/*
 * This file is part of the DXX-Rebirth project <https://www.dxx-rebirth.com/>.
 * It is copyright by its individual contributors, as recorded in the
 * project's Git history.  See COPYING.txt at the top level for license
 * terms and a link to the Git history.
 */

/*
 *
 * C version of vecmat library
 *
 */

#include <stdlib.h>
#include <stdint.h>
#include <math.h>           // for sqrt

#include "maths.h"
#include "vecmat.h"
#include "dxxerror.h"

namespace dcx {

//#define USE_ISQRT 1

constexpr vms_matrix vmd_identity_matrix = IDENTITY_MATRIX;
constexpr vms_vector vmd_zero_vector{};

//adds two vectors, fills in dest, returns ptr to dest
//ok for dest to equal either source, but should use vm_vec_add2() if so
vms_vector &vm_vec_add(vms_vector &dest,const vms_vector &src0,const vms_vector &src1)
{
	dest.x = src0.x + src1.x;
	dest.y = src0.y + src1.y;
	dest.z = src0.z + src1.z;
	return dest;
}


//subs two vectors, fills in dest, returns ptr to dest
//ok for dest to equal either source, but should use vm_vec_sub2() if so
vms_vector &_vm_vec_sub(vms_vector &dest,const vms_vector &src0,const vms_vector &src1)
{
	dest.x = src0.x - src1.x;
	dest.y = src0.y - src1.y;
	dest.z = src0.z - src1.z;
	return dest;
}

//adds one vector to another. returns ptr to dest
//dest can equal source
void vm_vec_add2(vms_vector &dest,const vms_vector &src)
{
	dest.x += src.x;
	dest.y += src.y;
	dest.z += src.z;
}

//subs one vector from another, returns ptr to dest
//dest can equal source
void vm_vec_sub2(vms_vector &dest,const vms_vector &src)
{
	dest.x -= src.x;
	dest.y -= src.y;
	dest.z -= src.z;
}

static inline fix avg_fix(fix64 a, fix64 b)
{
	return (a + b) / 2;
}

//averages two vectors. returns ptr to dest
//dest can equal either source
void vm_vec_avg(vms_vector &dest,const vms_vector &src0,const vms_vector &src1)
{
	dest.x = avg_fix(src0.x, src1.x);
	dest.y = avg_fix(src0.y, src1.y);
	dest.z = avg_fix(src0.z, src1.z);
}

//scales a vector in place.  returns ptr to vector
vms_vector &vm_vec_scale(vms_vector &dest,fix s)
{
	return vm_vec_copy_scale(dest, dest, s);
}

//scales and copies a vector.  returns ptr to dest
vms_vector &vm_vec_copy_scale(vms_vector &dest,const vms_vector &src,fix s)
{
	dest.x = fixmul(src.x,s);
	dest.y = fixmul(src.y,s);
	dest.z = fixmul(src.z,s);
	return dest;
}

//scales a vector, adds it to another, and stores in a 3rd vector
//dest = src1 + k * src2
void vm_vec_scale_add(vms_vector &dest,const vms_vector &src1,const vms_vector &src2,fix k)
{
	dest.x = src1.x + fixmul(src2.x,k);
	dest.y = src1.y + fixmul(src2.y,k);
	dest.z = src1.z + fixmul(src2.z,k);
}

//scales a vector and adds it to another
//dest += k * src
void vm_vec_scale_add2(vms_vector &dest,const vms_vector &src,fix k)
{
	dest.x += fixmul(src.x,k);
	dest.y += fixmul(src.y,k);
	dest.z += fixmul(src.z,k);
}

//scales a vector in place, taking n/d for scale.  returns ptr to vector
//dest *= n/d
void vm_vec_scale2(vms_vector &dest,fix n,fix d)
{
#if 0 // DPH: Kludge: this was overflowing a lot, so I made it use the FPU.
	float nd;
	nd = f2fl(n) / f2fl(d);
	dest.x = fl2f( f2fl(dest.x) * nd);
	dest.y = fl2f( f2fl(dest.y) * nd);
	dest.z = fl2f( f2fl(dest.z) * nd);
#else
	dest.x = fixmuldiv(dest.x,n,d);
	dest.y = fixmuldiv(dest.y,n,d);
	dest.z = fixmuldiv(dest.z,n,d);
#endif
}

