import matplotlib.pyplot as plt
import numpy as np

import xanes_analyzeRun
import mcutils as mc

colors = "#a6cee3 #1f78b4 #b2df8a #33a02c #fb9a99 #e31a1c #fdbf6f #ff7f00 #cab2d6 #6a3d9a #ffff99 #b15928".split()

def myc(i):
  return colors[i%len(colors)]

def doRun(run=77):
  """ Since p1 is transformed, for the calibration look at p2 """
  r = xanes_analyzeRun.AnalyzeRun(run)
  r.load()
  calibs = list(r.results.keys())
  calibs.sort()
  p2 = [ np.nanmedian(r.results[c].p1,axis=0) for c in calibs ]
  p2 = np.asarray(p2)
  ref = np.nanmedian( p2 ,axis=0)
  # normalize p2
  p2 = p2/ref
  # remove non-flat part
  _x = np.arange(p2.shape[1],dtype=float)
  p2 = [s-mc.poly_approximant(_x,s,order=20)(_x) for s in p2]
  p2 = np.asarray(p2)
  x = np.asarray( [_x for _ in p2])
  x[p2>-0.15] = np.nan
  x[x<100] = np.nan; # below pixel 200 is noise
  pos = np.nanmean(x,axis=1)
  fig,ax=plt.subplots(1,2,sharey=True)
  xshift = 0.2
  pcalib = []
  for icalib,(xi,p2i) in enumerate(zip(x,p2)):
    ppar = dict ( color = myc(icalib) )
    ax[0].plot(p2i+icalib*xshift,_x,'--',lw=1,alpha=0.5,**ppar)
    ax[0].plot(p2i+icalib*xshift,xi,lw=2,**ppar)
    pcalib.append( np.nanmedian(xi) )
    ax[1].plot(r.scanpos[icalib],pcalib[icalib],'o',**ppar)
  polycal = np.polyfit(pcalib,r.scanpos,1)
  ax[1].plot(np.polyval(polycal,_x),_x)
  print("polynomial calibration",polycal)