub* unit) {
    if (INRATE(0) == calc_FullRate)
        SETCALC(LFCub_next_a);
    else
        SETCALC(LFCub_next_k);

    unit->mFreqMul = 2.0 * unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1) + 0.5;

    LFCub_next_k(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

void LFTri_next_a(LFTri* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;
    LOOP1(inNumSamples, float z = phase > 1.f ? 2.f - phase : phase; phase += ZXP(freq) * freqmul;
          if (phase >= 3.f) phase -= 4.f; ZXP(out) = z;);

    unit->mPhase = phase;
}

void LFTri_next_k(LFTri* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;

    double phase = unit->mPhase;
    LOOP1(inNumSamples, float z = phase > 1.f ? 2.f - phase : phase; phase += freq; if (phase >= 3.f) phase -= 4.f;
          ZXP(out) = z;);

    unit->mPhase = phase;
}

void LFTri_Ctor(LFTri* unit) {
    if (INRATE(0) == calc_FullRate) {
        SETCALC(LFTri_next_a);
    } else {
        SETCALC(LFTri_next_k);
    }

    unit->mFreqMul = 4.0 * unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1);

    LFTri_next_k(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

void LFGauss_next_k(LFGauss* unit, int inNumSamples) {
    float* out = ZOUT(0);

    float dur = ZIN0(0);
    float c = ZIN0(1);
    float b = ZIN0(2);
    float loop = ZIN0(3);

    // offset phase by b
    double x = unit->mPhase - b;

    // for a full cycle from -1 to 1 in duration, double the step.
    float step = 2.f / (dur * unit->mRate->mSampleRate);

    // calculate exponent only once per loop
    float factor = -1.f / (2.f * c * c);

    LOOP1(
        inNumSamples,

        if (x > 1.f) {
            if (loop) {
                x -= 2.f;
            } else {
                DoneAction(ZIN0(4), unit);
            }
        } ZXP(out) = exp(x * x * factor);
        x += step;);

    unit->mPhase = x + b;
}

void LFGauss_next_a(LFGauss* unit, int inNumSamples) {
    float* out = ZOUT(0);

    float* dur = ZIN(0);

    float c = ZIN0(1);
    float b = ZIN0(2);
    float loop = ZIN0(3);
    float sr = unit->mRate->mSampleRate;

    // offset phase by b
    double x = unit->mPhase - b;
    float factor = -1.f / (2.f * c * c);

    LOOP1(
        inNumSamples,

        if (x > 1.f) {
            if (loop) {
                x -= 2.f;
            } else {
                DoneAction(ZIN0(4), unit);
            }
        }

        // for a full cycle from -1 to 1 in duration, double the step.
        float step = 2.f / (ZXP(dur) * sr);

        ZXP(out) = exp(x * x * factor);

        x += step;);

    unit->mPhase = x + b;
}


void LFGauss_next_aa(LFGauss* unit, int inNumSamples) {
    float* out = ZOUT(0);

    float* dur = ZIN(0);
    float* c = ZIN(1);

    float b = ZIN0(2);
    float loop = ZIN0(3);
    float sr = unit->mRate->mSampleRate;

    // offset phase by b
    double x = unit->mPhase - b;

    LOOP1(
        inNumSamples,

        if (x > 1.f) {
            if (loop) {
                x -= 2.f;
            } else {
                DoneAction(ZIN0(4), unit);
            }
        }

        // for a full cycle from -1 to 1 in duration, double the step.
        float step = 2.f / (ZXP(dur) * sr);

        float cval = ZXP(c);

        float factor = -1.f / (2.f * cval * cval); ZXP(out) = exp(x * x * factor);

        x += step;);

    unit->mPhase = x + b;
}


void LFGauss_Ctor(LFGauss* unit) {
    if (INRATE(0) == calc_FullRate) {
        if (INRATE(1) == calc_FullRate) {
            SETCALC(LFGauss_next_aa);
        } else {
            SETCALC(LFGauss_next_a);
        }
    } else {
        SETCALC(LFGauss_next_k);
    }
    unit->mPhase = -1.0;

    LFGauss_next_k(unit, 1);
    // reset phase
    unit->mPhase = -1.0;
}


//////////////////////////////////////////////////////////////////////////////////////////////////

// detect if phasor is out-of-bounds, trigger and wrap [0, 1]
static inline float Impulse_testWrapPhase(double prev_inc, double& phase) {
    if (prev_inc < 0.f) { // negative freqs
        if (phase <= 0.f) {
            phase += 1.f;
            if (phase <= 0.f) { // catch large phase jumps
                phase -= sc_ceil(phase);
            }
            return 1.f;
        } else {
            return 0.f;
        }
    } else { // positive freqs
        if (phase >= 1.f) {
            phase -= 1.f;
            if (phase >= 1.f) {
                phase -= sc_floor(phase);
            }
            return 1.f;
        } else {
            return 0.f;
        }
    }
}

void Impulse_next_ii(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;
    double inc = unit->mPhaseIncrement;

    LOOP1(inNumSamples, ZXP(out) = Impulse_testWrapPhase(inc, phase); phase += inc;);

    unit->mPhase = phase;
}

void Impulse_next_ik(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double inc = unit->mPhaseIncrement;

    double prev_off = unit->mPhaseOffset;
    double off = ZIN0(1);
    double phaseSlope = CALCSLOPE(off, prev_off);
    bool phOffChanged = phaseSlope != 0.f;

    LOOP1(
        inNumSamples, ZXP(out) = Impulse_testWrapPhase(inc, phase);

        if (phOffChanged) {
            phase += phaseSlope;
            Impulse_testWrapPhase(inc, phase);
        } phase += inc;);

    unit->mPhase = phase;
    unit->mPhaseOffset = off;
}

void Impulse_next_ki(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double prev_inc = unit->mPhaseIncrement;
    double inc = ZIN0(0) * unit->mFreqMul;
    double incSlope = CALCSLOPE(inc, prev_inc);

    LOOP1(inNumSamples, ZXP(out) = Impulse_testWrapPhase(prev_inc, phase);

          prev_inc += incSlope; phase += prev_inc;);

    unit->mPhase = phase;
    unit->mPhaseIncrement = inc;
}

void Impulse_next_kk(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double prev_inc = unit->mPhaseIncrement;
    double inc = ZIN0(0) * unit->mFreqMul;
    double incSlope = CALCSLOPE(inc, prev_inc);

    double prev_off = unit->mPhaseOffset;
    double off = ZIN0(1);
    double phaseSlope = CALCSLOPE(off, prev_off);
    bool phOffChanged = phaseSlope != 0.f;

    LOOP1(
        inNumSamples, ZXP(out) = Impulse_testWrapPhase(prev_inc, phase);

        if (phOffChanged) {
            phase += phaseSlope;
            Impulse_testWrapPhase(prev_inc, phase);
        } prev_inc += incSlope;
        phase += prev_inc;);

    unit->mPhase = phase;
    unit->mPhaseOffset = off;
    unit->mPhaseIncrement = inc;
}

void Impulse_next_ak(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double inc = unit->mPhaseIncrement;
    float* freqIn = ZIN(0);
    float freqMul = unit->mFreqMul;

    double prev_off = unit->mPhaseOffset;
    double off = ZIN0(1);
    double offSlope = CALCSLOPE(off, prev_off);
    bool offChanged = offSlope != 0.f;

    LOOP1(
        inNumSamples, float z = Impulse_testWrapPhase(inc, phase); if (offChanged) {
            phase += offSlope;
            Impulse_testWrapPhase(inc, phase);
        } inc = ZXP(freqIn) * freqMul;
        ZXP(out) = z; phase += inc;);

    unit->mPhase = phase;
    unit->mPhaseOffset = off;
    unit->mPhaseIncrement = inc;
}

void Impulse_next_aa(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double inc = unit->mPhaseIncrement;
    float* freqin = ZIN(0);
    float freqmul = unit->mFreqMul;

    double prev_off = unit->mPhaseOffset;
    float* offIn = ZIN(1);

    LOOP1(inNumSamples, float z = Impulse_testWrapPhase(inc, phase); float off = ZXP(offIn);
          float offInc = off - prev_off; phase += offInc; Impulse_testWrapPhase(inc, phase);
          inc = ZXP(freqin) * freqmul; ZXP(out) = z;

          phase += inc; prev_off = off;);

    unit->mPhase = phase;
    unit->mPhaseOffset = prev_off;
    unit->mPhaseIncrement = inc;
}

void Impulse_next_ai(Impulse* unit, int inNumSamples) {
    float* out = ZOUT(0);
    double phase = unit->mPhase;

    double inc = unit->mPhaseIncrement;
    float* freqin = ZIN(0);
    float freqmul = unit->mFreqMul;

    LOOP1(inNumSamples, float z = Impulse_testWrapPhase(inc, phase); inc = ZXP(freqin) * freqmul; ZXP(out) = z;
          phase += inc;);

    unit->mPhase = phase;
    unit->mPhaseIncrement = inc;
}

// Impulse is based on a wrapping phasor. When the phase wraps, an impulse is
// output. Phase _increments_ according to its frequency and an additional phase
// _offset_ is applied.
// Order of operations:
// 1. Phase _offset_ is applied to the current phase (if offset has changed).
// 2. Phase is wrapped into range.
// 3. Phase _increment_ is added (according to the frequency).
// 4. Phase is checked for being out of range, in which case a trigger is fired
//    and the phase is again wrapped.
// Therefore, phase increment (freq) triggers an impulse, but not phase offset.
void Impulse_Ctor(Impulse* unit) {
    unit->mPhaseOffset = ZIN0(1);
    unit->mFreqMul = unit->mRate->mSampleDur;
    unit->mPhaseIncrement = ZIN0(0) * unit->mFreqMul;

    double initOff = unit->mPhaseOffset;
    double initInc = unit->mPhaseIncrement;
    double initPhase = sc_wrap(initOff, 0.0, 1.0);

    // Initial phase offset of 0 means output of 1 on first sample.
    // Set phase to wrap point to trigger impulse on first sample
    if (initPhase == 0.0 && initInc >= 0.0) {
        initPhase = 1.0; // positive frequency trigger/wrap position
    }
    unit->mPhase = initPhase;

    UnitCalcFunc func;
    switch (INRATE(0)) {
    case calc_FullRate:
        switch (INRATE(1)) {
        case calc_ScalarRate:
            func = (UnitCalcFunc)Impulse_next_ai;
            break;
        case calc_BufRate:
            func = (UnitCalcFunc)Impulse_next_ak;
            break;
        case calc_FullRate:
            func = (UnitCalcFunc)Impulse_next_aa;
            break;
        }
        break;
    case calc_BufRate:
        if (INRATE(1) == calc_ScalarRate) {
            func = (UnitCalcFunc)Impulse_next_ki;
        } else {
            func = (UnitCalcFunc)Impulse_next_kk;
        }
        break;
    case calc_ScalarRate:
        if (INRATE(1) == calc_ScalarRate) {
            func = (UnitCalcFunc)Impulse_next_ii;
        } else {
            func = (UnitCalcFunc)Impulse_next_ik;
        }
        break;
    }
    unit->mCalcFunc = func;
    func(unit, 1);

    unit->mPhase = initPhase;
    unit->mPhaseOffset = initOff;
    unit->mPhaseIncrement = initInc;
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void VarSaw_next_a(VarSaw* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);
    float nextDuty = ZIN0(2);
    float duty = unit->mDuty;
    float invduty = unit->mInvDuty;
    float inv1duty = unit->mInv1Duty;

    float freqmul = unit->mFreqMul;
    double phase = unit->mPhase;

