plot_session.py

# -*- coding: utf-8 -*-
"""
Created on %(date)s

@author: %(username)s
"""

#import cv
import bisect
import itertools
import numpy as np
import matplotlib.dates as mdates
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import scipy.cluster.hierarchy as hcl
from scipy.stats import norm
import analysis_utilities as utils
#import image_processing as imgproc
import signal_processing as signalproc
import process_session
from rasterplot import rasterplot
import matplotlib.gridspec as gridspec
import matplotlib.cbook as cbook
from mpl_toolkits.mplot3d import Axes3D

max_height_cm = 24.0
height_pixel_to_cm = max_height_cm / 680.0
width_pixel_to_cm = 50.0 / 1280.0
rail_height_pixels = 100
frames_per_second = 120.0

protocol_colors = {
    'stable':'g',
    'centerfree':'r',
    'stabletocenterfree':'orange',
    'centerfreetostable':'orange',
    'randomizedcenterfree_day1':'b',
    'randomizedcenterfree_day2':'b',
    'randomizedcenterfree_day3':'b',
    'randomizedcenterfree_day4':'b',
    'degradationrandomizedcenterfree_day1':'b',
    'degradationrandomizedcenterfree_day2':'b',
    'permutationfreepair_day1':'violet',
    'permutationfreepair_day2':'violet',
    'permutationfreepair_day3':'violet',
    'permutationfreepair_day4':'violet',
    'permutationtofullyreleased':'purple',
    'fullyreleased':'purple'
}

protocol_artists = [
    plt.Rectangle((0,0),1,1,fc='g'),
    plt.Rectangle((0,0),1,1,fc='orange'),
    plt.Rectangle((0,0),1,1,fc='r'),
    plt.Rectangle((0,0),1,1,fc='b'),
    plt.Rectangle((0,0),1,1,fc='violet'),
    plt.Rectangle((0,0),1,1,fc='purple'),
]

protocol_labels = [
    'stable',
    'transition',
    'center unstable',
    'center random',
    'random permutation',
    'all released',
]

def get_protocol_color(label):
    return protocol_colors.get(label,'g')
    
def shade_session_protocols(sessions,session_xdata,axis=None,plotzero=True):
    if axis is None:
        axis = plt.gca()    
    
    colors = []
    session_labels = [session.session_labels.get('protocol') for session in sessions]
    base_x = 0
    prev_x = 0
    prev_color = None
    for i,(label,x) in enumerate(itertools.chain(zip(session_labels,session_xdata),[(None,0)])):
        if label is not None:
            color = get_protocol_color(label)
        if label is None or prev_color != color:
            if prev_color is not None:
                axis.axvspan(base_x,prev_x,facecolor=prev_color,alpha=0.5,lw=0,edgecolor='none')
                colors.append(color)
            base_x = prev_x
        #if i > 0 or plotzero:
        prev_x = x
        prev_color = color
    return colors

def click_data_action(figure,ondataclick):
    def onclick(event):
        if event.button == 3 and event.xdata is not None and event.ydata is not None:
            ondataclick(event)
    figure.canvas.mpl_connect('button_press_event',onclick)
    
def get_color_cycle(data,colormap=plt.cm.jet):
    return [colormap(i) for i in np.linspace(0, 0.9, len(data))]
    
def get_alternating_color_cycle(data,colors):
    return [colors[i % len(colors)] for i in range(len(data))]
    
def alternating_color_map(data,colors):
    plt.gca().set_color_cycle(get_alternating_color_cycle(data,colors))

def time_color_map(data,colormap=plt.cm.jet):
    plt.gca().set_color_cycle(get_color_cycle(data,colormap))

def plot_time_var(trials):
    time_color_map(trials)
    for trial in trials:
        plt.plot(trial)
        
def plot_time_distributions(name,sessions):
    plt.figure(name + ' spatial time distribution')
    plt.hist(utils.flatten([utils.flatten(process_session.get_clipped_trajectories(session)) for session in sessions]),100)
    plt.xlabel('horizontal progression (pixels)')
    plt.ylabel('total time (frames)')
    
def iter_poly_fit(x,y,order):
    done = False
    coeffs = None
    while not done:
      done = True
      coeffs = np.polyfit(x,y,order)
      fit = np.poly1d(coeffs)
      residuals = y - fit(x)
      std = np.std(residuals)
      badindices = np.where(residuals > 2*std)[0]
      if badindices.size > 0:
        done = False
        x = np.delete(x, badindices)
        y = np.delete(y, badindices)
    return coeffs
    
def plot_epoch_fit(data,deg=15):
    def _polynomial(x, *p):
        """Polynomial fitting function of arbitrary degree."""
        poly = 0.
        for i, n in enumerate(p):
            poly += n * x**i
        return poly

    data = utils.flatten([x for x in data])
    indices = range(len(data))
    coeffs = iter_poly_fit(indices,data,deg)
    polyline = np.poly1d(coeffs)
    yfit = polyline(indices)
    plt.plot(indices,yfit)
    
    
# Iterative poly fit
#    offset = 0
#    for epoch in data:
#        epochlen = len(epoch)
#        indices = range(offset,offset+epochlen)
#        coeffs = iter_poly_fit(indices,epoch,deg)
#        polyline = np.poly1d(coeffs)
#        yfit = polyline(indices)
        
# Polynomial curve fit
#        indices = np.array(range(epochlen))
#        p0 = np.ones(deg,)
#        sigma = np.std(epoch)
#        coeff, var_matrix = curve_fit(_polynomial, indices, np.array(epoch), p0=p0, sigma=sigma)
#        yfit = [_polynomial(xx, *tuple(coeff)) for xx in indices]
#        indices = range(offset,offset+epochlen)
#        offset += epochlen
#        plt.plot(indices,yfit)
        
def plot_epoch_average(data,label=None,offset=0,scale=1):
    mean = [np.mean(epoch) for epoch in data]
    std = [np.std(epoch) for i,epoch in enumerate(data)]
    ax = plt.gca()
    ax.xaxis.set_major_locator(ticker.IndexLocator(1,0))
    c = next(ax._get_lines.color_cycle)
    x = (np.arange(len(data))*scale)+offset
    plt.plot(x,mean,'--',zorder=0,color=c)
    plt.errorbar(x,mean,std,fmt='o',label=label,zorder=100,color=c)
    plt.xlim(-0.5-offset/2,(len(data)*scale)-0.5)
        
def plot_end_to_end(data):
    offset = 0
    for epoch in data:
        epochlen = len(epoch)
        indices = range(offset,offset+epochlen)
        offset += epochlen
        plt.plot(indices,epoch,'.')
        