[[nodiscard]]
static fix vm_vec_dot3(fix x,fix y,fix z,const vms_vector &v)
{
#if 0
	quadint q;

	q.low = q.high = 0;

	fixmulaccum(&q,x,v->x);
	fixmulaccum(&q,y,v->y);
	fixmulaccum(&q,z,v->z);

	return fixquadadjust(&q);
#else
	int64_t x0 = x;
	int64_t x1 = v.x;
	int64_t y0 = y;
	int64_t y1 = v.y;
	int64_t z0 = z;
	int64_t z1 = v.z;
	int64_t p = (x0 * x1) + (y0 * y1) + (z0 * z1);
	/* Convert back to fix and return. */
	return p >> 16;
#endif
}

fix vm_vec_dot(const vms_vector &v0,const vms_vector &v1)
{
	return vm_vec_dot3(v0.x, v0.y, v0.z, v1);
}

vm_magnitude_squared vm_vec_mag2(const vms_vector &v)
{
	const int64_t x = v.x;
	const int64_t y = v.y;
	const int64_t z = v.z;
	return vm_magnitude_squared{static_cast<uint64_t>((x * x) + (y * y) + (z * z))};
}

vm_magnitude vm_vec_mag(const vms_vector &v)
{
	quadint q;
	q.q = static_cast<uint64_t>(vm_vec_mag2(v));
	return vm_magnitude{quad_sqrt(q)};
}

//computes the distance between two points. (does sub and mag)
vm_distance vm_vec_dist(const vms_vector &v0,const vms_vector &v1)
{
	return vm_vec_mag(vm_vec_sub(v0,v1));
}

vm_distance_squared vm_vec_dist2(const vms_vector &v0,const vms_vector &v1)
{
	return vm_vec_mag2(vm_vec_sub(v0,v1));
}

//computes an approximation of the magnitude of the vector
//uses dist = largest + next_largest*3/8 + smallest*3/16
vm_magnitude vm_vec_mag_quick(const vms_vector &v)
{
	fix a,b,c,bc;

	a = labs(v.x);
	b = labs(v.y);
	c = labs(v.z);

	if (a < b) {
		std::swap(a, b);
	}

	if (b < c) {
		std::swap(b, c);
		if (a < b) {
			std::swap(a, b);
		}
	}

	bc = (b>>2) + (c>>3);

	return vm_magnitude{static_cast<uint32_t>(a + bc + (bc>>1))};
}


//computes an approximation of the distance between two points.
//uses dist = largest + next_largest*3/8 + smallest*3/16
vm_distance vm_vec_dist_quick(const vms_vector &v0,const vms_vector &v1)
{
	return vm_vec_mag_quick(vm_vec_sub(v0,v1));
}

//normalize a vector. returns mag of source vec
vm_magnitude vm_vec_copy_normalize(vms_vector &dest,const vms_vector &src)
{
	auto m = vm_vec_mag(src);
	if (m) {
		vm_vec_divide(dest, src, m);
	}
	return m;
}

void vm_vec_divide(vms_vector &dest,const vms_vector &src, fix m)
{
	dest.x = fixdiv(src.x,m);
	dest.y = fixdiv(src.y,m);
	dest.z = fixdiv(src.z,m);
}

//normalize a vector. returns mag of source vec
vm_magnitude vm_vec_normalize(vms_vector &v)
{
	return vm_vec_copy_normalize(v,v);
}

//normalize a vector. returns mag of source vec. uses approx mag
vm_magnitude vm_vec_copy_normalize_quick(vms_vector &dest,const vms_vector &src)
{
	auto m = vm_vec_mag_quick(src);
	if (m) {
		vm_vec_divide(dest, src, m);
	}
	return m;
}

//normalize a vector. returns 1/mag of source vec. uses approx 1/mag
vm_magnitude vm_vec_normalize_quick(vms_vector &v)
{
	return vm_vec_copy_normalize_quick(v,v);
}

