# -*- coding: utf-8 -*-

from mutatorMath.objects.error import MutatorError
from mutatorMath.objects.location import Location, sortLocations, biasFromLocations

import sys, warnings
from operator import itemgetter


__all__ = ['Mutator', 'buildMutator']

_EPSILON = sys.float_info.epsilon


def noBend(loc): return loc


def buildMutator(items, axes=None, bias=None):
    """
        Build a mutator with the (location, obj) pairs in items.
        Determine the bias based on the given locations.
    """
    from mutatorMath.objects.bender import Bender
    m = Mutator()
    if axes is not None:
        bender = Bender(axes)
        m.setBender(bender)
    else:
        bender = noBend
    # the order itself does not matter, but we should always build in the same order.
    items = sorted(items)
    if not bias:
        bias = biasFromLocations([bender(Location(loc)) for loc, obj in items], True)
    else:
        # note: this means that the actual bias might be different from the initial value. 
        bias = bender(bias)
    m.setBias(bias)
    n = None
    ofx = []
    onx = []
    for loc, obj in items:
        loc = bender(Location(loc))
        if (loc-bias).isOrigin():
            m.setNeutral(obj)
            break
    if m.getNeutral() is None:
        raise MutatorError("Did not find a neutral for this system", m)
    for loc, obj in items:
        locbent = bender(Location(loc))
        #lb = loc-bias
        lb = locbent-bias
        if lb.isOrigin(): continue
        if lb.isOnAxis():
            onx.append((lb, obj-m.getNeutral()))
        else:
            ofx.append((lb, obj-m.getNeutral()))
    for loc, obj in onx:
        m.addDelta(loc, obj, punch=False,  axisOnly=True)
    for loc, obj in ofx:
        m.addDelta(loc, obj, punch=True,  axisOnly=True)
    return bias, m


class Mutator(dict):

    """
        Calculator for multi dimensional interpolations.

    ::

        # The mutator needs one neutral object.
        m = Mutator(myNeutralMathObject)

        # The mutator needs one or more deltas.
        m.addDelta(Location(pop=1), myMasterMathObject-myNeutralMathObject)

        # The mutator calculates instances at other locations. Remember to inflate.
        m.getInstance(Location(pop=0.5)) + myNeutralMathObject

    """

    def __init__(self, neutral=None):
        self._axes = {}
        self._tags = {}
        self._bender = noBend
        self._neutral = neutral
        self._bias = Location()

    def setBender(self, bender):
        self._bender = bender

    def setBias(self, bias):
        self._bias = bias

    def getBias(self):
        return self._bias

    def setNeutral(self, aMathObject, deltaName="origin"):
        """Set the neutral object."""
        self._neutral = aMathObject
        self.addDelta(Location(), aMathObject-aMathObject, deltaName, punch=False, axisOnly=True)

    def getNeutral(self):
        """Get the neutral object."""
        return self._neutral

    def addDelta(self, location, aMathObject, deltaName = None, punch=False,  axisOnly=True):
        """ Add a delta at this location.
            *   location:   a Location object
            *   mathObject: a math-sensitive object
            *   deltaName: optional string/token
            *   punch:
                *   True: add the difference with the instance value at that location and the delta
                *   False: just add the delta.
        """
        #location = self._bender(location)
        if punch:
            r = self.getInstance(location, axisOnly=axisOnly)
            if r is not None:
                self[location.asTuple()] = aMathObject-r, deltaName
            else:
                raise MutatorError("Could not get instance.")
        else:
            self[location.asTuple()] = aMathObject, deltaName

    #
    # info
    #

    def getAxisNames(self):
        """
            Collect a set of axis names from all deltas.
        """
        s = {}
        for l, x in self.items():
            s.update(dict.fromkeys([k for k, v in l], None))
        return set(s.keys())

    def _collectAxisPoints(self):
        """
            Return a dictionary with all on-axis locations.
        """
        for l, (value, deltaName) in self.items():
            location = Location(l)
            name = location.isOnAxis()
            if name is not None and name is not False:
                if name not in self._axes:
                    self._axes[name] = []
                if l not in self._axes[name]:
                    self._axes[name].append(l)
        return self._axes

    def _collectOffAxisPoints(self):
        """
            Return a dictionary with all off-axis locations.
        """
        offAxis = {}
        for l, (value, deltaName) in self.items():
            location = Location(l)
            name = location.isOnAxis()
            if name is None or name is False:
                offAxis[l] = 1
        return list(offAxis.keys())


    def collectLocations(self):
        """
            Return a dictionary with all objects.
        """
        pts = []
        for l, (value, deltaName) in self.items():
            pts.append(Location(l))
        return pts

    def _allLocations(self):
        """
            Return a list of all locations of all objects.
        """
        l = []
        for locationTuple in self.keys():
            l.append(Location(locationTuple))
        return l