def plot_end_to_end_xy(data):
    offset = 0
    for x,y,epochlen in data:
        x += offset
        offset += epochlen
        plt.plot(x,y,'o',markersize=3)
        
def bar_end_to_end(data,colorcycle):
    offset = 0
    rects = []
    for i in range(len(data)):
        epoch = data[i]
        epochlen = len(epoch)
        indices = range(offset,offset+epochlen)
        offset += epochlen
        rects.append(plt.bar(indices,epoch,facecolor=colorcycle[i]))
    return rects
        
def plot_crossing_times(session,fmt='.'):
    plt.figure(session.name + ' crossing times')
    plt.plot(process_session.get_first_crossing_trial_times(session),fmt)
    
def plot_average_crossing_times(merged):
    plt.figure(merged.name + ' average crossing times')
    trial_miu = []
    trial_err = []
    for session in merged.merged_sessions:
        trial_times = process_session.get_first_crossing_trial_times(session)
        miu = np.mean(trial_times)
        sigma = np.std(trial_times)
        trial_miu.append(miu)
        trial_err.append(sigma)
    plt.bar(range(len(merged.merged_sessions)), trial_miu, yerr = [np.zeros(len(trial_err)),trial_err])
    
def plot_average_tip_height_all_conditions(name,sessions):
    plot_average_tip_height_light_trials(name,sessions)
    plot_average_tip_height_light_change(name,sessions)
    plot_average_tip_height_direction_trials(name,sessions)
    
def plot_average_tip_height_light_trials(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_height_two_conditions(name,'light','dark',sessions,lambda s:s.light_trials,crop=crop)
    
def plot_average_tip_height_light_change(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_height_two_conditions(name,'change','no change',sessions,lambda s:np.insert(np.diff(s.light_trials),0,1),crop=crop)
    
def plot_average_tip_height_direction_trials(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_height_two_conditions(name,'left','right',sessions,lambda s:np.array(s.crossing_direction),crop=crop)
    
def plot_average_tip_height_two_conditions(name,condition1,condition2,sessions,selector,crop=process_session.default_crop):
    figurename = "%s %s/%s" % (name,condition1,condition2)
    plot_average_tip_height_end_to_end(figurename,sessions,selector,colors='r',crop=crop)
    ax = plt.gca()
    line = ax.lines[0]
    line.set_label(condition1)
    
    fig = plot_average_tip_height_end_to_end(figurename,sessions,lambda s:~selector(s),colors='b',invert_y=False,crop=crop)
    ax = plt.gca()
    line = ax.lines[-1]
    line.set_label(condition2)
    plt.legend()
    plt.xlabel('crossings (single animal)')
    return fig
    
def plot_average_tip_height_end_to_end(name,sessions,conditionselector=lambda x:None,colors=['b','r'],invert_y=True,crop=[0,1280],legend=True):
    fig = plt.figure(name + ' average tip height')
    alternating_color_map(sessions,colors)
    def scale_height(avg):
        avg[1] = max_height_cm - ((avg[1] + rail_height_pixels) * height_pixel_to_cm)
        return avg
    average_height = [scale_height(process_session.get_average_crossing_tip_height(session,conditionselector(session),crop)) for session in sessions]
    plot_end_to_end_xy(average_height)
    shade_session_protocols(sessions,np.cumsum([epochlen for x,y,epochlen in average_height]))
    plt.xlabel('crossings')
    plt.ylabel('average nose height above steps (cm)')
    plt.title('average height of the nose during successive crossings')
    if legend:
        plt.legend(protocol_artists,protocol_labels,bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)    
    return fig
    
def plot_average_tip_speed_light_trials(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_speed_two_conditions(name,'light','dark',sessions,lambda s:s.light_trials,crop=crop)
    
def plot_average_tip_speed_light_change(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_speed_two_conditions(name,'change','no change',sessions,lambda s:np.insert(np.diff(s.light_trials),0,1),crop=crop)
    
def plot_average_tip_speed_direction_trials(name,sessions,crop=process_session.default_crop):
    return plot_average_tip_speed_two_conditions(name,'left','right',sessions,lambda s:np.array(s.crossing_direction),crop=crop)    
    
def plot_average_tip_speed_two_conditions(name,condition1,condition2,sessions,selector,crop=process_session.default_crop):
    figurename = "%s %s/%s" % (name,condition1,condition2)
    plot_average_tip_speed_end_to_end(figurename,sessions,selector,colors='r',crop=crop)
    ax = plt.gca()
    line = ax.lines[0]
    line.set_label(condition1)
    
    fig = plot_average_tip_speed_end_to_end(figurename,sessions,lambda s:~selector(s),colors='b',crop=crop)
    ax = plt.gca()
    line = ax.lines[-1]
    line.set_label(condition2)
    plt.legend()
    plt.xlabel('crossings (single animal)')
    return fig
        
def plot_average_tip_speed_end_to_end(name,sessions,conditionselector=lambda x:None,colors=['b','r'],crop=process_session.default_crop):
    fig = plt.figure(name + ' average tip speed')
    alternating_color_map(sessions,colors)
    def scale_speed(avg):
        avg[1] = (avg[1] * width_pixel_to_cm) * frames_per_second
        return avg
    average_speed = [scale_speed(process_session.get_average_crossing_tip_speed(session,conditionselector(session),crop)) for session in sessions]
    plot_end_to_end_xy(average_speed)
    shade_session_protocols(sessions,np.cumsum([epochlen for x,y,epochlen in average_speed]))
    plt.xlabel('crossings')
    plt.ylabel('average tip speed (cm / s)')
    plt.title('average speed of the nose during successive crossings')
    return fig
    