    LOOP1(
        inNumSamples,
        if (phase >= 1.f) {
            phase -= 1.f;
            duty = unit->mDuty = sc_clip(nextDuty, 0.001, 0.999);
            invduty = unit->mInvDuty = 2.f / duty;
            inv1duty = unit->mInv1Duty = 2.f / (1.f - duty);
        } float z = phase < duty ? phase * invduty : (1.f - phase) * inv1duty;
        phase += ZXP(freq) * freqmul; ZXP(out) = z - 1.f;);

    unit->mPhase = phase;
}

void VarSaw_next_k(VarSaw* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq = ZIN0(0) * unit->mFreqMul;
    float nextDuty = ZIN0(2);
    float duty = unit->mDuty;
    float invduty = unit->mInvDuty;
    float inv1duty = unit->mInv1Duty;

    double phase = unit->mPhase;

    LOOP1(
        inNumSamples,
        if (phase >= 1.f) {
            phase -= 1.f;
            duty = unit->mDuty = sc_clip(nextDuty, 0.001, 0.999);
            invduty = unit->mInvDuty = 2.f / duty;
            inv1duty = unit->mInv1Duty = 2.f / (1.f - duty);
        } float z = phase < duty ? phase * invduty : (1.f - phase) * inv1duty;
        phase += freq; ZXP(out) = z - 1.f;);

    unit->mPhase = phase;
}

void VarSaw_Ctor(VarSaw* unit) {
    if (INRATE(0) == calc_FullRate) {
        SETCALC(VarSaw_next_a);
    } else {
        SETCALC(VarSaw_next_k);
    }

    unit->mFreqMul = unit->mRate->mSampleDur;
    unit->mPhase = ZIN0(1);
    float duty = ZIN0(2);
    duty = unit->mDuty = sc_clip(duty, 0.001, 0.999);
    unit->mInvDuty = 2.f / duty;
    unit->mInv1Duty = 2.f / (1.f - duty);

    ZOUT0(0) = 0.f;
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void SyncSaw_next_aa(SyncSaw* unit, int inNumSamples) {
    float freqmul = unit->mFreqMul;
    float* out = ZOUT(0);
    float* freq1 = ZIN(0);
    float* freq2 = ZIN(1);

    double phase1 = unit->mPhase1;
    double phase2 = unit->mPhase2;

    LOOP1(
        inNumSamples, float freq1x = ZXP(freq1) * freqmul; float freq2x = ZXP(freq2) * freqmul; float z = phase2;
        phase2 += freq2x; if (phase2 >= 1.f) phase2 -= 2.f; phase1 += freq1x; if (phase1 >= 1.f) {
            phase1 -= 2.f;
            phase2 = (phase1 + 1.f) * freq2x / freq1x - 1.f;
        } ZXP(out) = z;);

    unit->mPhase1 = phase1;
    unit->mPhase2 = phase2;
}

void SyncSaw_next_ak(SyncSaw* unit, int inNumSamples) {
    float freqmul = unit->mFreqMul;
    float* out = ZOUT(0);
    float* freq1 = ZIN(0);
    float freq2x = ZIN0(1) * freqmul;

    double phase1 = unit->mPhase1;
    double phase2 = unit->mPhase2;

    LOOP1(
        inNumSamples, float freq1x = ZXP(freq1) * freqmul; float z = phase2; phase2 += freq2x;
        if (phase2 >= 1.f) phase2 -= 2.f; phase1 += freq1x; if (phase1 >= 1.f) {
            phase1 -= 2.f;
            phase2 = (phase1 + 1.f) * freq2x / freq1x - 1.f;
        } ZXP(out) = z;);

    unit->mPhase1 = phase1;
    unit->mPhase2 = phase2;
}

void SyncSaw_next_ka(SyncSaw* unit, int inNumSamples) {
    float freqmul = unit->mFreqMul;
    float* out = ZOUT(0);
    float freq1x = ZIN0(0) * freqmul;
    float* freq2 = ZIN(1);

    double phase1 = unit->mPhase1;
    double phase2 = unit->mPhase2;

    LOOP1(
        inNumSamples, float freq2x = ZXP(freq2) * freqmul; float z = phase2; phase2 += freq2x;
        if (phase2 >= 1.f) phase2 -= 2.f; phase1 += freq1x; if (phase1 >= 1.f) {
            phase1 -= 2.f;
            phase2 = (phase1 + 1.f) * freq2x / freq1x - 1.f;
        } ZXP(out) = z;);

    unit->mPhase1 = phase1;
    unit->mPhase2 = phase2;
}

void SyncSaw_next_kk(SyncSaw* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float freq1x = ZIN0(0) * unit->mFreqMul;
    float freq2x = ZIN0(1) * unit->mFreqMul;
    double phase1 = unit->mPhase1;
    double phase2 = unit->mPhase2;

    LOOP1(
        inNumSamples, float z = phase2; phase2 += freq2x; if (phase2 >= 1.f) phase2 -= 2.f; phase1 += freq1x;
        if (phase1 >= 1.f) {
            phase1 -= 2.f;
            phase2 = (phase1 + 1.f) * freq2x / freq1x - 1.f;
        } ZXP(out) = z;);

    unit->mPhase1 = phase1;
    unit->mPhase2 = phase2;
}

void SyncSaw_Ctor(SyncSaw* unit) {
    if (INRATE(0) == calc_FullRate) {
        if (INRATE(1) == calc_FullRate) {
            SETCALC(SyncSaw_next_aa);
        } else {
            SETCALC(SyncSaw_next_ak);
        }
    } else {
        if (INRATE(1) == calc_FullRate) {
            SETCALC(SyncSaw_next_ka);
        } else {
            SETCALC(SyncSaw_next_kk);
        }
    }
    unit->mFreqMul = 2.0 * unit->mRate->mSampleDur;
    unit->mPhase1 = 0.;
    unit->mPhase2 = 0.;

    SyncSaw_next_kk(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

struct K2A : SIMD_Unit {
    ControlRateInput<0> mLevel;

    K2A(void) {
        mLevel.init(this);
        if (inRate(0) == calc_ScalarRate)
            set_unrolled_calc_function<K2A, &K2A::next_i<unrolled_64>, &K2A::next_i<unrolled>, &K2A::next_i<scalar>>();
        else
            set_unrolled_calc_function<K2A, &K2A::next_k<unrolled_64>, &K2A::next_k<unrolled>, &K2A::next_k<scalar>>();
    }

    template <int type> void next_k(int inNumSamples) {
        if (mLevel.changed(this))
            slope_vec<type>(out(0), mLevel.slope(this), inNumSamples);
        else
            next_i<type>(inNumSamples);
    }

    template <int type> void next_i(int inNumSamples) { set_vec<type>(out(0), mLevel, inNumSamples); }
};

//////////////////////////////////////////////////////////////////////////////////////////////////

void A2K_next(A2K* unit, int inNumSamples) {
    ZOUT0(0) = ZIN0(0); // return first sample in block
}

void A2K_Ctor(A2K* unit) {
    SETCALC(A2K_next);
    A2K_next(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void T2K_next(T2K* unit, int inNumSamples) {
    float out = 0.f, val;
    float* in = ZIN(0);
    int n = unit->mWorld->mBufLength;
    LOOP1(n, val = ZXP(in); if (val > out) out = val;);
    ZOUT0(0) = out;
}

void T2K_Ctor(T2K* unit) {
    SETCALC(T2K_next);
    ZOUT0(0) = ZIN0(0);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

static inline void T2A_write_trigger(T2A* unit, float level) {
    float* out = OUT(0);
    int offset = (int)IN0(1);
    out[offset] = level;
}

void T2A_next(T2A* unit, int inNumSamples) {
    float level = IN0(0);

    ZClear(inNumSamples, ZOUT(0));
    if ((unit->mLevel <= 0.f && level > 0.f))
        T2A_write_trigger(unit, level);

    unit->mLevel = level;
}

#ifdef NOVA_SIMD
FLATTEN void T2A_next_nova(T2A* unit, int inNumSamples) {
    float level = IN0(0);

    nova::zerovec_simd(OUT(0), inNumSamples);
    if ((unit->mLevel <= 0.f && level > 0.f))
        T2A_write_trigger(unit, level);

    unit->mLevel = level;
}

FLATTEN void T2A_next_nova_64(T2A* unit, int inNumSamples) {
    float level = IN0(0);

    nova::zerovec_simd<64>(OUT(0));
    if ((unit->mLevel <= 0.f && level > 0.f))
        T2A_write_trigger(unit, level);

    unit->mLevel = level;
}

#endif

void T2A_Ctor(T2A* unit) {
#ifdef NOVA_SIMD
    if (BUFLENGTH == 64)
        SETCALC(T2A_next_nova_64);
    else if (boost::alignment::is_aligned(BUFLENGTH, 16))
        SETCALC(T2A_next_nova);
    else
#endif
        SETCALC(T2A_next);
    T2A_next(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

struct DC : SIMD_Unit {
    float value;

    DC(void) {
        value = in0(0);
        if (value == 0)
            set_unrolled_calc_function<DC, &DC::next_i<unrolled_64, true>, &DC::next_i<unrolled, true>,
                                       &DC::next_i<scalar, true>>();
        else
            set_unrolled_calc_function<DC, &DC::next_i<unrolled_64, false>, &DC::next_i<unrolled, false>,
                                       &DC::next_i<scalar, false>>();
    }

    template <int type, bool isZero> void next_i(int inNumSamples) {
        if (isZero)
            zero_vec<type>(out(0), inNumSamples);
        else
            set_vec<type>(out(0), value, inNumSamples);
    }
};

//////////////////////////////////////////////////////////////////////////////////////////////////

static inline void Line_next_loop(Line* unit, int& counter, int remain, double& level) {
    float* out = ZOUT(0);
    double slope = unit->mSlope;

    do {
        if (counter == 0) {
            int nsmps = remain;
            remain = 0;
            float endlevel = unit->mEndLevel;
            LOOP(nsmps, ZXP(out) = endlevel;);
        } else {
            int nsmps = sc_min(remain, counter);
            counter -= nsmps;
            remain -= nsmps;
            LOOP(nsmps, ZXP(out) = level; level += slope;);
            if (counter == 0) {
                unit->mDone = true;
                int doneAction = (int)ZIN0(3);
                DoneAction(doneAction, unit);
            }
        }
    } while (remain);
}


void Line_next(Line* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;
    Line_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