    #
    #   get instances
    #

    def getInstance(self, aLocation, axisOnly=False, getFactors=False):

        """ Calculate the delta at aLocation.
            *   aLocation:  a Location object, expected to be in bent space
            *   axisOnly:
                *   True: calculate an instance only with the on-axis masters.
                *   False: calculate an instance with on-axis and off-axis masters.
            *   getFactors:
                *   True: return a list of the calculated factors.
        """
        self._collectAxisPoints()
        factors = self.getFactors(aLocation, axisOnly)
        total = None
        for f, item, name in factors:
            if total is None:
                total = f * item
                continue
            total += f * item
        if total is None:
            total = 0 * self._neutral
        if getFactors:
            return total, factors
        return total

    def makeInstance(self, aLocation, bend=True):
        """
            Calculate an instance with the right bias and add the neutral.
            aLocation: expected to be in input space
        """
        if bend:
            aLocation = self._bender(Location(aLocation))
        if not aLocation.isAmbivalent():
            instanceObject = self.getInstance(aLocation-self._bias)
        else:
            locX, locY = aLocation.split()
            instanceObject = self.getInstance(locX-self._bias)*(1,0)+self.getInstance(locY-self._bias)*(0,1)
        return instanceObject+self._neutral

    def getFactors(self, aLocation, axisOnly=False, allFactors=False):
        """
            Return a list of all factors and math items at aLocation.
            factor, mathItem, deltaName
            all = True: include factors that are zero or near-zero
        """
        deltas = []
        aLocation.expand(self.getAxisNames())
        limits = getLimits(self._allLocations(), aLocation)
        for deltaLocationTuple, (mathItem, deltaName) in sorted(self.items()):
            deltaLocation = Location(deltaLocationTuple)
            deltaLocation.expand( self.getAxisNames())
            factor = self._accumulateFactors(aLocation, deltaLocation, limits, axisOnly)
            if not (factor-_EPSILON < 0 < factor+_EPSILON) or allFactors:
                # only add non-zero deltas.
                deltas.append((factor, mathItem, deltaName))
        deltas = sorted(deltas, key=itemgetter(0), reverse=True)
        return deltas

    #
    #   calculate
    #

    def _accumulateFactors(self, aLocation, deltaLocation, limits, axisOnly):
        """
            Calculate the factors of deltaLocation towards aLocation,
        """
        relative = []
        deltaAxis = deltaLocation.isOnAxis()
        if deltaAxis is None:
            relative.append(1)
        elif deltaAxis:
            deltasOnSameAxis = self._axes.get(deltaAxis, [])
            d = ((deltaAxis, 0),)
            if d not in deltasOnSameAxis:
                deltasOnSameAxis.append(d)
            if len(deltasOnSameAxis) == 1:
                relative.append(aLocation[deltaAxis] * deltaLocation[deltaAxis])
            else:
                factor =  self._calcOnAxisFactor(aLocation, deltaAxis, deltasOnSameAxis, deltaLocation)
                relative.append(factor)
        elif not axisOnly:
            factor = self._calcOffAxisFactor(aLocation, deltaLocation, limits)
            relative.append(factor)
        if not relative:
            return 0
        f = None
        for v in relative:
            if f is None: f = v
            else:
                f *= v
        return f

    def _calcOnAxisFactor(self, aLocation, deltaAxis, deltasOnSameAxis, deltaLocation):
        """
            Calculate the on-axis factors.
        """
        if deltaAxis == "origin":
            f = 0
            v = 0
        else:
            f = aLocation[deltaAxis]
            v = deltaLocation[deltaAxis]
        i = []
        iv = {}
        for value in deltasOnSameAxis:
            iv[Location(value)[deltaAxis]]=1
        i = sorted(iv.keys())
        r = 0
        B, M, A = [], [], []
        mA, mB, mM = None, None, None
        for value in i:
            if value < f: B.append(value)
            elif value > f: A.append(value)
            else: M.append(value)
        if len(B) > 0:
            mB = max(B)
            B.sort()
        if len(A) > 0:
            mA = min(A)
            A.sort()
        if len(M) > 0:
            mM = min(M)
            M.sort()
        if mM is not None:
            if ((f-_EPSILON <  v) and (f+_EPSILON > v)) or f==v: r = 1
            else: r = 0
        elif mB is not None and mA is not None:
            if v < mB or v > mA: r = 0
            else:
                if v == mA:
                    r = float(f-mB)/(mA-mB)
                else:
                    r = float(f-mA)/(mB-mA)
        elif mB is None and mA is not None:
            if v==A[1]:
                r = float(f-A[0])/(A[1]-A[0])
            elif v == A[0]:
                r = float(f-A[1])/(A[0]-A[1])
            else:
                r = 0
        elif  mB is not None and mA is None:
            if v == B[-2]:
                r = float(f-B[-1])/(B[-2]-B[-1])
            elif v == mB:
                r = float(f-B[-2])/(B[-1]-B[-2])
            else:
                r = 0
        return r

