# coding: utf-8

# In[1]:

#Collape the data!
#Plot I vs wavelength/frequency/both

get_ipython().magic(u'pylab inline')


# In[2]:

import numpy as np
import matplotlib.pyplot as pl
colors = ['m-', 'm-', 'b-', 'b-', 'g-', 'g-', 'y-', 'y-', 'r-', 'r-', 'k-']
colors2 = ['m--', 'm--', 'b--', 'b--', 'g--', 'g--', 'y--', 'y--', 'r--', 'r--', 'k--']


# In[3]:

#Functions
def normalize(array):
    #Normalize an array of data
    return array/np.max(array)

def collapse(array, time):
    #Collapse an array of data
    amid = len(array)/2 #Note that I purposefully left this an integer. You can't index an array with a float.
    upperh = array[amid:]
    uppert = time[amid:]
    lowerh = array[-amid:]
    lowert = time[0:amid] + amid #This will mostly be the same as uppert, but if an array contains an odd number of
                                 #points, you will be different and you will need this for plotting.
    
    #print shape(lowerh), shape(lowert)
    return upperh, lowerh, uppert, lowert


# In[4]:

#Classes
class IO:
    def __init__(self, filename):
        self.filename = filename
    
    def extract_data(self): #Don't bother with this function. It can't open csvs. It's meant for space seperated files.
        f = open(self.filename, "r")
    
        data = []
        for line in f:
            row = []
            for item in line.split():
                flt = True
                try:
                    row.append(float(item))
                except ValueError:
                    flt = False
                    #print "Line ", line, " contains values that are not floats and will be excluded."
            if flt == True:
                data.append(row)
    
        f.close()
    
        data = np.array(data)
        dataCols = data.T
        self.x = dataCols[0]
        self.y = dataCols[1]
        return dataCols[0], dataCols[1]
    
    def extract_csv(self): #This is the function you want.
        f = open(self.filename, "r")
    
        data = []
        for line in f:
            row = []
            for item in line.split(","):
                flt = True
                try:
                    row.append(float(item))
                except ValueError:
                    flt = False
                    """Commented out the print statements because this function gets looped 7 times and raises
                    exceptions each time since the first line in each file contains strings."""
                    #print "Line ", line, " contains values that are not floats and will be excluded."
                    #print "Item that raised this exception: ", item
            if flt == True:
                data.append(row)
    
        f.close()
    
        data = np.array(data)
        dataCols = data.T
        return dataCols

class Plot:
    def __init__(self, x, y, title="x vs y", xlabel="x", ylabel="y"):
        
        self.x = x
        self.y = y
        self.title = title
        self.xlabel = xlabel
        self.ylabel = ylabel
    
    def plot(self, lintype="b-"):
        # Dont use "k-" for your lintype. That is the type the axis uses.
        
        axis = np.zeros(len(self.x))
        pl.plot(self.x, axis, "k-")
        pl.plot(self.x, self.y, lintype)
        pl.title(self.title)
        pl.xlabel(self.xlabel)
        pl.ylabel(self.ylabel)
        pl.show()
        pl.clf()
        
    def overplot(self, y, lbl, lintype="b-", x=0):
        # Use this function if you plan on overplotting more than one data set.
        if type(x) != numpy.ndarray:
            x = self.x
        pl.plot(x, y, lintype, label=lbl)
        pl.title(self.title)
        pl.xlabel(self.xlabel)
        pl.ylabel(self.ylabel)
        pl.legend(loc='center left', bbox_to_anchor=(1, 0.5)) # This bbox_to_anchor setting will place the legend outside the plot.
        
    def scatter(self):
        axis = np.zeros(len(self.x))
        pl.plot(self.x, axis)
        pl.scatter(self.x, self.y)
        pl.title(self.title)
        pl.xlabel(self.xlabel)
        pl.ylabel(self.ylabel)
        pl.show()
        pl.clf()

    def overscatter(self, y, lbl, colors="b"):
        # Use this function if you plan on overplotting more than one data set.
        pl.scatter(self.x, y, color=colors, label=lbl)
        pl.title(self.title)
        pl.xlabel(self.xlabel)
        pl.ylabel(self.ylabel)
        pl.legend(loc='center left', bbox_to_anchor=(1, 0.5)) # This bbox_to_anchor setting will place the legend outside the plot.