#ifdef NOVA_SIMD
FLATTEN void Line_next_nova(Line* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;

    if (counter == 0) {
        nova::setvec_simd(OUT(0), unit->mEndLevel, inNumSamples);
        return;
    }

    if (counter > inNumSamples) {
        double slope = unit->mSlope;
        nova::set_slope_vec_simd(OUT(0), (float)level, (float)slope, inNumSamples);
        unit->mLevel = level + inNumSamples * slope;
        unit->mCounter = counter - inNumSamples;
        return;
    }
    Line_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

FLATTEN void Line_next_nova_64(Line* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;

    if (counter == 0) {
        nova::setvec_simd<64>(OUT(0), unit->mEndLevel);
        return;
    }

    if (counter > inNumSamples) {
        double slope = unit->mSlope;
        nova::set_slope_vec_simd(OUT(0), (float)level, (float)slope, 64);
        unit->mLevel = level + inNumSamples * slope;
        unit->mCounter = counter - inNumSamples;
        return;
    }

    Line_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

#endif

void Line_Ctor(Line* unit) {
#ifdef NOVA_SIMD
    if (BUFLENGTH == 64)
        SETCALC(Line_next_nova);
    else if (boost::alignment::is_aligned(BUFLENGTH, 16))
        SETCALC(Line_next_nova);
    else
#endif
        SETCALC(Line_next);
    double start = ZIN0(0);
    double end = ZIN0(1);
    double dur = ZIN0(2);

    int counter = (int)(dur * unit->mRate->mSampleRate + .5f);
    unit->mCounter = sc_max(1, counter);
    if (counter == 0) {
        unit->mLevel = end;
        unit->mSlope = 0.;
    } else {
        unit->mLevel = start;
        unit->mSlope = (end - start) / unit->mCounter;
        unit->mLevel += unit->mSlope;
    }
    unit->mEndLevel = end;
    ZOUT0(0) = unit->mLevel;
}

//////////////////////////////////////////////////////////////////////////////////////////////////

static inline void Xline_next_loop(XLine* unit, int& counter, int remain, double& level) {
    float* out = ZOUT(0);
    double grow = unit->mGrowth;

    do {
        if (counter == 0) {
            int nsmps = remain;
            remain = 0;
            LOOP(nsmps, ZXP(out) = level;);
        } else {
            int nsmps = sc_min(remain, counter);
            counter -= nsmps;
            remain -= nsmps;
            LOOP(nsmps, ZXP(out) = level; level *= grow;);
            if (counter == 0) {
                level = unit->mEndLevel;
                unit->mDone = true;
                int doneAction = (int)ZIN0(3);
                DoneAction(doneAction, unit);
            }
        }
    } while (remain);
}

void XLine_next(XLine* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;

    Xline_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

#ifdef NOVA_SIMD
FLATTEN void XLine_next_nova(XLine* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;

    if (counter == 0) {
        nova::setvec_simd(OUT(0), (float)level, inNumSamples);
        return;
    }
    if (counter > inNumSamples) {
        double grow = unit->mGrowth;
        nova::set_exp_vec_simd(OUT(0), (float)level, (float)grow, inNumSamples);
        level *= sc_powi(grow, inNumSamples);
        counter -= inNumSamples;
    } else
        Xline_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

FLATTEN void XLine_next_nova_64(XLine* unit, int inNumSamples) {
    double level = unit->mLevel;
    int counter = unit->mCounter;

    if (counter == 0) {
        nova::setvec_simd<64>(OUT(0), (float)level);
        return;
    }
    if (counter > 64) {
        double grow = unit->mGrowth;
        nova::set_exp_vec_simd(OUT(0), (float)level, (float)grow, 64);
        level *= sc_powi(grow, inNumSamples);
        counter -= inNumSamples;
    } else
        Xline_next_loop(unit, counter, inNumSamples, level);
    unit->mCounter = counter;
    unit->mLevel = level;
}

#endif

void XLine_Ctor(XLine* unit) {
#ifdef NOVA_SIMD
    if (BUFLENGTH == 64)
        SETCALC(XLine_next_nova_64);
    else if (boost::alignment::is_aligned(BUFLENGTH, 16))
        SETCALC(XLine_next_nova);
    else
#endif

        SETCALC(XLine_next);
    double start = ZIN0(0);
    double end = ZIN0(1);
    double dur = ZIN0(2);

    int counter = (int)(dur * unit->mRate->mSampleRate + .5f);

    unit->mEndLevel = end;

    if (counter == 0) {
        ZOUT0(0) = end;
        unit->mLevel = end;
        unit->mCounter = 0;
        unit->mGrowth = 0;
    } else {
        ZOUT0(0) = start;
        unit->mCounter = counter;
        unit->mGrowth = pow(end / start, 1.0 / counter);
        unit->mLevel = start * unit->mGrowth;
    }
}

//////////////////////////////////////////////////////////////////////////////////////////////////
/*
void Wrap_next(Wrap* unit, int inNumSamples)
{
    float *out = ZOUT(0);
    float *in   = ZIN(0);
    float lo = unit->m_lo;
    float hi = unit->m_hi;
    float range = unit->m_range;

    LOOP1(inNumSamples,
        ZXP(out) = sc_wrap(ZXP(in), lo, hi, range);
    );
}

void Wrap_Ctor(Wrap* unit)
{

    SETCALC(Wrap_next);
    unit->m_lo = ZIN0(1);
    unit->m_hi = ZIN0(2);

    if (unit->m_lo > unit->m_hi) {
        float temp = unit->m_lo;
        unit->m_lo = unit->m_hi;
        unit->m_hi = temp;
    }
    unit->m_range = unit->m_hi - unit->m_lo;

    Wrap_next(unit, 1);
}
*/


void Wrap_next_kk(Wrap* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_lo = ZIN0(1);
    float next_hi = ZIN0(2);
    float lo = unit->m_lo;
    float lo_slope = CALCSLOPE(next_lo, lo);
    float hi = unit->m_hi;
    float hi_slope = CALCSLOPE(next_hi, hi);
    LOOP1(inNumSamples, float range = hi - lo; ZXP(out) = sc_wrap(ZXP(in), lo, hi, range); lo += lo_slope;
          hi += hi_slope;);
    unit->m_lo = lo;
    unit->m_hi = hi;
}

void Wrap_next_ka(Wrap* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_lo = ZIN0(1);
    float* hi = ZIN(2);
    float lo = unit->m_lo;
    float lo_slope = CALCSLOPE(next_lo, lo);
    LOOP1(inNumSamples, float curhi = ZXP(hi); ZXP(out) = sc_wrap(ZXP(in), lo, curhi, curhi - lo); lo += lo_slope;);
    unit->m_lo = lo;
}

void Wrap_next_ak(Wrap* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float next_hi = ZIN0(2);
    float hi = unit->m_hi;
    float hi_slope = CALCSLOPE(next_hi, hi);

    LOOP1(inNumSamples, float curlo = ZXP(lo); ZXP(out) = sc_wrap(ZXP(in), curlo, hi, hi - curlo); hi += hi_slope;);
    unit->m_hi = hi;
}

void Wrap_next_aa(Wrap* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float* hi = ZIN(2);

    LOOP1(inNumSamples, float curhi = ZXP(hi); float curlo = ZXP(lo);
          ZXP(out) = sc_wrap(ZXP(in), curlo, curhi, curhi - curlo););
}

void Wrap_Ctor(Wrap* unit) {
    if (BUFLENGTH == 1) {
        // _aa? Well, yes - that calc func doesn't interpolate
        // and interpolation is not needed for kr (1 sample/block)
        SETCALC(Wrap_next_aa);
    } else {
        if (INRATE(1) == calc_FullRate) {
            if (INRATE(2) == calc_FullRate)
                SETCALC(Wrap_next_aa);
            else
                SETCALC(Wrap_next_ak);
        } else {
            if (INRATE(2) == calc_FullRate)
                SETCALC(Wrap_next_ka);
            else
                SETCALC(Wrap_next_kk);
        }
    }

    unit->m_lo = ZIN0(1);
    unit->m_hi = ZIN0(2);

    Wrap_next_kk(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////
/*
void Fold_next(Fold* unit, int inNumSamples)
{
    float *out = ZOUT(0);
    float *in   = ZIN(0);
    float lo = unit->m_lo;
    float hi = unit->m_hi;
    float range = unit->m_range;
    float range2 = unit->m_range2;

    LOOP1(inNumSamples,
        ZXP(out) = sc_fold(ZXP(in), lo, hi, range, range2);
    );
}

void Fold_Ctor(Fold* unit)
{

    SETCALC(Fold_next);
    unit->m_lo = ZIN0(1);
    unit->m_hi = ZIN0(2);

    if (unit->m_lo > unit->m_hi) {
        float temp = unit->m_lo;
        unit->m_lo = unit->m_hi;
        unit->m_hi = temp;
    }
    unit->m_range = unit->m_hi - unit->m_lo;
    unit->m_range2 = 2.f * unit->m_range;

    Fold_next(unit, 1);
}
*/
void Fold_next_kk(Fold* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_lo = ZIN0(1);
    float next_hi = ZIN0(2);
    float lo = unit->m_lo;
    float lo_slope = CALCSLOPE(next_lo, lo);
    float hi = unit->m_hi;
    float hi_slope = CALCSLOPE(next_hi, hi);

    LOOP1(inNumSamples, float range = hi - lo; float range2 = range * 2.f;
          ZXP(out) = sc_fold(ZXP(in), lo, hi, range, range2);

          lo += lo_slope; hi += hi_slope;);
    unit->m_lo = lo;
    unit->m_hi = hi;
}

void Fold_next_ka(Fold* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_lo = ZIN0(1);
    float* hi = ZIN(2);
    float lo = unit->m_lo;
    float lo_slope = CALCSLOPE(next_lo, lo);

    LOOP1(inNumSamples, float curhi = ZXP(hi); float range = curhi - lo; float range2 = range * 2.f;
          ZXP(out) = sc_fold(ZXP(in), lo, curhi, range, range2); lo += lo_slope;);
    unit->m_lo = lo;
}

void Fold_next_ak(Fold* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float next_hi = ZIN0(2);
    float hi = unit->m_hi;
    float hi_slope = CALCSLOPE(next_hi, hi);

    LOOP1(inNumSamples, float curlo = ZXP(lo); float range = hi - curlo; float range2 = range * 2.f;
          ZXP(out) = sc_fold(ZXP(in), curlo, hi, range, range2); hi += hi_slope;);
    unit->m_hi = hi;
}

void Fold_next_aa(Fold* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float* hi = ZIN(2);

    LOOP1(inNumSamples, float curhi = ZXP(hi); float curlo = ZXP(lo); float range = curhi - curlo;
          float range2 = range * 2.0; ZXP(out) = sc_fold(ZXP(in), curlo, curhi, range, range2););
}

void Fold_Ctor(Fold* unit) {
    if (BUFLENGTH == 1) {
        // _aa? Well, yes - that calc func doesn't interpolate
        // and interpolation is not needed for kr (1 sample/block)
        SETCALC(Fold_next_aa);
    } else {
        if (INRATE(1) == calc_FullRate) {
            if (INRATE(2) == calc_FullRate)
                SETCALC(Fold_next_aa);
            else
                SETCALC(Fold_next_ak);
        } else {
            if (INRATE(2) == calc_FullRate)
                SETCALC(Fold_next_ka);
            else
                SETCALC(Fold_next_kk);
        }
    }

    unit->m_lo = ZIN0(1);
    unit->m_hi = ZIN0(2);