    def _calcOffAxisFactor(self, aLocation, deltaLocation, limits):
        """
            Calculate the off-axis factors.
        """
        relative = []
        for dim in limits.keys():
            f = aLocation[dim]
            v = deltaLocation[dim]
            mB, M, mA = limits[dim]
            r = 0
            if mA is not None and v > mA:
                relative.append(0)
                continue
            elif mB is not None and v < mB:
                relative.append(0)
                continue
            if f < v-_EPSILON:
                if mB is None:
                    if M is not None and mA is not None:
                        if v == M:
                            r = (float(max(f,mA)-min(f, mA))/float(max(M,mA)-min(M, mA)))
                        else:
                            r = -(float(max(f,mA)-min(f, mA))/float(max(M,mA)-min(M, mA)) -1)
                    else: r = 0
                elif mA is None: r = 0
                else: r = float(f-mB)/(mA-mB)
            elif f > v+_EPSILON:
                if mB is None: r = 0
                elif mA is None:
                    if M is not None and mB is not None:
                        if v == M:
                            r = (float(max(f,mB)-min(f, mB))/(max(mB, M)-min(mB, M)))
                        else:
                            r = -(float(max(f,mB)-min(f, mB))/(max(mB, M)-min(mB, M)) - 1)
                    else: r = 0
                else: r = float(mA-f)/(mA-mB)
            else: r = 1
            relative.append(r)
        f = 1
        for i in relative:
            f *= i
        return f


def getLimits(locations, current, sortResults=True, verbose=False):
    """
        Find the projections for each delta in the list of locations, relative to the current location.
        Return only the dimensions that are relevant for current.
    """
    limit = {}
    for l in locations:
        a, b = current.common(l)
        if a is None:
            continue
        for name, value in b.items():
            f = a[name]
            if name not in limit:
                limit[name] = {}
                limit[name]['<'] = {}
                limit[name]['='] = {}
                limit[name]['>'] = {}
                if f > 0:
                    limit[name]['>'] = {0: [Location()]}
                elif f<0:
                    limit[name]['<'] = {0: [Location()]}
                else:
                    limit[name]['='] = {0: [Location()]}
            if current[name] < value - _EPSILON:
                if value not in limit[name]["<"]:
                    limit[name]["<"][value] = []
                limit[name]["<"][value].append(l)
            elif current[name] > value + _EPSILON:
                if value not in limit[name][">"]:
                    limit[name][">"][value] = []
                limit[name][">"][value].append(l)
            else:
                if value not in limit[name]["="]:
                    limit[name]["="][value] = []
                limit[name]["="][value].append(l)
    if not sortResults:
        return limit
    # now we have all the data, let's sort to the relevant values
    l = {}
    for name, lims in  limit.items():
        less = []
        more = []
        if lims[">"].keys():
            less = sorted(lims[">"].keys())
            lim_min = less[-1]
        else:
            lim_min = None
        if lims["<"].keys():
            more = sorted(lims["<"].keys())
            lim_max = more[0]
        else:
            lim_max = None
        if lim_min is None and lim_max is not None:
            # extrapolation < min
            if len(limit[name]['='])>0:
                l[name] = (None, list(limit[name]['='].keys())[0], None)
            elif len(more) > 1 and len(limit[name]['='])==0:
                # extrapolation
                l[name] = (None,  more[0], more[1])
        elif lim_min is not None and lim_max is None:
            # extrapolation < max
            if len(limit[name]['='])>0:
                # less > 0, M > 0, more = None
                # -> end of a limit
                l[name] = (None, limit[name]['='], None)
            elif len(less) > 1 and len(limit[name]['='])==0:
                # less > 0, M = None, more = None
                # extrapolation
                l[name] = (less[-2], less[-1], None)
        else:
            if len(limit[name]['=']) > 0:
                l[name] = (None, list(limit[name]['='].keys())[0], None)
            else:
                l[name] = (lim_min,  None, lim_max)
    return l


if __name__ == "__main__":
    import doctest
    sys.exit(doctest.testmod().failed)