def plot_trial_times_end_to_end(name,sessions,conditionselector=lambda x:None):
    fig = plt.figure(name + ' trial times')
    #time_color_map(sessions,plt.cm.jet)
    trial_time_color = get_alternating_color_cycle(sessions,['b'])
    effective_time_color = get_alternating_color_cycle(sessions,['r'])
    effective_trial_times = [process_session.get_first_crossing_trial_times(session,conditionselector(session)) for session in sessions]
    trial_times = [[i.total_seconds() for i in session.inter_reward_intervals] for session in sessions]
    rects_total = bar_end_to_end(trial_times,trial_time_color)
    rects_effective = bar_end_to_end(effective_trial_times,effective_time_color)
    boundaries = process_session.get_boundary_indices(trial_times)
    if len(boundaries) > 0:
        ylim = plt.gca().get_ylim()
        plt.vlines(boundaries,ylim[0],ylim[1])
    
    plt.xlabel('trials (single animal)')
    plt.ylabel('time to reward (s)')
    plt.title('session trial times')
    plt.legend([rects_total[0][0],rects_effective[0][0]],('total time','from first crossing'))
    
    # Interactive video playback feature
    def click_playback(figure,sessions):
        def ondataclick(event):
            x = int(round(event.xdata))
            datasession = None
            time = None
            trial_total = 0
            for i in range(len(sessions)):
                len_session = len(trial_times[i])
                print x,len_session,trial_total
                if x < trial_total + len_session:
                    datasession = sessions[i]
                    x = x - trial_total - 1
                    if x >= 0:
                        time = datasession.reward_times[x]
                    else:
                        time = datasession.start_time
                    break
                trial_total = trial_total + len_session
            
            pos_msec = process_session.get_time_video_pos_msec(datasession,time)
            fps = 120
            video = '\\..\\top_video.avi'
            if event.key == 'f':
                video = '\\..\\front_video.avi'
            imgproc.play_video(datasession.path[0] + video,datasession.name + ' ' + str(pos_msec) + 'msec',pos_msec,fps)
        click_data_action(figure,ondataclick)
    click_playback(fig,sessions)
    return fig
    
def plot_effective_trial_times_end_to_end(name,sessions,conditionselector=lambda x:None,colors=['b','r'],shade_sessions=True):
    fig = plt.figure(name + ' effective trial times')
    alternating_color_map(sessions,colors)
    effective_trial_times = [process_session.get_first_crossing_trial_times(session,conditionselector(session)) for session in sessions]
    plot_end_to_end(effective_trial_times)
    if shade_sessions:
        shade_session_protocols(sessions,process_session.get_all_boundary_indices(effective_trial_times))
    plt.xlabel('trials (single animal, session colored)')
    plt.ylabel('time to reward (s)')
    plt.title('interval between start of first crossing and poke head entry')
    return fig
    
def plot_effective_trial_times_in_time(name,sessions,conditionselector=lambda x:None,colors=['b','r'],shade_sessions=True):
    fig = plt.figure(name + ' effective trial times in time')
    alternating_color_map(sessions,colors)
    
    def get_trial_times_in_time(start_stop_times,session):
        x = np.array([(reward_time - session.start_time).total_seconds() / 60 for start_time,reward_time in start_stop_times])
        y = np.array([(reward_time - start_time).total_seconds() for start_time,reward_time in start_stop_times])
        epochlen = process_session.get_session_duration(session).total_seconds() / 60
        return x,y,epochlen
    
    effective_trial_times_in_time = [get_trial_times_in_time(process_session.get_first_crossing_start_stop_times(session,conditionselector(session)),session)
    for session in sessions]
    plot_end_to_end_xy(effective_trial_times_in_time)
    if shade_sessions:
        cumulative_session_time = np.cumsum([epochlen for x,y,epochlen in effective_trial_times_in_time])
        shade_session_protocols(sessions,cumulative_session_time)
    plt.xlabel('time in assay (min)')
    plt.ylabel('time to reward (s)')
    plt.title('interval between start of first crossing and poke head entry')
    return fig
    
def plot_smooth_trial_times(name,sessions,**kwargs):
    fig = plt.figure(name + ' smooth trial times')
    effective_trial_times = [process_session.get_first_crossing_trial_times(session) for session in sessions]
    trial_time_series = np.array([x for x in cbook.flatten(effective_trial_times)])
    smooth_series = signalproc.smooth(trial_time_series,kwargs.pop('window_len',15),kwargs.pop('window','blackman'))
    plt.plot(smooth_series)
    return fig
    
def plot_average_trial_times(name,sessions,label=None,offset=0,scale=1,makefig=True):
    if makefig:
        fig = plt.figure(name + ' average trial times')
    trial_times = [[i.total_seconds() for i in session.inter_reward_intervals]
                   if len(session.inter_reward_intervals) > 0
                   else [process_session.get_session_duration(session).total_seconds()]
                   for session in sessions]
    plot_epoch_average(trial_times,label,offset,scale)
    if makefig:
        plt.xlabel('sessions')
        plt.ylabel('time between rewards (s)')
        #plt.title('average trial time across sessions')
        return fig
    
def plot_average_reward_rate(name,sessions):
    fig = plt.figure(name + ' average reward rate')
    reward_rates = [[60.0 / i.total_seconds() for i in session.inter_reward_intervals] for session in sessions]
    plot_epoch_average(reward_rates)
    plt.xlabel('sessions')
    plt.ylabel('reward rate (trials / min)')
    plt.title('average reward rate across sessions')
    return fig
    