    Fold_next_kk(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void Clip_next_ii(Clip* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    LOOP1(inNumSamples, ZXP(out) = sc_clip(ZXP(in), lo, hi););
}

void Clip_next_kk(Clip* unit, int inNumSamples) {
    float next_lo = ZIN0(1);
    float next_hi = ZIN0(2);
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    if (lo == next_lo && hi == next_hi) {
        Clip_next_ii(unit, inNumSamples);
        return;
    }

    float* out = ZOUT(0);
    float* in = ZIN(0);
    float lo_slope = CALCSLOPE(next_lo, lo);
    float hi_slope = CALCSLOPE(next_hi, hi);

    LOOP1(inNumSamples, ZXP(out) = sc_clip(ZXP(in), lo, hi); lo += lo_slope; hi += hi_slope;);
    unit->m_lo = lo;
    unit->m_hi = hi;
}

void Clip_next_ka(Clip* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_lo = ZIN0(1);
    float* hi = ZIN(2);
    float lo = unit->m_lo;
    float lo_slope = CALCSLOPE(next_lo, lo);

    LOOP1(inNumSamples, ZXP(out) = sc_clip(ZXP(in), lo, ZXP(hi)); lo += lo_slope;);
    unit->m_lo = lo;
}

void Clip_next_ak(Clip* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float next_hi = ZIN0(2);
    float hi = unit->m_hi;
    float hi_slope = CALCSLOPE(next_hi, hi);

    LOOP1(inNumSamples, ZXP(out) = sc_clip(ZXP(in), ZXP(lo), hi); hi += hi_slope;);
    unit->m_hi = hi;
}

void Clip_next_aa(Clip* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* lo = ZIN(1);
    float* hi = ZIN(2);

    LOOP1(inNumSamples, ZXP(out) = sc_clip(ZXP(in), ZXP(lo), ZXP(hi)););
}

void Clip_next_k(Clip* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float lo = ZIN0(1);
    float hi = ZIN0(2);

    ZXP(out) = sc_clip(ZXP(in), lo, hi);
}

#ifdef NOVA_SIMD
void Clip_next_nova_ii(Clip* unit, int inNumSamples) {
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    nova::clip_vec_simd(OUT(0), IN(0), lo, hi, inNumSamples);
}

void Clip_next_nova_ki(Clip* unit, int inNumSamples) {
    float next_lo = ZIN0(1);
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    if (lo == next_lo) {
        Clip_next_nova_ii(unit, inNumSamples);
        return;
    }

    float lo_slope = CALCSLOPE(next_lo, lo);
    nova::clip_vec_simd(OUT(0), IN(0), slope_argument(lo, lo_slope), hi, inNumSamples);
    unit->m_lo = next_lo;
}

void Clip_next_nova_ik(Clip* unit, int inNumSamples) {
    float next_hi = ZIN0(2);
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    if (hi == next_hi) {
        Clip_next_nova_ii(unit, inNumSamples);
        return;
    }

    float hi_slope = CALCSLOPE(next_hi, hi);
    nova::clip_vec_simd(OUT(0), IN(0), lo, slope_argument(hi, hi_slope), inNumSamples);
    unit->m_hi = next_hi;
}

void Clip_next_nova_kk(Clip* unit, int inNumSamples) {
    float next_lo = ZIN0(1);
    float next_hi = ZIN0(2);
    float lo = unit->m_lo;
    float hi = unit->m_hi;

    if (lo == next_lo && hi == next_hi) {
        Clip_next_nova_ii(unit, inNumSamples);
        return;
    }

    if (lo == next_lo) {
        Clip_next_nova_ik(unit, inNumSamples);
        return;
    }

    if (hi == next_hi) {
        Clip_next_nova_ki(unit, inNumSamples);
        return;
    }

    float lo_slope = CALCSLOPE(next_lo, lo);
    float hi_slope = CALCSLOPE(next_hi, hi);

    nova::clip_vec_simd(OUT(0), IN(0), slope_argument(lo, lo_slope), slope_argument(hi, hi_slope), inNumSamples);
    unit->m_lo = next_lo;
    unit->m_hi = next_hi;
}

void Clip_next_nova_ai(Clip* unit, int inNumSamples) {
    float hi = unit->m_hi;
    nova::clip_vec_simd(OUT(0), IN(0), IN(1), hi, inNumSamples);
}

void Clip_next_nova_ak(Clip* unit, int inNumSamples) {
    float next_hi = ZIN0(2);
    float hi = unit->m_hi;

    if (hi == next_hi) {
        Clip_next_nova_ai(unit, inNumSamples);
        return;
    }

    float hi_slope = CALCSLOPE(next_hi, hi);

    nova::clip_vec_simd(OUT(0), IN(0), IN(1), slope_argument(hi, hi_slope), inNumSamples);
    unit->m_hi = next_hi;
}

void Clip_next_nova_ia(Clip* unit, int inNumSamples) {
    float lo = unit->m_lo;
    nova::clip_vec_simd(OUT(0), IN(0), lo, IN(2), inNumSamples);
}

void Clip_next_nova_ka(Clip* unit, int inNumSamples) {
    float next_lo = ZIN0(1);
    float lo = unit->m_lo;

    if (lo == next_lo) {
        Clip_next_nova_ia(unit, inNumSamples);
        return;
    }

    float lo_slope = CALCSLOPE(next_lo, lo);
    nova::clip_vec_simd(OUT(0), IN(0), slope_argument(lo, lo_slope), IN(2), inNumSamples);
    unit->m_lo = next_lo;
}


void Clip_next_nova_aa(Clip* unit, int inNumSamples) { nova::clip_vec_simd(OUT(0), IN(0), IN(1), IN(2), inNumSamples); }

#endif

typedef void (*ClipCalcFunc)(Clip*, int);

static ClipCalcFunc Clip_SelectCalc(Clip* unit) {
    if (BUFLENGTH == 1)
        return Clip_next_k;

    int loRate = INRATE(1);
    int hiRate = INRATE(2);

#ifdef NOVA_SIMD
    if (boost::alignment::is_aligned(BUFLENGTH, 16)) {
        switch (loRate) {
        case calc_FullRate:
            switch (hiRate) {
            case calc_FullRate:
                return Clip_next_nova_aa;
            case calc_BufRate:
                return Clip_next_nova_ak;
            case calc_ScalarRate:
                return Clip_next_nova_ai;
            }
            break;

        case calc_BufRate:
            switch (hiRate) {
            case calc_FullRate:
                return Clip_next_nova_ka;
            case calc_BufRate:
                return Clip_next_nova_kk;
            case calc_ScalarRate:
                return Clip_next_nova_ki;
            }
            break;

        case calc_ScalarRate:
            switch (hiRate) {
            case calc_FullRate:
                return Clip_next_nova_ia;
            case calc_BufRate:
                return Clip_next_nova_ik;
            case calc_ScalarRate:
                return Clip_next_nova_ii;
            }
            break;
        }
    }
#endif

    if (loRate == calc_FullRate && hiRate == calc_FullRate)
        return Clip_next_aa;

    if (loRate == calc_ScalarRate && hiRate == calc_ScalarRate)
        return Clip_next_ii;

    if (loRate == calc_FullRate && hiRate != calc_FullRate)
        return Clip_next_ak;

    if (loRate != calc_FullRate && hiRate == calc_FullRate)
        return Clip_next_ak;

    return Clip_next_kk;
}

void Clip_Ctor(Clip* unit) {
    ClipCalcFunc fn = Clip_SelectCalc(unit);
    unit->mCalcFunc = (UnitCalcFunc)fn;

    unit->m_lo = ZIN0(1);
    unit->m_hi = ZIN0(2);

    Clip_next_ii(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void Unwrap_next(Unwrap* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float range = unit->m_range;
    float half = unit->m_half;
    float prev = unit->m_prev;
    float offset = unit->m_offset;

    LOOP1(
        inNumSamples, float zin = ZXP(in); float diff = zin - prev; if (fabs(diff) > half) {
            if (zin < prev)
                offset += range;
            else
                offset -= range;
        } ZXP(out) = zin + offset;
        prev = zin;);
    unit->m_prev = prev;
    unit->m_offset = offset;
}

void Unwrap_Ctor(Unwrap* unit) {
    SETCALC(Unwrap_next);
    float in = ZIN0(0);
    float lo = ZIN0(1);
    float hi = ZIN0(2);

    if (lo > hi) {
        float temp = lo;
        lo = hi;
        hi = temp;
    }
    unit->m_range = fabs(hi - lo);
    unit->m_half = unit->m_range * 0.5f;

    if (in < lo || in >= hi)
        unit->m_offset = floor((lo - in) / unit->m_range) * unit->m_range;
    else
        unit->m_offset = 0.f;

    Unwrap_next(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void ModDif_next_kk(ModDif* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_dif = ZIN0(1);
    float next_mod = ZIN0(2);
    float dif = unit->m_dif;
    float dif_slope = CALCSLOPE(next_dif, dif);
    float mod = unit->m_mod;
    float mod_slope = CALCSLOPE(next_mod, mod);

    LOOP1(inNumSamples, float inval = ZXP(in); float diff = std::fmod(std::abs(inval - dif), mod);
          float modhalf = mod * 0.5; ZXP(out) = modhalf - std::fabs(diff - modhalf); dif += dif_slope;
          mod += mod_slope;);
    unit->m_dif = dif;
    unit->m_mod = mod;
}

void ModDif_next_ka(ModDif* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float next_dif = ZIN0(1);
    float* mod = ZIN(2);
    float dif = unit->m_dif;
    float dif_slope = CALCSLOPE(next_dif, dif);

    LOOP1(inNumSamples, float inval = ZXP(in); float curmod = ZXP(mod);
          float diff = std::fmod(std::abs(inval - dif), curmod); float modhalf = curmod * 0.5;
          ZXP(out) = modhalf - std::abs(diff - modhalf); dif += dif_slope;);
    unit->m_dif = dif;
}

void ModDif_next_ak(ModDif* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* dif = ZIN(1);
    float next_mod = ZIN0(2);
    float mod = unit->m_mod;
    float mod_slope = CALCSLOPE(next_mod, mod);

    LOOP1(inNumSamples, float inval = ZXP(in); float diff = std::fmod(std::abs(inval - ZXP(dif)), mod);
          float modhalf = mod * 0.5; ZXP(out) = modhalf - std::abs(diff - modhalf); mod += mod_slope;);
    unit->m_mod = mod;
}

void ModDif_next_aa(ModDif* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* dif = ZIN(1);
    float* mod = ZIN(2);