//return the normalized direction vector between two points
//dest = normalized(end - start).  Returns 1/mag of direction vector
//NOTE: the order of the parameters matches the vector subtraction
vm_magnitude vm_vec_normalized_dir_quick(vms_vector &dest,const vms_vector &end,const vms_vector &start)
{
	return vm_vec_normalize_quick(vm_vec_sub(dest,end,start));
}

//return the normalized direction vector between two points
//dest = normalized(end - start).  Returns mag of direction vector
//NOTE: the order of the parameters matches the vector subtraction
vm_magnitude vm_vec_normalized_dir(vms_vector &dest,const vms_vector &end,const vms_vector &start)
{
	return vm_vec_normalize(vm_vec_sub(dest,end,start));
}

//computes surface normal from three points. result is normalized
//returns ptr to dest
//dest CANNOT equal either source
void vm_vec_normal(vms_vector &dest,const vms_vector &p0,const vms_vector &p1,const vms_vector &p2)
{
	vm_vec_perp(dest,p0,p1,p2);
	vm_vec_normalize(dest);
}

//make sure a vector is reasonably sized to go into a cross product
static void check_vec(vms_vector *v)
{
	fix check;
	int cnt = 0;

	check = labs(v->x) | labs(v->y) | labs(v->z);
	
	if (check == 0)
		return;

	if (check & 0xfffc0000) {		//too big

		while (check & 0xfff00000) {
			cnt += 4;
			check >>= 4;
		}

		while (check & 0xfffc0000) {
			cnt += 2;
			check >>= 2;
		}

		v->x >>= cnt;
		v->y >>= cnt;
		v->z >>= cnt;
	}
	else												//maybe too small
		if ((check & 0xffff8000) == 0) {		//yep, too small

			while ((check & 0xfffff000) == 0) {
				cnt += 4;
				check <<= 4;
			}

			while ((check & 0xffff8000) == 0) {
				cnt += 2;
				check <<= 2;
			}

			v->x >>= cnt;
			v->y >>= cnt;
			v->z >>= cnt;
		}
}

//computes cross product of two vectors. 
//Note: this magnitude of the resultant vector is the
//product of the magnitudes of the two source vectors.  This means it is
//quite easy for this routine to overflow and underflow.  Be careful that
//your inputs are ok.
void vm_vec_cross(vms_vector &dest,const vms_vector &src0,const vms_vector &src1)
{
	quadint q;

	Assert(&dest!=&src0 && &dest!=&src1);

	q.q = 0;
	fixmulaccum(&q,src0.y,src1.z);
	fixmulaccum(&q,-src0.z,src1.y);
	dest.x = fixquadadjust(&q);

	q.q = 0;
	fixmulaccum(&q,src0.z,src1.x);
	fixmulaccum(&q,-src0.x,src1.z);
	dest.y = fixquadadjust(&q);

	q.q = 0;
	fixmulaccum(&q,src0.x,src1.y);
	fixmulaccum(&q,-src0.y,src1.x);
	dest.z = fixquadadjust(&q);
}

//computes non-normalized surface normal from three points. 
//returns ptr to dest
//dest CANNOT equal either source
void vm_vec_perp(vms_vector &dest,const vms_vector &p0,const vms_vector &p1,const vms_vector &p2)
{
	auto t0 = vm_vec_sub(p1,p0);
	auto t1 = vm_vec_sub(p2,p1);
	check_vec(&t0);
	check_vec(&t1);
	vm_vec_cross(dest,t0,t1);
}


//computes the delta angle between two vectors. 
//vectors need not be normalized. if they are, call vm_vec_delta_ang_norm()
//the forward vector (third parameter) can be NULL, in which case the absolute
//value of the angle in returned.  Otherwise the angle around that vector is
//returned.
fixang vm_vec_delta_ang(const vms_vector &v0,const vms_vector &v1,const vms_vector &fvec)
{
	vms_vector t0,t1;

	if (!vm_vec_copy_normalize(t0,v0) || !vm_vec_copy_normalize(t1,v1))
		return 0;

	return vm_vec_delta_ang_norm(t0,t1,fvec);
}

//computes the delta angle between two normalized vectors. 
fixang vm_vec_delta_ang_norm(const vms_vector &v0,const vms_vector &v1,const vms_vector &fvec)
{
	fixang a;