def plot_average_effective_trial_times(name,sessions,conditionselector=lambda x:None):
    fig = plt.figure(name + ' average effective trial times')
    effective_trial_times = [process_session.get_first_crossing_trial_times(session,conditionselector(session)) for session in sessions]
    plot_epoch_average(effective_trial_times)
    plt.xlabel('sessions')
    plt.ylabel('time to reward (s)')
    plt.title('average interval between start of first crossing and reward')
    return fig
    
def plot_average_tip_speed(name,sessions,conditionselector=lambda x:None,crop=process_session.default_crop,label=None,offset=0,scale=1,trial_slices=slice(None)):
    fig = plt.figure(name + ' average tip speed across sessions')
    def scale_speed(avg):
        avg[1] = (avg[1] * width_pixel_to_cm) * frames_per_second
        return avg
    average_speed = [scale_speed(process_session.get_average_crossing_tip_speed(session,conditionselector(session),crop,trial_slices))[1] for session in sessions]
    plot_epoch_average(average_speed,label,offset,scale)
    plt.xlabel('sessions')
    plt.ylabel('average tip speed (cm / s)')
    plt.title('average speed of the tip of the nose across sessions')
    return fig
    
def plot_average_tip_height(name,sessions,conditionselector=lambda x:None,crop=process_session.default_crop,label=None,offset=0,scale=1):
    fig = plt.figure(name + ' average tip height across sessions')
    height_samples = [utils.flatten([process_session.get_clipped_trial_variable(session,i,trial,crop) for i,trial in enumerate(session.tip_vertical)]) for session in sessions]
    plot_epoch_average(height_samples,label,offset,scale)
    plt.xlabel('sessions')
    plt.ylabel('average tip height (pixels)')
    plt.title('average height of the tip of the nose across sessions')
    return fig
    
def plot_min_trial_times(name,sessions):
    fig = plt.figure(name + ' min trial times')
    sorted_session_trial_times = [np.sort(session.inter_reward_intervals) for session in sessions]
    min_trial_times = [(sorted_trial_times[0].total_seconds() if sorted_trial_times.any() else None) for sorted_trial_times in sorted_session_trial_times]

    valid_times = [min_trial_time if min_trial_time else 0 for min_trial_time in min_trial_times]
    valid_faces = plt.bar(range(len(sessions)),valid_times)
    
    ylim = plt.gca().get_ylim()[1]
    invalid_times = [0 if min_trial_time else ylim for min_trial_time in min_trial_times]
    if np.sum(invalid_times) > 0:
        invalid_faces = plt.bar(range(len(sessions)),invalid_times,color='r')
        plt.legend([valid_faces[0],invalid_faces[0]],('rewarded','no rewards'))
    else:
        plt.legend([valid_faces[0]],['rewarded'])
    
    plt.xlabel('sessions')
    plt.ylabel('time to reward (s)')
    plt.title('minimum trial time across sessions')
    return fig
        
def plot_progression(session):
    plt.figure(session.name + ' progression')
    time_color_map(session.tip_horizontal)
    for i in range(len(session.tip_horizontal)):
        normalized = np.array(session.tip_horizontal[i])
        invalid_indices = normalized == -1
        if i in session.left_crossings:
            normalized = 1078 + normalized * -1
        normalized[invalid_indices] = np.nan
        plt.plot(normalized)
        
def plot_progression_path(session):
    plt.figure(session.name + ' progression')
    time_color_map(session.tip_horizontal_path)
    for path in session.tip_horizontal_path:
        normalized = np.array(path)
        plt.plot(normalized)
    plt.xlabel('time (frames)')
    plt.ylabel('progression (pixels)')
        
def plot_progression_delta_path(session):
    plt.figure(session.name + ' progression delta')
    time_color_map(session.tip_horizontal_path)
    for path in session.tip_horizontal_path:
        normalized = np.array(path)
        plt.plot(np.diff(normalized))
    plt.xlabel('time (frames)')
    plt.ylabel('progression delta (pixels)')

def plot_tip_spatial_speed_trial(session,i):
    plt.figure(session.name + ' ' + session.session_type + ' speed')
    speed_spline,speed,decision = process_session.get_tip_spatial_speed_trial(session,i)
    if decision:
        plt.plot(speed_spline)
    
def plot_tip_spatial_speed(session):
    plt.figure(session.name + ' ' + session.session_type + ' speed')
    data = []
    trial_map = {}
    for i in range(len(session.tip_horizontal)):
        print i,len(data)
        speed_spline,speed,decision = process_session.get_tip_spatial_speed_trial(session,i)
        if decision:
            trial_map[len(data)] = i
            data.append(speed_spline)
    im = plt.imshow(data,vmin=0,vmax=10)
    plt.colorbar(im)
    return trial_map
    
def plot_spatial_variable_interpolation_split(fig,session,selector,vmin=None,vmax=None):
    data = []
    adjustprops = dict(left=0.1, bottom=0.1, right=0.97, top=0.93, wspace=0.2, hspace=0.2)
    fig.subplots_adjust(**adjustprops)
    for i in range(len(session.tip_horizontal)):
        value = selector(session,i)
        data.append(value)
        
    if session.session_type == 'merge':
        sessions = session.merged_sessions
    else:
        sessions = [session]
        
    root = None
    for i in range(len(sessions)):
        if root is None:
            ax = fig.add_subplot(len(sessions),1,i + 1)
            for label in ax.get_xticklabels():
                label.visible = False
            root = ax
        else:
            ax = fig.add_subplot(len(sessions),1,i + 1, sharex=root, sharey=root)
        
        im = plt.imshow(data,vmin=vmin,vmax=vmax)
        plt.colorbar(im)