    LOOP1(inNumSamples, float inval = ZXP(in); float curmod = ZXP(mod);
          float diff = std::fmod(std::abs(inval - ZXP(dif)), curmod); float modhalf = curmod * 0.5;
          ZXP(out) = modhalf - std::abs(diff - modhalf););
}

void ModDif_Ctor(ModDif* unit) {
    if (BUFLENGTH == 1) {
        // _aa? Well, yes - that calc func doesn't interpolate
        // and interpolation is not needed for kr (1 sample/block)
        SETCALC(ModDif_next_aa);
    } else {
        if (INRATE(1) == calc_FullRate) {
            if (INRATE(2) == calc_FullRate)
                SETCALC(ModDif_next_aa);
            else
                SETCALC(ModDif_next_ak);
        } else {
            if (INRATE(2) == calc_FullRate)
                SETCALC(ModDif_next_ka);
            else
                SETCALC(ModDif_next_kk);
        }
    }

    unit->m_dif = ZIN0(1);
    unit->m_mod = ZIN0(2);

    ModDif_next_kk(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void AmpComp_next(AmpComp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);
    float rootmul = unit->m_rootmul;
    float xb = unit->m_exponent;

    LOOP1(inNumSamples, float xa = ZXP(freq); ZXP(out) = xa >= 0.f ? pow(xa, xb) * rootmul : -pow(-xa, xb) * rootmul;);
}

void AmpComp_next_kk(AmpComp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);
    float root = ZIN0(1);
    float xb = ZIN0(2);

    LOOP1(inNumSamples, float xa = root / ZXP(freq); ZXP(out) = xa >= 0.f ? pow(xa, xb) : -pow(-xa, xb););
}

void AmpComp_Ctor(AmpComp* unit) {
    if (INRATE(1) != calc_ScalarRate || INRATE(2) != calc_ScalarRate) {
        SETCALC(AmpComp_next_kk);
    } else {
        float exp = ZIN0(2);
        unit->m_rootmul = pow(ZIN0(1), exp);
        unit->m_exponent = -1.f * exp;
        SETCALC(AmpComp_next);
    }
    AmpComp_next(unit, 1);
}


//////////////////////////////////////////////////////////////////////////////////////////////////

const double AMPCOMP_K = 3.5041384 * 10e15;
const double AMPCOMP_C1 = 20.598997 * 20.598997;
const double AMPCOMP_C2 = 107.65265 * 107.65265;
const double AMPCOMP_C3 = 737.86223 * 737.86223;
const double AMPCOMP_C4 = 12194.217 * 12194.217;
const double AMPCOMP_MINLEVEL = -0.1575371167435;

double AmpCompA_calcLevel(double freq) {
    double r = freq * freq;
    double level = (AMPCOMP_K * r * r * r * r);
    double n1 = AMPCOMP_C1 + r;
    double n2 = AMPCOMP_C4 + r;
    level = level / (n1 * n1 * (AMPCOMP_C2 + r) * (AMPCOMP_C3 + r) * n2 * n2);
    level = 1. - sqrt(level);
    return level;
}

void AmpCompA_next(AmpCompA* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* freq = ZIN(0);

    double scale = unit->m_scale;
    double offset = unit->m_offset;

    LOOP1(inNumSamples, ZXP(out) = AmpCompA_calcLevel(ZXP(freq)) * scale + offset;);
}

void AmpCompA_Ctor(AmpCompA* unit) {
    double rootFreq = ZIN0(1);
    double rootLevel = AmpCompA_calcLevel(rootFreq);
    float minLevel = ZIN0(2);
    unit->m_scale = (ZIN0(3) - minLevel) / (rootLevel - AMPCOMP_MINLEVEL);
    unit->m_offset = minLevel - unit->m_scale * AMPCOMP_MINLEVEL;

    SETCALC(AmpCompA_next);
    AmpCompA_next(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void InRange_next(InRange* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float lo = ZIN0(1);
    float hi = ZIN0(2);

    LOOP1(inNumSamples, float zin = ZXP(in); ZXP(out) = zin >= lo && zin <= hi ? 1.f : 0.f;);
}

void InRange_Ctor(InRange* unit) {
    SETCALC(InRange_next);
    InRange_next(unit, 1);
}

//////////////////////////////////////////////////////////////////////////////////////////////////

void InRect_next(InRect* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* inx = ZIN(0);
    float* iny = ZIN(1);
    float left = ZIN0(2);
    float top = ZIN0(3);
    float right = ZIN0(4);
    float bottom = ZIN0(5);

    LOOP1(inNumSamples, float x = ZXP(inx); float y = ZXP(iny);
          ZXP(out) = x >= left && x <= right && y >= top && y <= bottom ? 1.f : 0.f;);
}

void InRect_Ctor(InRect* unit) {
    SETCALC(InRect_next);
    InRect_next(unit, 1);
}

////////////////////////////////////////////////////////////////////////////////////////////////////////


void LinExp_next(LinExp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float dstlo = unit->m_dstlo;

    float dstratio = unit->m_dstratio;
    float rsrcrange = unit->m_rsrcrange;
    float rrminuslo = unit->m_rrminuslo;

    LOOP1(inNumSamples, ZXP(out) = dstlo * pow(dstratio, ZXP(in) * rsrcrange + rrminuslo););
}

#ifdef NOVA_SIMD
static inline void LinExp_next_nova_loop(float* out, const float* in, int inNumSamples, nova::vec<float> dstlo,
                                         nova::vec<float> dstratio, nova::vec<float> rsrcrange,
                                         nova::vec<float> rrminuslo) {
    const int vecSize = nova::vec<float>::size;
    int unroll = inNumSamples / (2 * vecSize);

    do {
        nova::vec<float> val0, val1;
        val0.load_aligned(in);
        val1.load_aligned(in + vecSize);

        val0 = dstlo * pow(dstratio, val0 * rsrcrange + rrminuslo);
        val1 = dstlo * pow(dstratio, val1 * rsrcrange + rrminuslo);

        val0.store_aligned(out);
        val1.store_aligned(out + vecSize);

        in += 2 * vecSize;
        out += 2 * vecSize;
    } while (--unroll);
}

FLATTEN static void LinExp_next_nova(LinExp* unit, int inNumSamples) {
    float* out = OUT(0);
    float* in = IN(0);

    LinExp_next_nova_loop(out, in, inNumSamples, unit->m_dstlo, unit->m_dstratio, unit->m_rsrcrange, unit->m_rrminuslo);
}

FLATTEN static void LinExp_next_nova_kk(LinExp* unit, int inNumSamples) {
    float* out = OUT(0);
    float* in = IN(0);

    float srclo = ZIN0(1);
    float srchi = ZIN0(2);
    float dstlo = ZIN0(3);
    float dsthi = ZIN0(4);
    float dstratio = dsthi / dstlo;
    float rsrcrange = sc_reciprocal(srchi - srclo);
    float rrminuslo = rsrcrange * -srclo;

    LinExp_next_nova_loop(out, in, inNumSamples, dstlo, dstratio, rsrcrange, rrminuslo);
}

#endif

void LinExp_next_kk(LinExp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float srclo = ZIN0(1);
    float srchi = ZIN0(2);
    float dstlo = ZIN0(3);
    float dsthi = ZIN0(4);
    float dstratio = dsthi * sc_reciprocal(dstlo);
    float rsrcrange = sc_reciprocal(srchi - srclo);
    float rrminuslo = rsrcrange * -srclo;

    LOOP1(inNumSamples, ZXP(out) = dstlo * pow(dstratio, ZXP(in) * rsrcrange + rrminuslo););
}

void LinExp_next_aa(LinExp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* srclo = ZIN(1);
    float* srchi = ZIN(2);
    float* dstlo = ZIN(3);
    float* dsthi = ZIN(4);


    LOOP1(inNumSamples, float zdsthi = ZXP(dsthi); float zdstlo = ZXP(dstlo); float zsrchi = ZXP(srchi);
          float zsrclo = ZXP(srclo); float dstratio = zdsthi / zdstlo; float rsrcrange = sc_reciprocal(zsrchi - zsrclo);
          float rrminuslo = rsrcrange * -zsrclo; ZXP(out) = zdstlo * pow(dstratio, ZXP(in) * rsrcrange + rrminuslo););
}

void LinExp_next_ak(LinExp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float* srclo = ZIN(1);
    float* srchi = ZIN(2);
    float dstlo = ZIN0(3);
    float dsthi = ZIN0(4);
    float dstratio = dsthi / dstlo;

    LOOP1(inNumSamples, float zsrchi = ZXP(srchi); float zsrclo = ZXP(srclo);

          float rsrcrange = sc_reciprocal(zsrchi - zsrclo); float rrminuslo = rsrcrange * -zsrclo;
          ZXP(out) = dstlo * pow(dstratio, ZXP(in) * rsrcrange + rrminuslo););
}

void LinExp_next_ka(LinExp* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* in = ZIN(0);
    float srclo = ZIN0(1);
    float srchi = ZIN0(2);
    float* dstlo = ZIN(3);
    float* dsthi = ZIN(4);
    float rsrcrange = sc_reciprocal(srchi - srclo);
    float rrminuslo = rsrcrange * -srclo;

    LOOP1(inNumSamples, float zdsthi = ZXP(dsthi); float zdstlo = ZXP(dstlo); float dstratio = zdsthi / zdstlo;
          ZXP(out) = zdstlo * pow(dstratio, ZXP(in) * rsrcrange + rrminuslo););
}


static void LinExp_SetCalc(LinExp* unit) {
    if (INRATE(1) == calc_FullRate || INRATE(2) == calc_FullRate) {
        if (INRATE(3) == calc_FullRate || INRATE(4) == calc_FullRate) {
            SETCALC(LinExp_next_aa);
            return;
        } else {
            SETCALC(LinExp_next_ak);
            return;
        }
    } else {
        if (INRATE(3) == calc_FullRate || INRATE(4) == calc_FullRate) {
            SETCALC(LinExp_next_ka);
            return;
        }
    }

    bool allScalar = true;
    for (int i = 1; i < 5; i++) {
        if (INRATE(i) != calc_ScalarRate) {
            allScalar = false;
            break;
        }
    };

#ifdef NOVA_SIMD
    if ((BUFLENGTH % (2 * nova::vec<float>::size)) == 0)
        if (allScalar)
            SETCALC(LinExp_next_nova);
        else
            SETCALC(LinExp_next_nova_kk);
    else
#endif
        if (allScalar)
        SETCALC(LinExp_next);
    else
        SETCALC(LinExp_next_kk);

    if (!allScalar)
        return;

    float srclo = ZIN0(1);
    float srchi = ZIN0(2);
    float dstlo = ZIN0(3);
    float dsthi = ZIN0(4);
    unit->m_dstlo = dstlo;
    unit->m_dstratio = dsthi / dstlo;
    unit->m_rsrcrange = sc_reciprocal(srchi - srclo);
    unit->m_rrminuslo = unit->m_rsrcrange * -srclo;
}

void LinExp_Ctor(LinExp* unit) {
    LinExp_SetCalc(unit);
    float srclo = ZIN0(1);
    float srchi = ZIN0(2);
    float dstlo = ZIN0(3);
    float dsthi = ZIN0(4);
    unit->m_dstlo = dstlo;
    unit->m_dstratio = dsthi / dstlo;
    unit->m_rsrcrange = 1. / (srchi - srclo);
    unit->m_rrminuslo = unit->m_rsrcrange * -srclo;
    LinExp_next(unit, 1);
}

