import math
import numpy as np
from scipy.optimize import fsolve


def antiparallel2gap(K, phi, undudict):
    gLH = K2gap(K, undudict['K-value_LH'])
    if phi >= 0.0:
        gLV = K2gap(K, undudict['K-value_apLV+'])
        gC = K2gap(K, undudict['K-value_45+'])
        dgLV = gLV - gLH
        dgC = gC - gLH - dgLV/2
    else:
        gLV = K2gap(K, undudict['K-value_apLV-'])
        gC = K2gap(K, undudict['K-value_45-'])
        dgLV = gLV - gLH
        dgC = gC - gLH - dgLV/2
    return gLH + dgLV * np.sin(0.5 * phi)**2 + dgC * np.sin(phi)**2


def K2gap(Kval, fitparam):
    g2K_func = np.poly1d(fitparam[::-1])
    tau_init = 1.0
    k_log = float(np.log(Kval))
    return float(fsolve(k_log - g2K_func, tau_init))



## TODO: implement the proper model:


def antiparallel_rad2k_fit(radial, phi_in_degree, undudict):
    key = "anti-parallel-" if phi_in_degree < 0 else "anti-parallel+"
    undudict = undudict[key]["K-value"]
    phi = math.radians(phi_in_degree)
    amp1 = getfit(radial, undudict['fitpars_amp1'])
    amp2 = getfit(radial, undudict['fitpars_amp2'])
    amp3 = getfit(radial, undudict['fitpars_amp3'])
    KLH = np.exp(getfit(radial, undudict['fitpars_KLH']))
    KLV = np.exp(getfit(radial, undudict['fitpars_KLV']))
    K45 = np.exp(getfit(radial, undudict['fitpars_K45']))
    dKLV = KLV - KLH
    dK45 = K45 - KLH - dKLV/2
    #print(radial, phi_in_degree, amp1, amp2, amp3, KLH, dKLV, dK45)
    if phi >= 0.0:
        return KLH + dKLV * np.sin(0.5*phi)**2 + dK45 * np.sin(phi)**2 + amp1 * np.sin(2*phi) + amp2 * np.sin(2.0*phi)**2 + amp3 * np.cos(6*phi)
    else:
        return KLH + dKLV * np.sin(0.5*phi)**2 + dK45 * np.sin(phi)**2 + amp1 * np.sin(2*phi) + amp2 * np.sin(2.0*phi)**2 + amp3 * np.cos(8*phi)


def getfit(val, fitparam):
    fit_func = np.poly1d(fitparam)
    val = float(val)
    return fit_func(val)


def antiparallel_k2rad_fit(Kval, phi_in_degree, undudict):
    key = "anti-parallel-" if phi_in_degree < 0 else "anti-parallel+"
    undudict = undudict[key]["K-value"]
    phi = math.radians(phi_in_degree)
    amp1 = getfit(Kval, undudict['fitpars_amp1'])
    amp2 = getfit(Kval, undudict['fitpars_amp2'])
    amp3 = getfit(Kval, undudict['fitpars_amp3'])
    RadLH = getfit_inverse(np.log(Kval), undudict['fitpars_KLH'])
    RadLV = getfit_inverse(np.log(Kval), undudict['fitpars_KLV'])
    Rad45 = getfit_inverse(np.log(Kval), undudict['fitpars_K45'])
    dRadLV = RadLV - RadLH
    dRad45 = Rad45 - RadLH - dRadLV/2
    #print(radial, phi_in_degree, amp1, amp2, amp3, RadLH, RadLV, Rad45)
    if phi >= 0.0:
        return RadLH + dRadLV * np.sin(0.5*phi)**2 + dRad45 * np.sin(phi)**2 + amp1 * np.sin(2*phi) + amp2 * np.sin(2.0*phi)**2 + amp3 * np.cos(6*phi)
    else:
        return RadLH + dRadLV * np.sin(0.5*phi)**2 + dRad45 * np.sin(phi)**2 + amp1 * np.sin(2*phi) + amp2 * np.sin(2.0*phi)**2 + amp3 * np.cos(8*phi)


def getfit_inverse(val, fitparam):
    fit_func = np.poly1d(fitparam)
    tau_init = 1.0
    val = float(val)
    return float(fsolve(val - fit_func, tau_init))