    [plt.axvline(step-100,linewidth=2,color='r') for step in process_session.roi_step_centers]        
    plt.xlabel('horizontal position (pixels)')
    plt.ylabel('trials')
    #plt.xticks(np.array(process_session.roi_step_centers)-100)
    
def plot_spatial_variable_interpolation(fig,session,selector,vmin=None,vmax=None):
    data = []
    for i in range(len(session.tip_horizontal)):
        value = selector(session,i)
        data.append(value)
        
    im = plt.imshow(data,vmin=vmin,vmax=vmax)
    cb = plt.colorbar(im)
    [plt.axvline(step-process_session.interpolation_range[0],linewidth=2,color='k') for step in process_session.roi_step_centers]
    if session.session_type == 'merge':
        for i,boundary in enumerate(np.insert(np.cumsum([len(s.trial_time) for s in session.merged_sessions][:-1]),0,0)):
            if i == 0:
                continue
            color = 'k'
            if session.merged_sessions[i].session_type == 'manipulation':
                color = 'r'
            #plt.axhline(boundary,linewidth=5,color='b')
            #annotation = session.merged_sessions[i].name + ' ' + session.merged_sessions[i].session_type
            im.axes.annotate('', xy=(0,boundary), xycoords='data',
                         xytext=(-50,0),textcoords='offset points',
                         arrowprops=dict(facecolor=color,arrowstyle='wedge'))
            
            #im.axes.annotate(annotation, xy=(0,boundary), xycoords='data',
            #             xytext=(-130,-4),textcoords='offset points')
        
        #[plt.axhline(boundary,linewidth=2,color='b') for boundary in np.cumsum([len(s.trial_time) for s in session.merged_sessions][:-1])]

    #im.axes.set_yticklabels(labels)
    #im.axes.get_yaxis().set_ticks([])
    plt.xlabel('horizontal position (pixels)')
    plt.ylabel('crossings')
    plt.xticks(np.array(process_session.roi_step_centers)-process_session.interpolation_range[0],np.array(process_session.roi_step_centers))
    
    # Interactive video playback feature
    def click_playback(figure,session,selector = lambda y:y):
        def ondataclick(event):
            y = int(round(event.ydata))
            datasession = session
            if session.session_type == 'merge':
                trial_total = 0
                for i in range(len(session.merged_sessions)):
                    len_session = len(session.merged_sessions[i].trial_time)
                    if y < trial_total + len_session:
                        datasession = session.merged_sessions[i]
                        y = y - trial_total
                        break
                    trial_total = trial_total + len_session
            
            y = selector(y)
            pos_msec = process_session.get_trial_video_pos_msec(datasession,y)
            imgproc.play_video(datasession.path[0] + '\\..\\front_video.avi',datasession.name,pos_msec,0)
        click_data_action(figure,ondataclick)
    click_playback(fig,session)
    return cb
    
def plot_tip_spatial_height_interp(session,vmin=600,vmax=700):
    fig = plt.figure(session.name + ' position tip height ')
    cb = plot_spatial_variable_interpolation(fig,session,(lambda s,i:process_session.get_spatial_tip_height_trial(s,i)),vmin,vmax)
    plt.title('distribution of nose tip height with position')
    if cb is not None:
        cb.set_label('tip height (pixels)')
        cb.ax.invert_yaxis()
    return fig
    
def plot_tip_spatial_speed_interp(session,vmin=0,vmax=20):
    fig = plt.figure(session.name + ' position tip speed')
    cb = plot_spatial_variable_interpolation(fig,session,(lambda s,i:process_session.get_tip_horizontal_speed_trial(s,i)),vmin,vmax)
    plt.title('distribution of nose tip speed with position')
    cb.set_label('tip speed (pixels / frame)')
    return fig

def plot_tip_trajectory(session,i):
    plt.figure(session.name + ' ' + session.session_type + ' trajectory')
    x,y,decision = process_session.get_tip_trajectory_trial(session,i)
    if decision:
        plt.plot(x,y)
    ax = plt.gca()
    ax.set_ylim(ax.get_ylim()[::-1])
    
def plot_tip_trajectories(session,crop=[0,1280]):
    lplot = None
    rplot = None
    fig = plt.figure(session.name + ' ' + session.session_type + ' trajectory')
    for i in range(1,len(session.tip_horizontal)-1):
        x,y,rdir = process_session.get_tip_trajectory_trial(session,i,crop)
        #plt.plot(x,y)
        if rdir:
            rplot, = plt.plot(x,y,'b')
        else:
            lplot, = plt.plot(x,y,'r')
   
    #[plt.axvline(center,linewidth=2,color='k') for center in np.array(process_session.roi_step_centers)]
    plt.xlabel('horizontal position (pixels)')
    plt.ylabel('vertical position (pixels)')
    plt.title('tracked nose tip trajectories')
#    plt.xticks(np.array(process_session.roi_step_centers))
    plt.legend([lplot,rplot],('left','right'),loc=4)
                
    ax = plt.gca()
    ax.set_ylim(ax.get_ylim()[::-1])
    return fig
    
def plot_tip_trajectories_3d(session,crop=[0,1280]):
    lplot = None
    rplot = None
    fig = plt.figure(session.name + ' ' + session.session_type + ' trajectory')
    ax = fig.add_subplot(111, projection = '3d')
    for i in range(1,len(session.tip_horizontal)-1):
        x,y,rdir = process_session.get_tip_trajectory_trial(session,i,crop)
        #plt.plot(x,y)
        if rdir:
            rplot, = plt.plot(x,range(len(x)),y,'b')
        else:
            lplot, = plt.plot(x,range(len(x)),y,'r')

    ax.invert_zaxis()
    ax.invert_yaxis()
    #[plt.axvline(center,linewidth=2,color='k') for center in np.array(process_session.roi_step_centers)]
    ax.set_xlabel('horizontal position (pixels)')
    ax.set_ylabel('time (frames)')
    ax.set_zlabel('vertical position (pixels)')
    plt.title('tracked nose tip trajectories')
#    plt.xticks(np.array(process_session.roi_step_centers))
    plt.legend([lplot,rplot],('left','right'),loc=4)
                
    ax = plt.gca()
    ax.set_ylim(ax.get_ylim()[::-1])
    return fig
        
def plot_latency_curves(name,sessions):
    plt.figure(name + ' cumulative latency')
    time_color_map(sessions)
    for session in sessions:
        latency = np.cumsum([interval.total_seconds() for interval in session.inter_reward_intervals])
        plt.plot(latency)
        for i in range(len(latency)):
            plt.plot(i,latency[i],'k|')
    plt.xlabel('trials')
    plt.ylabel('latency (s)')
    plt.title('cumulative latency')
    