////////////////////////////////////////////////////////////////////////////////////////////////////////


enum {
    kEnvGen_gate,
    kEnvGen_levelScale,
    kEnvGen_levelBias,
    kEnvGen_timeScale,
    kEnvGen_doneAction,
    kEnvGen_initLevel,
    kEnvGen_numStages,
    kEnvGen_releaseNode,
    kEnvGen_loopNode,
    // 'kEnvGen_nodeOffset' must always be last
    // if you need to add an arg, put it before this one
    kEnvGen_nodeOffset
};


enum {
    shape_Step,
    shape_Linear,
    shape_Exponential,
    shape_Sine,
    shape_Welch,
    shape_Curve,
    shape_Squared,
    shape_Cubed,
    shape_Hold,
    shape_Sustain = 9999
};


#ifdef NOVA_SIMD
void EnvGen_next_ak_nova(EnvGen* unit, int inNumSamples);
#endif

#define ENVGEN_NOT_STARTED 1000000000

void EnvGen_Ctor(EnvGen* unit) {
    // Print("EnvGen_Ctor A\n");
    if (unit->mCalcRate == calc_FullRate) {
        if (INRATE(0) == calc_FullRate) {
            SETCALC(EnvGen_next_aa);
        } else {
#ifdef NOVA_SIMD
            if (boost::alignment::is_aligned(BUFLENGTH, 16))
                SETCALC(EnvGen_next_ak_nova);
            else
#endif
                SETCALC(EnvGen_next_ak);
        }
    } else {
        SETCALC(EnvGen_next_k);
    }

    // gate = 1.0, levelScale = 1.0, levelBias = 0.0, timeScale
    // level0, numstages, releaseNode, loopNode,
    // [level, dur, shape, curve]

    unit->m_endLevel = unit->m_level = ZIN0(kEnvGen_initLevel) * ZIN0(kEnvGen_levelScale) + ZIN0(kEnvGen_levelBias);
    unit->m_counter = 0;
    unit->m_stage = ENVGEN_NOT_STARTED;
    unit->m_shape = shape_Hold;
    unit->m_prevGate = 0.f;
    unit->m_released = false;
    unit->m_releaseNode = (int)ZIN0(kEnvGen_releaseNode);

    float** envPtr = unit->mInBuf + kEnvGen_nodeOffset;
    const int initialShape = (int32)*envPtr[2];
    if (initialShape == shape_Hold)
        unit->m_level = *envPtr[0]; // we start at the end level;

    EnvGen_next_k(unit, 1);
}

// called by nextSegment and check_gate:
// - counter: num samples to next segment
// - level: current envelope value
// - dur: if supplied and >= 0, stretch segment to last dur seconds (used in forced release)
static bool EnvGen_initSegment(EnvGen* unit, int& counter, double& level, double dur = -1) {
    // Print("stage %d\n", unit->m_stage);
    // Print("initSegment\n");
    // out = unit->m_level;
    int stageOffset = (unit->m_stage << 2) + kEnvGen_nodeOffset;

    if (stageOffset + 4 > unit->mNumInputs) {
        // oops.
        Print("envelope went past end of inputs.\n");
        ClearUnitOutputs(unit, 1);
        NodeEnd(&unit->mParent->mNode);
        return false;
    }

    float previousEndLevel = unit->m_endLevel;
    if (unit->m_shape == shape_Hold)
        level = previousEndLevel;
    float** envPtr = unit->mInBuf + stageOffset;
    double endLevel = *envPtr[0] * ZIN0(kEnvGen_levelScale) + ZIN0(kEnvGen_levelBias); // scale levels
    if (dur < 0)
        dur = *envPtr[1] * ZIN0(kEnvGen_timeScale);
    unit->m_shape = (int32)*envPtr[2];
    double curve = *envPtr[3];
    unit->m_endLevel = endLevel;

    counter = (int32)(dur * SAMPLERATE);
    counter = sc_max(1, counter);
    // Print("counter %d stageOffset %d   level %g   endLevel %g   dur %g   shape %d   curve %g\n", counter,
    // stageOffset, level, endLevel, dur, unit->m_shape, curve); Print("SAMPLERATE %g\n", SAMPLERATE);
    if (counter == 1)
        unit->m_shape = 1; // shape_Linear
    // Print("new counter = %d  shape = %d\n", counter, unit->m_shape);
    switch (unit->m_shape) {
    case shape_Step: {
        level = endLevel;
    } break;
    case shape_Hold: {
        level = previousEndLevel;
    } break;
    case shape_Linear: {
        unit->m_grow = (endLevel - level) / counter;
        // Print("grow %g\n", unit->m_grow);
    } break;
    case shape_Exponential: {
        unit->m_grow = pow(endLevel / level, 1.0 / counter);
    } break;
    case shape_Sine: {
        double w = pi / counter;

        unit->m_a2 = (endLevel + level) * 0.5;
        unit->m_b1 = 2. * cos(w);
        unit->m_y1 = (endLevel - level) * 0.5;
        unit->m_y2 = unit->m_y1 * sin(pi * 0.5 - w);
        level = unit->m_a2 - unit->m_y1;
    } break;
    case shape_Welch: {
        double w = (pi * 0.5) / counter;

        unit->m_b1 = 2. * cos(w);

        if (endLevel >= level) {
            unit->m_a2 = level;
            unit->m_y1 = 0.;
            unit->m_y2 = -sin(w) * (endLevel - level);
        } else {
            unit->m_a2 = endLevel;
            unit->m_y1 = level - endLevel;
            unit->m_y2 = cos(w) * (level - endLevel);
        }
        level = unit->m_a2 + unit->m_y1;
    } break;
    case shape_Curve: {
        if (fabs(curve) < 0.001) {
            unit->m_shape = 1; // shape_Linear
            unit->m_grow = (endLevel - level) / counter;
        } else {
            double a1 = (endLevel - level) / (1.0 - exp(curve));
            unit->m_a2 = level + a1;
            unit->m_b1 = a1;
            unit->m_grow = exp(curve / counter);
        }
    } break;
    case shape_Squared: {
        unit->m_y1 = sqrt(level);
        unit->m_y2 = sqrt(endLevel);
        unit->m_grow = (unit->m_y2 - unit->m_y1) / counter;
    } break;
    case shape_Cubed: {
        unit->m_y1 = pow(level, 1.0 / 3.0); // 0.33333333);
        unit->m_y2 = pow(endLevel, 1.0 / 3.0);
        unit->m_grow = (unit->m_y2 - unit->m_y1) / counter;
    } break;
    };
    return true;
}

static bool check_gate(EnvGen* unit, float prevGate, float gate, int& counter, double level, int counterOffset = 0) {
    if (prevGate <= 0.f && gate > 0.f) {
        unit->m_stage = -1;
        unit->m_released = false;
        unit->mDone = false;
        counter = counterOffset;
        return false;
    } else if (gate <= -1.f && prevGate > -1.f) {
        // forced release: jump to last segment overriding its duration
        double dur = -gate - 1.f;
        counter = (int32)(dur * SAMPLERATE);
        counter = sc_max(1, counter) + counterOffset;
        unit->m_stage = static_cast<int>(ZIN0(kEnvGen_numStages) - 1);
        unit->m_released = true;
        EnvGen_initSegment(unit, counter, level, dur);
        return false;
    } else if (prevGate > 0.f && gate <= 0.f && unit->m_releaseNode >= 0 && !unit->m_released) {
        counter = counterOffset;
        unit->m_stage = unit->m_releaseNode - 1;
        unit->m_released = true;
        return false;
    }
    return true;
}

static inline bool check_gate_ar(EnvGen* unit, int i, float& prevGate, float*& gatein, int& nsmps, int& counter,
                                 double level) {
    const float gate = ZXP(gatein);
    const bool result = check_gate(unit, prevGate, gate, counter, level, i);
    if (!result) {
        --gatein;
        nsmps = i;
    }
    prevGate = gate;
    return result;
}

static inline bool EnvGen_nextSegment(EnvGen* unit, int& counter, double& level) {
    // Print("stage %d rel %d\n", unit->m_stage, (int)ZIN0(kEnvGen_releaseNode));
    int numstages = (int)ZIN0(kEnvGen_numStages);

    // Print("stage %d   numstages %d\n", unit->m_stage, numstages);
    if (unit->m_stage + 1 >= numstages) { // num stages
        // Print("stage+1 > num stages\n");
        counter = INT_MAX;
        unit->m_shape = 0;
        level = unit->m_endLevel;
        unit->mDone = true;
        int doneAction = (int)ZIN0(kEnvGen_doneAction);
        DoneAction(doneAction, unit);
    } else if (unit->m_stage == ENVGEN_NOT_STARTED) {
        counter = INT_MAX;
        return true;
    } else if (unit->m_stage + 1 == (int)ZIN0(kEnvGen_releaseNode) && !unit->m_released) { // sustain stage
        int loopNode = (int)ZIN0(kEnvGen_loopNode);
        if (loopNode >= 0 && loopNode < numstages) {
            unit->m_stage = loopNode;
            return EnvGen_initSegment(unit, counter, level);
        } else {
            counter = INT_MAX;
            unit->m_shape = shape_Sustain;
            level = unit->m_endLevel;
        }
        // Print("sustain\n");
    } else {
        unit->m_stage++;
        return EnvGen_initSegment(unit, counter, level);
    }

    return true;
}

template <bool CheckGateOnSustain, typename GateCheck>
static inline void EnvGen_perform(EnvGen* unit, float*& out, double& level, int& nsmps, GateCheck const& gateCheck) {
    switch (unit->m_shape) {
    case shape_Step:
    case shape_Hold: {
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
        }
    } break;
    case shape_Linear: {
        double grow = unit->m_grow;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            level += grow;
        }
    } break;
    case shape_Exponential: {
        double grow = unit->m_grow;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            level *= grow;
        }
    } break;
    case shape_Sine: {
        double a2 = unit->m_a2;
        double b1 = unit->m_b1;
        double y2 = unit->m_y2;
        double y1 = unit->m_y1;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            double y0 = b1 * y1 - y2;
            level = a2 - y0;
            y2 = y1;
            y1 = y0;
        }
        unit->m_y1 = y1;
        unit->m_y2 = y2;
    } break;
    case shape_Welch: {
        double a2 = unit->m_a2;
        double b1 = unit->m_b1;
        double y2 = unit->m_y2;
        double y1 = unit->m_y1;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            double y0 = b1 * y1 - y2;
            level = a2 + y0;
            y2 = y1;
            y1 = y0;
        }
        unit->m_y1 = y1;
        unit->m_y2 = y2;
    } break;
    case shape_Curve: {
        double a2 = unit->m_a2;
        double b1 = unit->m_b1;
        double grow = unit->m_grow;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            b1 *= grow;
            level = a2 - b1;
        }
        unit->m_b1 = b1;
    } break;
    case shape_Squared: {
        double grow = unit->m_grow;
        double y1 = unit->m_y1;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            y1 += grow;
            level = y1 * y1;
        }
        unit->m_y1 = y1;
    } break;
    case shape_Cubed: {
        double grow = unit->m_grow;
        double y1 = unit->m_y1;
        for (int i = 0; i < nsmps; ++i) {
            if (!gateCheck(i))
                break;
            ZXP(out) = level;
            y1 += grow;
            y1 = sc_max(y1, 0);
            level = y1 * y1 * y1;
        }
        unit->m_y1 = y1;
    } break;
    case shape_Sustain: {
        for (int i = 0; i < nsmps; ++i) {
            if (CheckGateOnSustain) {
                if (gateCheck(i))
                    ZXP(out) = level;
            } else
                ZXP(out) = level;
        }
    } break;
    }
}