# In[5]:

filenames = ["Max_Kourtney_160422_trial_1_cirrus.csv", "Max_Kourtney_160422_trial_2_clear.csv",              "Max_Kourtney_160422_trial_3_clear.csv", "Max_Kourtney_160422_trial_4_clear.csv",              "Max_Kourtney_160422_trial_5_clear.csv", "Max_Kourtney_160422_trial_6_clear.csv",
             "Max_Kourtney_160422_trial_7_cirrus.csv"]

trials = []
for item in filenames:
    #print item
    File = IO(item)
    data = File.extract_csv()
    trials.append(data)
    
#trials = np.array(trials)
wavelengths_nm = np.array([405.6, 441.9, 467.6, 493.7, 538.2, 610.4, 638.7, 654.9, 702.3, 750.5])

for i in xrange(len(trials)):
    s = trials[i][0]
    Title = "I vs time: Trial %i [Raw Data]" % (i+1)
    p = Plot(s, trials[i][1], title=Title, xlabel="Time [s]", ylabel="I")
    
    for j in xrange(len(trials[i]) - 1):
        p.overplot((trials[i][j+1]), lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])

    pl.show()
    pl.clf()


# In[6]:

for i in xrange(len(trials)):
    s = trials[i][0]
    Title = "I/I_max vs time: Trial %i [Raw Data]" % (i+1)
    p = Plot(s, trials[i][1], title=Title, xlabel="Time [s]", ylabel="I/I_max")
    
    for j in xrange(len(trials[i]) - 1):
        p.overplot(normalize(trials[i][j+1]), lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])

    pl.show()
    pl.clf()


# In[8]:

margin = 0.106

trials = []
for item in filenames:
    File = IO(item)
    data = File.extract_csv()
    trials.append(data)
    
#Here I want to try to put a mask on the data so that we only see the times when the sun is in the detector.
#For this I need to remove the data points where the normalized data is below a certain percentage given by "margin."
#The only way I could do this was by unpackaging the data, masking it, and repackaging it, so I needed this big, long loop.

newTrials = []
for i in xrange(len(trials)):
    for k in xrange(len(trials[i])):
        trials1 = []
        trials2 = []
        mins = []
        maxs = []
    
        for j in xrange(len(trials[i])-1):
            mins.append(np.min(np.where(normalize(trials[i][j+1]) > margin)[0]))
            maxs.append(np.max(np.where(normalize(trials[i][j+1]) > margin)[0]))
    
        lmin = np.min(np.array(mins))
        umax = np.max(np.array(maxs))
    
        dmin = np.linspace(0, lmin, lmin+1)
        dmax = np.linspace((umax-lmin), (len(trials[i][0])-lmin), int((len(trials[i][0])-lmin)-(umax-lmin)))
    
        for j in xrange(len(trials[i])):
            trials3 = np.delete(trials[i][j], dmin)
            trials4 = np.delete(trials3, dmax).tolist()
            trials2.append(trials4)
    
        trials1.append(trials2)
    newTrials.append(np.array(trials2))

#Now I take all the time arrays and set the start time back to zero
for i in xrange(len(newTrials)):
    newTrials[i][0] = newTrials[i][0] - np.min(newTrials[i][0])

for i in xrange(len(newTrials)):
    s = newTrials[i][0]
    Title = "I vs Time: Trial %i" % (i+1)
    p = Plot(s, newTrials[i][1], title=Title, xlabel="Time [s]", ylabel="I")
    
    for j in xrange(len(newTrials[i]) - 1):
        p.overplot((newTrials[i][j+1]), lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])

    pl.show()
    pl.clf()


# In[9]:

for i in xrange(len(newTrials)):
    s = newTrials[i][0]
    mu = np.sort(np.cos((90./np.max(newTrials[i][0]))*s - 90))
    Title = "I/I_max vs mu: Trial %i" % (i+1)
    p = Plot(mu, np.sort(normalize(newTrials[i][1])), title=Title, xlabel="mu", ylabel="I")
    #p = Plot(s, normalize(newTrials[i][1]), title=Title, xlabel="Time [s]", ylabel="I")
    
    for j in xrange(len(newTrials[i]) - 1):
        p.overplot((np.sort(normalize(newTrials[i][j+1]))), lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])

    pl.show()
    pl.clf()


# In[9]:

# Collapsed Data

for i in xrange(len(trials)):
    s = trials[i][0]
    #mu = timetomu(trials[i][0])
    Title = "Collapsed I vs time: Trial %i [Raw Data]" % (i+1)
    p = Plot(s, trials[i][1], title=Title, xlabel="Time [s]", ylabel="I")
    
    for j in xrange(len(trials[i]) - 1):
        up, lw, ut, lt = collapse(trials[i][j+1], s)
        p.overplot(up, x=ut, lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])
        p.overplot(lw, x=lt, lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors2[j])

    pl.show()
    pl.clf()


# In[13]:

for i in xrange(len(newTrials)):
    s = newTrials[i][0]
    #mu = timetomu(trials[i][0])
    Title = "Collapsed I vs time: Trial %i [Raw Data]" % (i+1)
    p = Plot(s, trials[i][1], title=Title, xlabel="Time [s]", ylabel="I")
    
    for j in xrange(len(trials[i]) - 1):
        up, lw, ut, lt = collapse(newTrials[i][j+1], s)
        p.overplot(up, x=ut, lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors[j])
        p.overplot(lw, x=lt, lbl=("%.1f nm" % wavelengths_nm[j]), lintype=colors2[j])

    pl.show()
    pl.clf()


# In[12]:

#As you can see, some of the collapsed data trends underneath the actual data, and some data sets don't.
#I'm not quite sure why, but my guess is passing clouds fudged the data since it isn't affecting all the sets.


# In[ ]: