"""Specialised plots for functions defined in blmfuncs.py.

fixme:
        -- Im normalsing all color mapping input vectors to [0,1]. This will
        destroy informative numerical values in colorbar (but we
        are not showing these anyway). A better fix would be to let the
        colorbar listen to the scalarmappable instance and corect itself, but
        I did not get that to work ...

fixme2:
        -- If scatterplot is not inited with a colorvector there will be no
        colorbar, but when adding colors the colorbar shoud be created.
"""
from fluents import plots
from scipy import dot,sum,diag,arange,log,mean,newaxis
from matplotlib import cm

class PcaScorePlot(plots.ScatterPlot):
    """PCA Score plot"""
    def __init__(self, model, absi=0, ordi=1):
        self._T = model.model['T']
        dataset_1 = model.as_dataset('T')
        dataset_2 = dataset_1
        id_dim = dataset_1.get_dim_name(0)
        sel_dim = dataset_1.get_dim_name(1)
        id_1, = dataset_1.get_identifiers(sel_dim, [absi])
        id_2, = dataset_1.get_identifiers(sel_dim, [ordi])
        plots.ScatterPlot.__init__(self, dataset_1, dataset_2, id_dim, sel_dim, id_1, id_2 ,c='b' ,s=40 , name='pca-scores')

    def set_absicca(self,n):
        self.xaxis_data = self._T[:,n]

    def set_ordinate(self,n):
        self.yaxis_data = self._T[:,n]

class PcaLoadingPlot(plots.ScatterPlot):
     """PCA Loading plot"""
     def __init__(self, model, absi=0, ordi=1):
         self._P = model.model['P']
         dataset_1 = model.as_dataset('P')
         dataset_2 = dataset_1
         id_dim = dataset_1.get_dim_name(0)
         sel_dim = dataset_1.get_dim_name(1)
         id_1, = dataset_1.get_identifiers(sel_dim, [absi])
         id_2, = dataset_1.get_identifiers(sel_dim, [ordi])
         if model.model.has_key('p_tsq'):
             col = model.model['p_tsq'].ravel()
             col = normalise(col)
         else:
             col = 'g'
         plots.ScatterPlot.__init__(self, dataset_1, dataset_2, id_dim, sel_dim, id_1, id_2,c=col,s=20, name='pls-loadings')

     def set_absicca(self,n):
         self.xaxis_data = self._P[:,n]

     def set_ordinate(self,n):
         self.yaxis_data = self._P[:,n]
    
class PlsScorePlot(plots.ScatterPlot):
    """PLS Score plot"""
    def __init__(self,model, absi=0, ordi=1):
        self._T = model.model['T']
        dataset_1 = model.as_dataset('T')
        dataset_2 = dataset_1
        id_dim = dataset_1.get_dim_name(0)
        sel_dim = dataset_1.get_dim_name(1)
        id_1, = dataset_1.get_identifiers(sel_dim, [absi])
        id_2, = dataset_1.get_identifiers(sel_dim, [ordi])
        
        plots.ScatterPlot.__init__(self, dataset_1, dataset_2,
                                   id_dim, sel_dim, id_1, id_2 ,
                                   c='b' ,s=40 , name='pls-scores')
        
    def set_absicca(self,n):
        self.xaxis_data = self._T[:,n]

    def set_ordinate(self,n):
        self.yaxis_data = self._T[:,n]


class PlsLoadingPlot(plots.ScatterPlot):    
    """PLS Loading plot"""
    def __init__(self,model,absi=0,ordi=1):
        self._P = model.model['P']
        dataset_1 = model.as_dataset('P')
        dataset_2 = dataset_1
        id_dim = dataset_1.get_dim_name(0)
        sel_dim = dataset_1.get_dim_name(1)
        id_1, = dataset_1.get_identifiers(sel_dim, [absi])
        id_2, = dataset_1.get_identifiers(sel_dim, [ordi])
        if model.model.has_key('w_tsq'):
            col = model.model['w_tsq'].ravel()
            col = normalise(col)
        else:
            col = 'g'
        plots.ScatterPlot.__init__(self, dataset_1, dataset_2,
                                   id_dim, sel_dim, id_1, id_2,
                                   c=col, s=20,                                   name='loadings')
        
    def set_absicca(self,n):
        self.xaxis_data = self._P[:,n]
        
    def set_ordinate(self,n):
        self.yaxis_data = self._T[:,n]

class LineViewXc(plots.LineViewPlot):
    """A line view of centered raw data
    """
    def __init__(self, func_class, name='Profiles'):
        # copy, center, plot
        x = func_class._dataset['X'].copy()
        x._array = x._array - mean(x._array,0)[newaxis]
        plots.LineViewPlot.__init__(self, x, 1, None, name)
        
class ParalellCoordinates(plots.Plot):
    """Parallell coordinates for score loads with many comp.
    """
    def __init__(self,model, p = 'loads'):
        pass

class PlsQvalScatter(plots.ScatterPlot):
    """A vulcano like plot of loads vs qvals
    """
    def __init__(self, func_class, pc=0):
        model = func_class.model
        if not model.has_key('w_tsq'):
            return
        self._W = model['P']
        dataset_1 = func_class.as_dataset('P')
        dataset_2 = func_class.as_dataset('w_tsq')
        id_dim = dataset_1.get_dim_name(0) #genes
        sel_dim = dataset_1.get_dim_name(1) #_comp
        sel_dim_2 = dataset_2.get_dim_name(1) #_zero_dim
        id_1, = dataset_1.get_identifiers(sel_dim, [0])
        id_2, = dataset_2.get_identifiers(sel_dim_2, [0])
        if model.has_key('w_tsq'):
            col = model['w_tsq'].ravel()
            col = normalise(col)
        else:
            col = 'g'
        plots.ScatterPlot.__init__(self, dataset_1, dataset_2,
                                   id_dim, sel_dim, id_1, id_2,
                                   c=col, s=20, sel_dim_2=sel_dim_2,
                                   name='Load Volcano')


class InfluencePlot(plots.ScatterPlot):
    """
    """
    pass
        

def normalise(x):
    """Scale vector x to [0,1]
    """
    x = x - x.min()
    x = x/x.max()
    return x