static inline void EnvGen_perform(EnvGen* unit, float*& out, double& level, int nsmps) {
    EnvGen_perform<false>(unit, out, level, nsmps, [](int i) { return true; });
}


void EnvGen_next_k(EnvGen* unit, int inNumSamples) {
    float gate = ZIN0(kEnvGen_gate);
    // Print("->EnvGen_next_k gate %g\n", gate);
    int counter = unit->m_counter;
    double level = unit->m_level;

    check_gate(unit, unit->m_prevGate, gate, counter, level);
    unit->m_prevGate = gate;

    // gate = 1.0, levelScale = 1.0, levelBias = 0.0, timeScale
    // level0, numstages, releaseNode, loopNode,
    // [level, dur, shape, curve]

    if (counter <= 0) {
        bool success = EnvGen_nextSegment(unit, counter, level);
        if (!success)
            return;
    }

    float* out = ZOUT(0);
    EnvGen_perform(unit, out, level, 1);

    // Print("x %d %d %d %g\n", unit->m_stage, counter, unit->m_shape, *out);
    unit->m_level = level;
    unit->m_counter = counter - 1;
}

void EnvGen_next_ak(EnvGen* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float gate = ZIN0(kEnvGen_gate);
    int counter = unit->m_counter;
    double level = unit->m_level;

    check_gate(unit, unit->m_prevGate, gate, counter, level);
    unit->m_prevGate = gate;

    int remain = inNumSamples;
    while (remain) {
        if (counter <= 0) {
            bool success = EnvGen_nextSegment(unit, counter, level);
            if (!success)
                return;
        }

        int nsmps = sc_min(remain, counter);
        EnvGen_perform(unit, out, level, nsmps);

        remain -= nsmps;
        counter -= nsmps;
    }
    // Print("x %d %d %d %g\n", unit->m_stage, counter, unit->m_shape, ZOUT0(0));
    unit->m_level = level;
    unit->m_counter = counter;
}

#ifdef NOVA_SIMD
FLATTEN void EnvGen_next_ak_nova(EnvGen* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float gate = ZIN0(kEnvGen_gate);
    int counter = unit->m_counter;
    double level = unit->m_level;

    check_gate(unit, unit->m_prevGate, gate, counter, level);
    unit->m_prevGate = gate;

    int remain = inNumSamples;
    if (counter > inNumSamples) {
        switch (unit->m_shape) {
        case shape_Step:
        case shape_Hold:
        case shape_Sustain:
            nova::setvec_simd(OUT(0), (float)level, inNumSamples);
            remain = 0;
            counter -= inNumSamples;
            break;
        case shape_Linear: {
            double slope = unit->m_grow;
            nova::set_slope_vec_simd(OUT(0), (float)level, (float)slope, inNumSamples);
            level += inNumSamples * slope;
            remain = 0;
            counter -= inNumSamples;
        } break;
        case shape_Exponential: {
            double grow = unit->m_grow;
            nova::set_exp_vec_simd(OUT(0), (float)level, (float)grow, inNumSamples);
            level *= sc_powi(grow, inNumSamples);
            remain = 0;
            counter -= inNumSamples;
        } break;
        }
    }

    while (remain) {
        if (counter <= 0) {
            bool success = EnvGen_nextSegment(unit, counter, level);
            if (!success)
                return;
        }

        int nsmps = sc_min(remain, counter);
        EnvGen_perform(unit, out, level, nsmps);

        remain -= nsmps;
        counter -= nsmps;
    }
    // Print("x %d %d %d %g\n", unit->m_stage, counter, unit->m_shape, ZOUT0(0));
    unit->m_level = level;
    unit->m_counter = counter;
}
#endif


void EnvGen_next_aa(EnvGen* unit, int inNumSamples) {
    float* out = ZOUT(0);
    float* gatein = ZIN(kEnvGen_gate);
    int counter = unit->m_counter;
    double level = unit->m_level;
    float gate = unit->m_prevGate;
    int remain = inNumSamples;
    while (remain) {
        if (counter <= 0) {
            bool success = EnvGen_nextSegment(unit, counter, level);
            if (!success)
                return;
        }

        int nsmps = sc_min(remain, counter);
        EnvGen_perform<true>(unit, out, level, nsmps,
                             [&](int i) { return check_gate_ar(unit, i, gate, gatein, nsmps, counter, level); });

        remain -= nsmps;
        counter -= nsmps;
    }
    unit->m_level = level;
    unit->m_counter = counter;
    unit->m_prevGate = gate;
}

////////////////////////////////////////////////////////////////////////////////////////////////////////

void Linen_Ctor(Linen* unit) {
    // gate attack level release
    SETCALC(Linen_next_k);

    unit->m_level = 0.f;
    unit->m_stage = 4;
    unit->m_prevGate = 0.f;
    if (ZIN0(0) <= -1.f) {
        unit->m_stage = 1;
    } // early release
    Linen_next_k(unit, 1);
}

void Linen_next_k(Linen* unit, int inNumSamples) {
    float gate = ZIN0(0);
    float* out = OUT(0);

    if (unit->m_prevGate <= 0.f && gate > 0.f) {
        unit->mDone = false;
        unit->m_stage = 0;
        float attackTime = ZIN0(1);
        float susLevel = ZIN0(2);
        int counter = (int)(attackTime * SAMPLERATE);
        counter = sc_max(1, counter);
        unit->m_slope = (susLevel - unit->m_level) / counter;
        unit->m_counter = counter;
    }

    switch (unit->m_stage) {
    case 0:
    case 2:
        *out = unit->m_level;
        unit->m_level += unit->m_slope;
        if (--unit->m_counter == 0)
            unit->m_stage++;
        break;
    case 1:
        *out = unit->m_level;
        if (gate <= -1.f) {
            // cutoff
            unit->m_stage = 2;
            float releaseTime = -gate - 1.f;
            int counter = (int)(releaseTime * SAMPLERATE);
            counter = sc_max(1, counter);
            unit->m_slope = -unit->m_level / counter;
            unit->m_counter = counter;
        } else if (gate <= 0.f) {
            unit->m_stage = 2;
            float releaseTime = ZIN0(3);
            int counter = (int)(releaseTime * SAMPLERATE);
            counter = sc_max(1, counter);
            unit->m_slope = -unit->m_level / counter;
            unit->m_counter = counter;

            // Print("release %d %d\n", unit->mParent->mNode.mID, counter);
        }
        break;
    case 3: {
        *out = 0.f;
        // Print("done %d\n", unit->mParent->mNode.mID);
        unit->mDone = true;
        unit->m_stage++;
        int doneAction = (int)ZIN0(4);
        DoneAction(doneAction, unit);
    } break;
    case 4:
        *out = 0.f;
        break;
    }
    unit->m_prevGate = gate;
}

////////////////////////////////////////////////////////////////////////////////////////////////////////

void EnvFill(World* world, struct SndBuf* buf, struct sc_msg_iter* msg) {
    if (buf->channels != 1)
        return;

    int size = buf->samples;
    int byteSize = size * sizeof(float);
    float* data = (float*)malloc(byteSize);

    double level = msg->getf();
    int numStages = msg->geti();
    /*int releaseNode =*/msg->geti(); // ignored
    /*int loopNode =*/msg->geti(); // ignored

    double pos = 0.;
    int32 index = 0;
    int32 remain = size;

    for (int j = 0; j < numStages; ++j) {
        double endLevel = msg->getf();
        double dur = msg->getf();
        int shape = msg->geti();
        double curve = msg->getf();

        int32 ipos = (int32)pos;
        double smpdur = dur * size;
        int32 nsmps = (int32)smpdur - ipos;
        nsmps = sc_min(nsmps, remain);

        switch (shape) {
        case shape_Step: {
            level = endLevel;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
            }
        } break;
        case shape_Hold: {
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
            }
            level = endLevel;
        } break;
        case shape_Linear: {
            double grow = (endLevel - level) / nsmps;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                level += grow;
            }
        } break;
        case shape_Exponential: {
            double grow = pow(endLevel / level, 1.0 / nsmps);
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                level *= grow;
            }
        } break;
        case shape_Sine: {
            double w = pi / nsmps;

            double a2 = (endLevel + level) * 0.5;
            double b1 = 2. * cos(w);
            double y1 = (endLevel - level) * 0.5;
            double y2 = y1 * sin(pi * 0.5 - w);
            level = a2 - y1;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                double y0 = b1 * y1 - y2;
                level = a2 - y0;
                y2 = y1;
                y1 = y0;
            }
        } break;
        case shape_Welch: {
            double w = (pi * 0.5) / nsmps;

            double b1 = 2. * cos(w);
            double a2, y1, y2;
            if (endLevel >= level) {
                a2 = level;
                y1 = 0.;
                y2 = -sin(w) * (endLevel - level);
            } else {
                a2 = endLevel;
                y1 = level - endLevel;
                y2 = cos(w) * (level - endLevel);
            }
            level = a2 + y1;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                double y0 = b1 * y1 - y2;
                level = a2 - y0;
                y2 = y1;
                y1 = y0;
            }
        } break;
        case shape_Curve: {
            if (fabs(curve) < 0.001) {
                double grow = (endLevel - level) / nsmps;
                for (int i = 0; i < nsmps; ++i) {
                    data[index++] = level;
                    level += grow;
                }
            } else {
                double a1 = (endLevel - level) / (1.0 - exp(curve));
                double a2 = level + a1;
                double b1 = a1;
                double grow = exp(curve / nsmps);
                for (int i = 0; i < nsmps; ++i) {
                    data[index++] = level;
                    b1 *= grow;
                    level = a2 - b1;
                }
            }
        } break;
        case shape_Squared: {
            double y1 = sqrt(level);
            double y2 = sqrt(endLevel);
            double grow = (y2 - y1) / nsmps;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                y1 += grow;
                level = y1 * y1;
            }
        } break;
        case shape_Cubed: {
            double y1 = pow(level, 0.33333333);
            double y2 = pow(endLevel, 0.33333333);
            double grow = (y2 - y1) / nsmps;
            for (int i = 0; i < nsmps; ++i) {
                data[index++] = level;
                y1 += grow;
                level = y1 * y1 * y1;
            }
        } break;
        }

        pos += smpdur;
        level = endLevel;
        remain -= nsmps;
    }
    memcpy(buf->data, data, byteSize);
    free(data);
}