def plot_step_dynamics(name,sessions,step,selector):
    title = name + ' step %s aligned dynamics' % step
    plt.figure(title)
    step_tip_trials = [process_session.get_step_aligned_data(selector(session),session,step,120,120,False)[1] for session in sessions]
    im = plt.imshow(utils.flatten(step_tip_trials))
    plt.colorbar(im)
    plt.xlabel('time (frames)')
    plt.ylabel('trials')
    plt.title(title)
    
    trial = 0
    for i,session in enumerate(step_tip_trials):
        im.axes.annotate(sessions[i].name + ' ' + sessions[i].session_type, xy=(0,trial), xycoords='data',
                         xytext=(-350,0),textcoords='offset points',
                         arrowprops=dict(arrowstyle="->"))
        trial += len(session)
    
def plot_step_tip_dynamics(name,sessions,step):
    plot_step_dynamics(name + ' horizontal tip',sessions,step,lambda s:s.tip_horizontal)
    
def plot_step_tip_height_dynamics(name,sessions,step):
    plot_step_dynamics(name + ' vertical tip',sessions,step,lambda s:s.tip_vertical)
    
def plot_step_activity_dynamics(name,sessions,step):
    plot_step_dynamics(name + ' step activity',sessions,step,lambda s:s.step_activity[step])
    
def plot_step_tip_progression(name,sessions):
    plt.figure(name + ' average step aligned tip position')
    step_tip_trials = [[utils.meanstd(process_session.get_step_aligned_data(session.tip_horizontal,session,step)[0]) for session in sessions] for step in range(6)]
    [plt.errorbar(range(len(session_data)),[mean for mean,std in session_data],[std for mean,std in session_data]) for session_data in step_tip_trials]
    
def plot_step_tip_trials_end_to_end(name,sessions):
    plt.figure(name + ' step aligned tip position')
    time_color_map(sessions,plt.cm.hsv)
    step_tip_trials = [[process_session.get_step_aligned_data(session.tip_horizontal,session,step)[0] for session in sessions] for step in range(6)]
    [plot_end_to_end(step_aligned_data) for step_aligned_data in step_tip_trials]
    plt.xlabel('trials')
    plt.ylabel('tip position (pixels)')
    
def plot_step_tip_height_trials_end_to_end(name,sessions,step):
    plt.figure(name + ' step aligned tip height')
    time_color_map(sessions,plt.cm.spectral)
    step_tip_trials = [process_session.get_step_aligned_data(session.tip_horizontal,session,step)[0] for session in sessions]
    #step_tip_trials = [utils.nanmean(process_session.get_step_aligned_data(session.tip_horizontal,session,step)[0]) for session in sessions]
    plot_end_to_end(step_tip_trials)
    #plt.plot(step_tip_trials)
    plt.xlabel('trials')
    plt.ylabel('tip height (pixels)')
    
def plot_step_tip_position(name,sessions,steps):
    plt.figure(name + ' step aligned tip position')
    gs = gridspec.GridSpec(2,2,width_ratios=[8,1],height_ratios=[1,8],wspace = 0.05,hspace = 0.05)
    #time_color_map(sessions,plt.cm.spectral)
    plt.subplot(gs[2])
    alternating_color_map(sessions,['black','orange'])
    step_tip_trials_x = [[process_session.get_step_aligned_data(session.tip_horizontal,session,step)[0] for session in sessions] for step in steps]
    step_tip_trials_y = [[process_session.get_step_aligned_data(session.tip_vertical,session,step)[0] for session in sessions] for step in steps]
    
    step_tip_trial_sessions_x = []
    step_tip_trial_sessions_y = []
    for i in range(len(step_tip_trials_x)):
        for s in range(len(step_tip_trials_x[i])):
            for k in range(len(step_tip_trials_x[i][s])):
                if s >= len(step_tip_trial_sessions_x):
                    step_tip_trial_sessions_x.append([])
                    step_tip_trial_sessions_y.append([])
                step_tip_trial_sessions_x[s].append(step_tip_trials_x[i][s][k])
                step_tip_trial_sessions_y[s].append(step_tip_trials_y[i][s][k])
    
    [plt.plot(step_tip_trial_sessions_x[i],step_tip_trial_sessions_y[i],'.') for i in range(len(step_tip_trial_sessions_x))]
    plt.xlabel('horizontal position (pixels)')
    plt.ylabel('vertical position (pixels)')
    plt.title('nose tip positions aligned on step events (single animal)')
    plt.xticks(np.array(process_session.roi_step_centers))
    plt.legend([session.session_type for session in sessions])
    ax = plt.gca()
    ax.set_ylim(ax.get_ylim()[::-1])
    
    plt.subplot(gs[0])
    alternating_color_map(sessions,['black','orange'])
    
    step_tip_trial_sessions_x = [utils.removenan(session) for session in step_tip_trial_sessions_x]
    step_tip_trial_sessions_y = [utils.removenan(session) for session in step_tip_trial_sessions_y]
    minx = np.min([np.min(session) for session in step_tip_trial_sessions_x])
    maxx = 1075.0
    miny = np.min([np.min(session) for session in step_tip_trial_sessions_y])
    maxy = np.max([np.max(session) for session in step_tip_trial_sessions_y])
    
    bins = np.linspace(minx, maxx, 100)
    [plt.hist(step_tip_trial_sessions_x[i],100,range=(minx,maxx),alpha=1-(i*0.5),histtype='stepfilled') for i in range(len(step_tip_trial_sessions_x))]
    plt.xticks([])
    plt.yticks([])
    plt.subplot(gs[3])
    alternating_color_map(sessions,['black','orange'])
    bins = np.linspace(miny, maxy, 100)
    [plt.hist(step_tip_trial_sessions_y[i],bins,alpha=1-(i*0.5),orientation='horizontal',histtype='stepfilled') for i in range(len(step_tip_trial_sessions_y))]
    ax = plt.gca()
    plt.xticks([])
    plt.yticks([])
    ax.set_ylim(ax.get_ylim()[::-1])
    #plt.xlim(0,14)
    plt.xlim(0,40)
    #return step_tip_trial_sessions_x
    
