# convert datestring to proper format for numpy.loadtext()

strconverter = strpdate2num(fmt)

def bytesconverter(b):

s = b.decode(encoding)

return strconverter(s)

C = np.loadtxt('/tmp/domog/mysql/sql29.csv', delimiter=';', converters={0: bytespdate2num("%Y-%m-%d %H:%M:%S")})

# 2=13= windspeed (Gilze-Rijen)

# 3=14= winddirection (Gilze-Rijen)

# !=15= WindChill

Wspd = np.array(C[:, 2]) # windspeeds [m/s]

Wdir = np.array(C[:, 3]) # windvector [deg]

# convert dates in C[:,0] to something matplotlib understands

# D = matplotlib.dates.num2date(C[:, 0])

d2r = (1/360.) * np.pi * 2. # constant to convert from degrees to radians

ms2kmh = 3.6 # constant to convert from m/s to km/h

# do the conversions

Wdir[:] = [(x * d2r) for x in Wdir]

Wspd[:] = [(x * ms2kmh) for x in Wspd]

hrsmpls = 60 # data contains this number of samples per hour

# the graph will show one slice per hour # noqa

lenWdir = len(Wdir)

last14 = Wdir[lenWdir - 1]

last13 = Wspd[lenWdir - 1]

# create intermediate arrays

B13 = Wspd

B14 = Wdir

# make the array-lengths a multiple of <hrsmpls>

for x in range(hrsmpls - lenWdir % hrsmpls):

B13 = np.append(B13, last13)

B14 = np.append(B14, last14)

# Determine average speed and direction per 1-hour-period.

radii = theta = width = np.array([])

for x in range(0, lenWdir - 1, hrsmpls):

radii = np.append(radii, np.mean(B13[x:x+5]))

# Averaging of the bearings as per:

# http://rosettacode.org/wiki/Averages/Mean_angle

avg_theta = phase(sum(rect(1, d) for d in B14[x:x+hrsmpls-1])/hrsmpls)

if (avg_theta < 0):

avg_theta = avg_theta + (2 * np.pi)

theta = np.append(theta, avg_theta)

w = (np.pi - abs(np.max(B14[x:x+hrsmpls-1]) - np.min(B14[x:x+hrsmpls-1]) - np.pi))

width = np.append(width, w)

ahpla = 0.3

# bar plot on a polar axis.

# number of datapoints to show

N = len(radii)

ax = pl.subplot(111, polar=True)

# 0deg position at the top

ax.set_theta_zero_location("N")

# 90deg position to the right; show compass bearings

ax.set_theta_direction(-1)

bars = ax.bar(theta, radii, width=width, bottom=0.0)

# Use custom colors and opacity

for r, bar in zip(list(range(N)), bars):

bar.set_facecolor(pl.cm.hot((r / float(N))))

bar.set_alpha(ahpla)

# highlight the last bar (most recent value) by giving it a different color

bar.set_facecolor(pl.cm.cool(1.))

bar.set_alpha(1.)

# pl.title('Windroos')