//////////////////// Add IEnvGen 06/06/2007 /////////////////////////////////


struct IEnvGen : public Unit {
    float m_level, m_offset;
    float m_startpoint, m_numvals, m_pointin;
    float* m_envvals;
};


extern "C" {
void IEnvGen_next_a(IEnvGen* unit, int inNumSamples);
void IEnvGen_next_k(IEnvGen* unit, int inNumSamples);
void IEnvGen_Ctor(IEnvGen* unit);
void IEnvGen_Dtor(IEnvGen* unit);
}

#define GET_ENV_VAL                                                                                                    \
    switch (shape) {                                                                                                   \
    case shape_Step:                                                                                                   \
        level = unit->m_level = endLevel;                                                                              \
        break;                                                                                                         \
    case shape_Hold:                                                                                                   \
        level = unit->m_level;                                                                                         \
        unit->m_level = endLevel;                                                                                      \
        break;                                                                                                         \
    case shape_Linear:                                                                                                 \
    default:                                                                                                           \
        level = unit->m_level = pos * (endLevel - begLevel) + begLevel;                                                \
        break;                                                                                                         \
    case shape_Exponential:                                                                                            \
        level = unit->m_level = begLevel * pow(endLevel / begLevel, pos);                                              \
        break;                                                                                                         \
    case shape_Sine:                                                                                                   \
        level = unit->m_level = begLevel + (endLevel - begLevel) * (-cos(pi * pos) * 0.5 + 0.5);                       \
        break;                                                                                                         \
    case shape_Welch: {                                                                                                \
        if (begLevel < endLevel)                                                                                       \
            level = unit->m_level = begLevel + (endLevel - begLevel) * sin(pi2 * pos);                                 \
        else                                                                                                           \
            level = unit->m_level = endLevel - (endLevel - begLevel) * sin(pi2 - pi2 * pos);                           \
        break;                                                                                                         \
    }                                                                                                                  \
    case shape_Curve:                                                                                                  \
        if (fabs((float)curve) < 0.0001) {                                                                             \
            level = unit->m_level = pos * (endLevel - begLevel) + begLevel;                                            \
        } else {                                                                                                       \
            double denom = 1. - exp((float)curve);                                                                     \
            double numer = 1. - exp((float)(pos * curve));                                                             \
            level = unit->m_level = begLevel + (endLevel - begLevel) * (numer / denom);                                \
        }                                                                                                              \
        break;                                                                                                         \
    case shape_Squared: {                                                                                              \
        double sqrtBegLevel = sqrt(begLevel);                                                                          \
        double sqrtEndLevel = sqrt(endLevel);                                                                          \
        double sqrtLevel = pos * (sqrtEndLevel - sqrtBegLevel) + sqrtBegLevel;                                         \
        level = unit->m_level = sqrtLevel * sqrtLevel;                                                                 \
        break;                                                                                                         \
    }                                                                                                                  \
    case shape_Cubed: {                                                                                                \
        double cbrtBegLevel = pow(begLevel, 0.3333333f);                                                               \
        double cbrtEndLevel = pow(endLevel, 0.3333333f);                                                               \
        double cbrtLevel = pos * (cbrtEndLevel - cbrtBegLevel) + cbrtBegLevel;                                         \
        level = unit->m_level = cbrtLevel * cbrtLevel * cbrtLevel;                                                     \
        break;                                                                                                         \
    }                                                                                                                  \
    }


void IEnvGen_Ctor(IEnvGen* unit) {
    if (INRATE(0) == calc_FullRate) {
        SETCALC(IEnvGen_next_a);
    } else {
        SETCALC(IEnvGen_next_k);
    }

    // pointer, offset
    // initlevel, numstages, totaldur,
    // [dur, shape, curve, level] * numvals
    int numStages = (int)IN0(3);
    int numvals = numStages * 4; // initlevel + (levels, dur, shape, curves) * stages
    float offset = unit->m_offset = IN0(1);
    float point = unit->m_pointin = IN0(0) - offset;
    unit->m_envvals = (float*)RTAlloc(unit->mWorld, (int)(numvals + 1.) * sizeof(float));
    ClearUnitIfMemFailed(unit->m_envvals);

    unit->m_envvals[0] = IN0(2);
    //  Print("offset of and initial  values %3,3f, %3.3f\n", offset, unit->m_envvals[0]);
    // fill m_envvals with the values;
    for (int i = 1; i <= numvals; i++) {
        unit->m_envvals[i] = IN0(4 + i);
        //      Print("val for: %d, %3.3f\n", i, unit->m_envvals[i]);
    }

    //  float out = OUT0(0);
    float totalDur = IN0(4);
    float level = 0.f;
    float newtime = 0.f;
    int stage = 0;
    float seglen = 0.f;
    if (point >= totalDur) {
        unit->m_level = level = unit->m_envvals[numStages * 4]; // grab the last value
    } else {
        if (point <= 0.0) {
            unit->m_level = level = unit->m_envvals[0];
        } else {
            float segpos = point;
            // determine which segment the current time pointer needs calculated
            for (int j = 0; point >= newtime; j++) {
                seglen = unit->m_envvals[(j * 4) + 1];
                newtime += seglen;
                segpos -= seglen;
                stage = j;
            }

            segpos = segpos + seglen;
            float begLevel = unit->m_envvals[(stage * 4)];
            int shape = (int)unit->m_envvals[(stage * 4) + 2];
            int curve = (int)unit->m_envvals[(stage * 4) + 3];
            float endLevel = unit->m_envvals[(stage * 4) + 4];
            float pos = (segpos / seglen);

            GET_ENV_VAL
        }
    }
    OUT0(0) = level;
}

void IEnvGen_Dtor(IEnvGen* unit) { RTFree(unit->mWorld, unit->m_envvals); }


void IEnvGen_next_a(IEnvGen* unit, int inNumSamples) {
    float* out = OUT(0);
    float level = unit->m_level;
    float* pointin = IN(0);
    float offset = unit->m_offset;
    int numStages = (int)IN0(3);
    float point; // = unit->m_pointin;

    float totalDur = IN0(4);

    int stagemul;
    // pointer, offset
    // level0, numstages, totaldur,
    // [initval, [dur, shape, curve, level] * N ]

    for (int i = 0; i < inNumSamples; i++) {
        if (pointin[i] == unit->m_pointin) {
            out[i] = level;
        } else {
            unit->m_pointin = point = sc_max(pointin[i] - offset, 0.0);
            float newtime = 0.f;
            int stage = 0;
            float seglen = 0.f;
            if (point >= totalDur) {
                unit->m_level = level = unit->m_envvals[numStages * 4]; // grab the last value
            } else {
                if (point <= 0.0) {
                    unit->m_level = level = unit->m_envvals[0];
                } else {
                    float segpos = point;
                    // determine which segment the current time pointer needs
                    for (int j = 0; point >= newtime; j++) {
                        seglen = unit->m_envvals[(j * 4) + 1];
                        newtime += seglen;
                        segpos -= seglen;
                        stage = j;
                    }
                    stagemul = stage * 4;
                    segpos = segpos + seglen;
                    float begLevel = unit->m_envvals[stagemul];
                    int shape = (int)unit->m_envvals[stagemul + 2];
                    int curve = (int)unit->m_envvals[stagemul + 3];
                    float endLevel = unit->m_envvals[stagemul + 4];
                    float pos = (segpos / seglen);

                    GET_ENV_VAL
                }
            }
            out[i] = level;
        }
    }
}

void IEnvGen_next_k(IEnvGen* unit, int inNumSamples) {
    float* out = OUT(0);
    float level = unit->m_level;
    float pointin = IN0(0);
    float offset = unit->m_offset;
    int numStages = (int)IN0(3);
    float point; // = unit->m_pointin;

    float totalDur = IN0(4);

    int stagemul;
    // pointer, offset
    // level0, numstages, totaldur,
    // [initval, [dur, shape, curve, level] * N ]

    for (int i = 0; i < inNumSamples; i++) {
        if (pointin == unit->m_pointin) {
            out[i] = level;
        } else {
            unit->m_pointin = point = sc_max(pointin - offset, 0.0);
            float newtime = 0.f;
            int stage = 0;
            float seglen = 0.f;
            if (point >= totalDur) {
                unit->m_level = level = unit->m_envvals[numStages * 4]; // grab the last value
            } else {
                if (point <= 0.0) {
                    unit->m_level = level = unit->m_envvals[0];
                } else {
                    float segpos = point;
                    // determine which segment the current time pointer needs
                    for (int j = 0; point >= newtime; j++) {
                        seglen = unit->m_envvals[(j * 4) + 1];
                        newtime += seglen;
                        segpos -= seglen;
                        stage = j;
                    }
                    stagemul = stage * 4;
                    segpos = segpos + seglen;
                    float begLevel = unit->m_envvals[stagemul];
                    int shape = (int)unit->m_envvals[stagemul + 2];
                    int curve = (int)unit->m_envvals[stagemul + 3];
                    float endLevel = unit->m_envvals[stagemul + 4];
                    float pos = (segpos / seglen);

                    GET_ENV_VAL
                }
            }
            out[i] = level;
        }
    }
}

////////////////////////////////////////////////////////////////////////////////////////////////////////


PluginLoad(LF) {
    ft = inTable;

    DefineSimpleUnit(Vibrato);
    DefineSimpleUnit(LFPulse);
    DefineSimpleUnit(LFSaw);
    DefineSimpleUnit(LFPar);
    DefineSimpleUnit(LFCub);
    DefineSimpleUnit(LFTri);
    DefineSimpleUnit(LFGauss);
    DefineSimpleUnit(Impulse);
    DefineSimpleUnit(VarSaw);
    DefineSimpleUnit(SyncSaw);
    registerUnit<K2A>(ft, "K2A");
    DefineSimpleUnit(A2K);
    DefineSimpleUnit(T2K);
    DefineSimpleUnit(T2A);
    registerUnit<DC>(ft, "DC");
    DefineSimpleUnit(Line);
    DefineSimpleUnit(XLine);

    DefineSimpleUnit(Wrap);
    DefineSimpleUnit(Fold);
    DefineSimpleUnit(Clip);
    DefineSimpleUnit(Unwrap);
    DefineSimpleUnit(ModDif);
    DefineSimpleUnit(AmpComp);
    DefineSimpleUnit(AmpCompA);
    DefineSimpleUnit(InRange);
    DefineSimpleUnit(InRect);
    DefineSimpleUnit(LinExp);
    DefineSimpleUnit(EnvGen);
    DefineSimpleUnit(Linen);

    DefineBufGen("env", EnvFill);

    DefineDtorUnit(IEnvGen);
}