def plot_step_average_variable(name,sessions,selector):
    plt.figure(name + ' step aligned tip height')
    time_color_map(sessions,plt.cm.spectral)
    
    step_tip_trials = [[process_session.get_step_aligned_data(selector(session),session,step)[0] for session in sessions] for step in range(6)]    
    step_tip_trials = [[np.ma.masked_array(session,np.isnan(session)) for session in step] for step in step_tip_trials]

    plt.boxplot(step_tip_trials[2])
    #plt.bar(range(8),step_tip_trials[2])
#    [plot_end_to_end(step_aligned_data) for step_aligned_data in step_tip_trials]
    plt.xlabel('trials')
    plt.ylabel('tip height (pixels)')
    
def plot_step_average_tip_height(name,sessions):
    plt.figure(name + ' step aligned tip height')
    time_color_map(sessions,plt.cm.spectral)
    
    step_tip_trials = [[process_session.get_step_aligned_data(session.tip_vertical,session,step)[0] for session in sessions] for step in range(6)]    
    step_tip_trials = [[np.ma.masked_array(session,np.isnan(session)) for session in step] for step in step_tip_trials]

    plt.boxplot(step_tip_trials[2])
    #plt.bar(range(8),step_tip_trials[2])
#    [plot_end_to_end(step_aligned_data) for step_aligned_data in step_tip_trials]
    plt.xlabel('trials')
    plt.ylabel('tip height (pixels)')
        
def plot_step_tip_distribution(session,stepindex):
    plt.figure("%s step %s tip distribution" % (session.name, stepindex))
    plt.hist(process_session.step_tip_distribution(session,stepindex))
        
def plot_average_intensity(name,mean):
    plt.figure(name + 'mean')
    plot_time_var(mean)
    plt.xlabel('time (frames)')
    plt.ylabel('average intensity')
    plt.title('trial-by-trial pixel average')
    ax = plt.gca()
    ylim = ax.get_ylim()
    ylim = (2,ylim[1])
    ax.set_ylim(ylim)
    
def plot_step_activity(session,step):
    plt.figure(session.name + ' step %s activity' % (step))
    [plt.plot(activity) for activity in session.step_activity[step]]
    plt.xlabel('time (frames)')
    plt.ylabel('roi activity (pixels)')
    plt.title('step activity')
    
def plot_step_trials(name,steps):
    plt.figure(name + 'steps')
    rasterplot(steps[0],'b.')
    rasterplot(steps[1],'g.')
    rasterplot(steps[2],'r.')
    rasterplot(steps[3],'c.')
    rasterplot(steps[4],'m.')
    rasterplot(steps[5],'y.')
    ax = plt.gca()
    xlim = ax.get_xlim()
    xlim = (xlim[0],700)
    ylim = ax.get_ylim()[::-1]
    ylim = (ylim[0],1)
    ax.set_ylim(ylim)
    ax.set_xlim(xlim)
    plt.xlabel('time (frames)')
    plt.ylabel('trials')
    plt.title('trial-by-trial step sequence')
    
    plt.figure(name + 'first step distributions')
    x = np.linspace(xlim[0],xlim[1])
    steps = [process_session.index_distribution(steps,slice(1)) for steps in steps]
    for distribution in steps:
        dmean = np.mean(distribution)
        dstd = np.std(distribution)
        plt.plot(x,norm.pdf(x,loc=dmean,scale=dstd))
    ax = plt.gca()
    ylim = ax.get_ylim()
    ax.set_ylim(ylim)
    plt.xlabel('time (frames)')
    plt.ylabel('probability')
    
    plt.figure(name + 'first step cumulative distributions')
    for distribution in steps:
        hist,bins=np.histogram(distribution,bins=100)
        cumulative = np.cumsum(hist)
#        width=0.7*(bins[1]-bins[0])
        center=(bins[:-1]+bins[1:])/2
        plt.plot(center,cumulative)
    
#def plot_step_probability(session):
#    plt.figure(session.name + 'step probability')
#    plt.bar(range(len(session.steps)),process_session.step_probabilities(session))
    
def plot_step_probability(name,sessions,colorcycle=None):
    width = 0.9 / (len(sessions) + 1)
    ind = np.arange(6)
    fig = plt.figure(name + ' skip probability')
    ax = fig.add_subplot(111)
    if colorcycle is None:
        colorcycle = get_color_cycle(sessions,plt.cm.spectral)
        
    barplots = []
    for i in range(len(sessions)):
        barplots.append(plt.bar(ind + i * width, process_session.step_probabilities(sessions[i]), width, color = colorcycle[i]))
        
    ax.set_xticks(ind + 0.5 * width * len(sessions))
    ax.set_xticklabels( ('1', '2', '3', '4', '5', '6') )
    
    display = (0,4)
    ax.legend([barplots[i] for i in display],('stable','manipulation'),bbox_to_anchor=(0., 1.02, 1., .102), loc=3,
       ncol=2, mode="expand", borderaxespad=0.)
    