	a = fix_acos(vm_vec_dot(v0,v1));
		if (vm_vec_dot(vm_vec_cross(v0,v1),fvec) < 0)
			a = -a;
	return a;
}

static void sincos_2_matrix(vms_matrix &m, const fixang bank, const fix_sincos_result p, const fix_sincos_result h)
{
#define DXX_S2M_DECL(V)	\
	const auto &sin##V = V.sin;	\
	const auto &cos##V = V.cos
	DXX_S2M_DECL(p);
	DXX_S2M_DECL(h);
	m.fvec.y = -sinp;								//m6
	m.fvec.x = fixmul(sinh,cosp);				//m3
	m.fvec.z = fixmul(cosh,cosp);				//m9
	const auto &&b = fix_sincos(bank);
	DXX_S2M_DECL(b);
#undef DXX_S2M_DECL
	m.rvec.y = fixmul(sinb,cosp);				//m4
	m.uvec.y = fixmul(cosb,cosp);				//m5

	const auto cbch = fixmul(cosb,cosh);
	const auto sbsh = fixmul(sinb,sinh);
	m.rvec.x = cbch + fixmul(sinp,sbsh);		//m1
	m.uvec.z = sbsh + fixmul(sinp,cbch);		//m8

	const auto sbch = fixmul(sinb,cosh);
	const auto cbsh = fixmul(cosb,sinh);
	m.uvec.x = fixmul(sinp,cbsh) - sbch;		//m2
	m.rvec.z = fixmul(sinp,sbch) - cbsh;		//m7
}

//computes a matrix from a set of three angles.  returns ptr to matrix
void vm_angles_2_matrix(vms_matrix &m,const vms_angvec &a)
{
	const auto al = a;
	sincos_2_matrix(m, al.b, fix_sincos(al.p), fix_sincos(al.h));
}

#if DXX_USE_EDITOR
//computes a matrix from a forward vector and an angle
void vm_vec_ang_2_matrix(vms_matrix &m,const vms_vector &v,fixang a)
{
	fix sinp,cosp,sinh,cosh;
	sinp = -v.y;
	cosp = fix_sqrt(f1_0 - fixmul(sinp,sinp));

	sinh = fixdiv(v.x,cosp);
	cosh = fixdiv(v.z,cosp);
	sincos_2_matrix(m, a, {sinp, cosp}, {sinh, cosh});
}
#endif

//computes a matrix from one or more vectors. The forward vector is required,
//with the other two being optional.  If both up & right vectors are passed,
//the up vector is used.  If only the forward vector is passed, a bank of
//zero is assumed
//returns ptr to matrix
void vm_vector_2_matrix(vms_matrix &m,const vms_vector &fvec,const vms_vector *uvec,const vms_vector *rvec)
{
	vms_vector &xvec=m.rvec,&yvec=m.uvec,&zvec=m.fvec;
	if (!vm_vec_copy_normalize(zvec,fvec)) {
		Int3();		//forward vec should not be zero-length
		return;
	}
	if (uvec == NULL) {
		if (rvec == NULL) {		//just forward vec
bad_vector2:
	;
			if (zvec.x==0 && zvec.z==0) {		//forward vec is straight up or down

				m.rvec.x = f1_0;
				m.uvec.z = (zvec.y < 0)?f1_0:-f1_0;

				m.rvec.y = m.rvec.z = m.uvec.x = m.uvec.y = 0;
			}
			else { 		//not straight up or down

				xvec.x = zvec.z;
				xvec.y = 0;
				xvec.z = -zvec.x;

				vm_vec_normalize(xvec);
				vm_vec_cross(yvec,zvec,xvec);
			}
		}
		else {						//use right vec

			if (!vm_vec_copy_normalize(xvec,*rvec))
				goto bad_vector2;

			vm_vec_cross(yvec,zvec,xvec);

			//normalize new perpendicular vector
			if (!vm_vec_normalize(yvec))
				goto bad_vector2;

			//now recompute right vector, in case it wasn't entirely perpendiclar
			vm_vec_cross(xvec,yvec,zvec);

		}
	}
	else {		//use up vec

		if (!vm_vec_copy_normalize(yvec,*uvec))
			goto bad_vector2;

		vm_vec_cross(xvec,yvec,zvec);
		