    #ax.legend(('black','red'))
    #ax.legend(barplots,[session.name for session in sessions],bbox_to_anchor=(0, 0, 1, 1), bbox_transform=fig.transFigure)
    #plt.xlabel('step skipped\n(single animal across all sessions)')
    #plt.ylabel('probability of skipping a step')
    #plt.title('probability of skipping each step')

def plot_session(session):
    plot_average_intensity(session.name,session.mean)
    plot_step_trials(session.name,session.steps)
    plot_step_probability(session)

def plot_speed_raster(name,trials,trial_slice=slice(0,140)):
    speed_trials = process_session.get_tip_horizontal_speed_trials(trials,trial_slice)
    plt.figure(name + 'speed raster')
    plt.imshow(speed_trials,vmin=0,vmax=10)
    
def plot_spatial_speed_distribution(trials):
    m = process_session.get_spatial_speed_distribution(trials)
    bins = m[0][1][1:]
    avg_speeds = (m[1][1:])/(m[2][1:])
    plt.bar(bins[0:len(bins)-1],avg_speeds[0:len(avg_speeds)-1],10)
    
def plot_distance_statistics(name,distances,clusters):
    plt.figure(name + ' distance matrix')
    plt.imshow(distances)
    
    plt.figure(name + ' distance distribution')
    plt.hist(utils.flatten(distances),100)
    
    plt.figure(name + ' cluster dendrogram')
    hcl.dendrogram(clusters)
    
def plot_session_activity(session):
    legend_plots = []
    legend_labels = []
    
    fig = plt.figure(session.name + ' session activity')
    ax1 = fig.add_subplot('111')
    legend_plots.append(ax1.plot_date(mdates.datestr2num(session.left_poke[1]),session.left_poke[0] - np.min(session.left_poke[0]),'y')[0])
    legend_labels.append('Left Poke')
    
    legend_plots.append(ax1.plot_date(mdates.datestr2num(session.right_poke[1]),session.right_poke[0] - np.min(session.right_poke[0]),'k')[0])
    legend_labels.append('Right Poke')
    plt.xlabel('Time')
    plt.ylabel('Poke Activation (a.u.)')
    
    if session.front_activity is not None:
        ax2 = ax1.twinx()
        activity_time = mdates.datestr2num([session.frame_time[i] for i in range(len(session.front_activity))])
        legend_plots.append(ax2.plot_date(activity_time,session.front_activity, 'g')[0])
        legend_labels.append('Front Activity')
        plt.ylabel('Front motion (pixels/frame)')

    ylim = ax1.get_ylim()    
    if len(session.left_rewards) > 0:
        left_rewards = mdates.datestr2num([session.left_poke[1][bisect.bisect_left(session.left_poke[1],reward)] for reward in session.left_rewards])
        legend_plots.append(ax1.vlines(left_rewards,ylim[0],ylim[1],'r'))
        legend_labels.append('Rewards')

    if len(session.right_rewards) > 0:
        right_rewards = mdates.datestr2num([session.right_poke[1][bisect.bisect_left(session.right_poke[1],reward)] for reward in session.right_rewards])
        ax1.vlines(right_rewards,ylim[0],ylim[1],'r')
    
    ax2.legend(tuple(legend_plots),tuple(legend_labels))
    plt.title('left/right poke activation')
    return fig
    
def plot_poke_statistics(name,sessions):
    barwidth = 0.35
    fig = plt.figure(name + ' poke activation')
    left_pokes = map(list, zip(*[utils.meanstd(session.left_poke[0][session.left_poke[0] > 200]) for session in sessions]))
    right_pokes = map(list, zip(*[utils.meanstd(session.right_poke[0][session.right_poke[0] > 200]) for session in sessions]))
    
    ax = fig.add_subplot(111)
    indices = np.arange(len(sessions))
    left_rects = ax.bar(indices, left_pokes[0], barwidth, color='r', yerr=left_pokes[1])
    right_rects = ax.bar(indices+barwidth, right_pokes[0], barwidth, color='y', yerr=right_pokes[1])

    ax.set_xlabel('Sessions')
    ax.set_ylabel('Activation (a.u.)')
    ax.set_title('Reward-triggered poke activation values')
    ax.set_xticks(indices+barwidth)
    ax.set_xticklabels( [session.name for session in sessions] )
    ax.legend( (left_rects[0], right_rects[0]), ('Left Poke', 'Right Poke') )
    
    def autolabel(rects):
        # attach some text labels
        for rect in rects:
            height = rect.get_height()
            ax.text(rect.get_x()+rect.get_width()/2., 1.05*height, '%d'%int(height),
                    ha='center', va='bottom')
                
    autolabel(left_rects)
    autolabel(right_rects)
    return fig
    
def plot_average(name,images):
    mean = imgproc.average_image_list(images)
    cv.NamedWindow(name,cv.CV_WINDOW_NORMAL)
    cv.ShowImage(name,mean)
    del mean
    
def plot_image_tiles(name,tiles,mosaicscale=2):
    if len(tiles) > 0:
        mosaic_size = mosaicscale * np.array(cv.GetSize(tiles[0]))
        mosaic = cv.CreateImage(mosaic_size,tiles[0].depth,tiles[0].channels)
        ntiles = np.ceil(np.sqrt(len(tiles)))
        tile_scale = mosaicscale / ntiles
        tile_size = (int(tiles[0].width * tile_scale),int(tiles[0].height * tile_scale))
        for i in xrange(len(tiles)):
            image = tiles[i]
            xtile = int(i % ntiles)
            ytile = int(i / ntiles)
            roi = (int(xtile * tile_size[0]),int(ytile * tile_size[1]),tile_size[0],tile_size[1])
            cv.SetImageROI(mosaic,roi)
            cv.Resize(image,mosaic)
        cv.ResetImageROI(mosaic)
        cv.NamedWindow(name,cv.CV_WINDOW_NORMAL)
        cv.ShowImage(name,mosaic)
        del mosaic
        
def plot_posture_clusters(session,images,clusters,t):
    fclusters = hcl.fcluster(clusters,t,'maxclust')
    img_clusters = process_session.split_clusters(images,fclusters)
    for i in xrange(len(img_clusters)):
        plot_image_tiles(session.name + ' Cluster%s' % (i), img_clusters[i])
    average_clusters = [imgproc.avgstd_image_list(imgs) for imgs in img_clusters]
    
    plot_image_tiles(session.name + ' Cluster Means and Stds',utils.flatten(zip([avgstd[0] for avgstd in average_clusters],[avgstd[1] for avgstd in average_clusters])))
    
    plt.figure(session.name + 'Cluster Std Comparison')
    plt.bar(range(len(average_clusters)),[cv.Sum(img[0])[0] for img in average_clusters])
    del average_clusters