		//normalize new perpendicular vector
		if (!vm_vec_normalize(xvec))
			goto bad_vector2;

		//now recompute up vector, in case it wasn't entirely perpendiclar
		vm_vec_cross(yvec,zvec,xvec);
	}
}


//rotates a vector through a matrix. returns ptr to dest vector
//dest CANNOT equal source
void vm_vec_rotate(vms_vector &dest,const vms_vector &src,const vms_matrix &m)
{
	dest.x = vm_vec_dot(src,m.rvec);
	dest.y = vm_vec_dot(src,m.uvec);
	dest.z = vm_vec_dot(src,m.fvec);
}

//mulitply 2 matrices, fill in dest.  returns ptr to dest
//dest CANNOT equal either source
void _vm_matrix_x_matrix(vms_matrix &dest,const vms_matrix &src0,const vms_matrix &src1)
{
	dest.rvec.x = vm_vec_dot3(src0.rvec.x,src0.uvec.x,src0.fvec.x, src1.rvec);
	dest.uvec.x = vm_vec_dot3(src0.rvec.x,src0.uvec.x,src0.fvec.x, src1.uvec);
	dest.fvec.x = vm_vec_dot3(src0.rvec.x,src0.uvec.x,src0.fvec.x, src1.fvec);

	dest.rvec.y = vm_vec_dot3(src0.rvec.y,src0.uvec.y,src0.fvec.y, src1.rvec);
	dest.uvec.y = vm_vec_dot3(src0.rvec.y,src0.uvec.y,src0.fvec.y, src1.uvec);
	dest.fvec.y = vm_vec_dot3(src0.rvec.y,src0.uvec.y,src0.fvec.y, src1.fvec);

	dest.rvec.z = vm_vec_dot3(src0.rvec.z,src0.uvec.z,src0.fvec.z, src1.rvec);
	dest.uvec.z = vm_vec_dot3(src0.rvec.z,src0.uvec.z,src0.fvec.z, src1.uvec);
	dest.fvec.z = vm_vec_dot3(src0.rvec.z,src0.uvec.z,src0.fvec.z, src1.fvec);
}

//extract angles from a matrix 
void vm_extract_angles_matrix(vms_angvec &a,const vms_matrix &m)
{
	fix cosp;

	if (m.fvec.x==0 && m.fvec.z==0)		//zero head
		a.h = 0;
	else
		a.h = fix_atan2(m.fvec.z,m.fvec.x);

	const auto &&ah = fix_sincos(a.h);
	const auto &sinh = ah.sin;
	const auto &cosh = ah.cos;

	if (abs(sinh) > abs(cosh))				//sine is larger, so use it
		cosp = fixdiv(m.fvec.x,sinh);
	else											//cosine is larger, so use it
		cosp = fixdiv(m.fvec.z,cosh);

	if (cosp==0 && m.fvec.y==0)
		a.p = 0;
	else
		a.p = fix_atan2(cosp,-m.fvec.y);


	if (cosp == 0)	//the cosine of pitch is zero.  we're pitched straight up. say no bank

		a.b = 0;

	else {
		fix sinb,cosb;

		sinb = fixdiv(m.rvec.y,cosp);
		cosb = fixdiv(m.uvec.y,cosp);

		if (sinb==0 && cosb==0)
			a.b = 0;
		else
			a.b = fix_atan2(cosb,sinb);

	}
}


//extract heading and pitch from a vector, assuming bank==0
static vms_angvec &vm_extract_angles_vector_normalized(vms_angvec &a,const vms_vector &v)
{
	a.b = 0;		//always zero bank
	a.p = fix_asin(-v.y);
	if (v.x == 0 && v.z == 0)
		a.h = 0;
	else
		a.h = fix_atan2(v.z,v.x);
	return a;
}

//extract heading and pitch from a vector, assuming bank==0
void vm_extract_angles_vector(vms_angvec &a,const vms_vector &v)
{
	vms_vector t;
	if (vm_vec_copy_normalize(t,v))
		vm_extract_angles_vector_normalized(a,t);
	else
		a = {};
}

//compute the distance from a point to a plane.  takes the normalized normal
//of the plane (ebx), a point on the plane (edi), and the point to check (esi).
//returns distance in eax
//distance is signed, so negative dist is on the back of the plane
fix vm_dist_to_plane(const vms_vector &checkp,const vms_vector &norm,const vms_vector &planep)
{
	return vm_vec_dot(vm_vec_sub(checkp,planep),norm);
}

// convert vms_matrix to vms_quaternion
void vms_quaternion_from_matrix(vms_quaternion &rq, const vms_matrix &m)
{
	const auto rx = m.rvec.x;
	const auto ry = m.rvec.y;
	const auto rz = m.rvec.z;
	const auto ux = m.uvec.x;
	const auto uy = m.uvec.y;
	const auto uz = m.uvec.z;
	const auto fx = m.fvec.x;
	const auto fy = m.fvec.y;
	const auto fz = m.fvec.z;
	const fix tr = rx + uy + fz;
	fix qw, qx, qy, qz;
	if (tr > 0) {
		fix s = fixmul(fix_sqrt(tr + fl2f(1.0)), fl2f(2.0));
		qw = fixmul(fl2f(0.25), s);
		qx = fixdiv(fy - uz, s);
		qy = fixdiv(rz - fx, s);
		qz = fixdiv(ux - ry, s);
	} else if ((rx > uy) & (rx > fz)) {
		fix s = fixmul(fix_sqrt(fl2f(1.0) + rx - uy - fz), fl2f(2.0));
		qw = fixdiv(fy - uz, s);
		qx = fixmul(fl2f(0.25), s);
		qy = fixdiv(ry + ux, s);
		qz = fixdiv(rz + fx, s);
	} else if (uy > fz) { 
		fix s = fixmul(fix_sqrt(fl2f(1.0) + uy - rx - fz), fl2f(2.0));
		qw = fixdiv(rz - fx, s);
		qx = fixdiv(ry + ux, s);
		qy = fixmul(fl2f(0.25), s);
		qz = fixdiv(uz + fy, s);
	} else { 
		fix s = fixmul(fix_sqrt(fl2f(1.0) + fz - rx - uy), fl2f(2.0));
		qw = fixdiv(ux - ry, s);
		qx = fixdiv(rz + fx, s);
		qy = fixdiv(uz + fy, s);
		qz = fixmul(fl2f(0.25), s);
	}
	rq.w = qw / 2;
	rq.x = qx / 2;
	rq.y = qy / 2;
	rq.z = qz / 2;
}

// convert vms_quaternion to vms_matrix
void vms_matrix_from_quaternion(vms_matrix &m, const vms_quaternion &q)
{
	const fix qw2 = q.w * 2;
	const fix qx2 = q.x * 2;
	const fix qy2 = q.y * 2;
	const fix qz2 = q.z * 2;
	const fix sqw = fixmul(qw2, qw2);
	const fix sqx = fixmul(qx2, qx2);
	const fix sqy = fixmul(qy2, qy2);
	const fix sqz = fixmul(qz2, qz2);
	const fix invs = fixdiv(fl2f(1.0), (sqw + sqx + sqy + sqz));

	m.rvec.x = fixmul(sqx - sqy - sqz + sqw, invs);
	m.uvec.y = fixmul(-sqx + sqy - sqz + sqw, invs);
	m.fvec.z = fixmul(-sqx - sqy + sqz + sqw, invs);
	
	const fix qxy = fixmul(qx2, qy2);
	const fix qzw = fixmul(qz2, qw2);
	m.uvec.x = fixmul(fixmul(fl2f(2.0), (qxy + qzw)), invs);
	m.rvec.y = fixmul(fixmul(fl2f(2.0), (qxy - qzw)), invs);
	
	const fix qxz = fixmul(qx2, qz2);
	const fix qyw = fixmul(qy2, qw2);
	m.fvec.x = fixmul(fixmul(fl2f(2.0), (qxz - qyw)), invs);
	m.rvec.z = fixmul(fixmul(fl2f(2.0), (qxz + qyw)), invs);
	
	const fix qyz = fixmul(qy2, qz2);
	const fix qxw = fixmul(qx2, qw2);
	m.fvec.y = fixmul(fixmul(fl2f(2.0), (qyz + qxw)), invs);
	m.uvec.z = fixmul(fixmul(fl2f(2.0), (qyz - qxw)), invs);
}

}