#!/usr/bin/python
# Read seasonal snowfalls from the Web
import urllib
import html2text

# print "The web stie used by this Python script was taken down in Spring, 2016."
# print "This script no longer works because of this."
# sys.exit(1)

# The first html produces two frames.  We want the second frame.
# html = urllib.urlopen("http://www.wrcc.dri.edu/cgi-bin/cliMAIN.pl?ma0770").read()
# html = urllib.urlopen("http://www.wrcc.dri.edu/WRCCWrappers.py?sodxtrmts+190770+por+por+snow+none+msum+5+07+F").read()


html = urllib.urlopen("https://course.ccs.neu.edu/cs7600/parent/python/snowfall-boston.html").read()


#=======================================
# Read the html data into a text file.
if "1891-92" in html:  # if URL taken from noaa.gov or a copy of it
  delim = "1891-92"
  delim_end = "\n\n"
elif "1936" in html:   # else if URL taken from wrcc.dri.edu
  delim = "1936"
  delim_end = "\n2016"
  # Convert raw html to text file.
  html2text.BODY_WIDTH = 0 # The stupid html2text was writing '\n\n'
  html = html2text.html2text(html)
  # html2text was producing unicode.  ASCII is good enough in the US of A. :-)
  html = html.encode("ascii", "ignore")  # ignore non-ASCII characters
  # The stupid html2text was writing '\n\n'; Finish patching the output.
  html = html.replace('\n', '\t')
  html = html.replace('\t\t', '\n')
else:
  print "*** unknown format ***"

#=======================================
# Now process the html data, to get the record years.
data1 = delim + html.split(delim)[1]
data2 = data1.split(delim_end)[0]

a = data2.split('\n')
b = [ (x.split()[0], x.split()[-1]) for x in a ]
c = [ (float(x[-1]), x[0]) for x in b ]
d = sorted( c, key=lambda x: x[0] )
d.reverse() # Largest snowfalls first

print( "Ten largest snowfalls:\n" +
       '\n'.join( [str((x[1], x[0])) for x in d[:10]] ) )

#=======================================
# And now graph the variable c from above (snowfall in chronological order)

import matplotlib.pyplot

matplotlib.pyplot.xlabel("Year")
# The list c is: [ (46.8, '1891-92'), ... ]
# Pick out labels for [0:len(c)+1:10]  (from 0 to len(c), for every 10th elt)
matplotlib.pyplot.xticks( range(0, len(c), 10),
                          [x[1] for x in c][0:len(c)+1:10] )
locs, labels = matplotlib.pyplot.xticks()   # Get the labels field for x-axix
matplotlib.pyplot.setp(labels, rotation=45) # Rotate the labels 45 degrees
matplotlib.pyplot.axis(xmin=0)
matplotlib.pyplot.axis(xmax=len(c)+1)

matplotlib.pyplot.ylabel("Snowfall (in.)")

# This must be done after adjusting the x- and y-axes.
matplotlib.pyplot.tight_layout(w_pad=6, h_pad=6) # padding in width, height

matplotlib.pyplot.scatter(range(len(c)), [x[0] for x in c])  # scatter plot
matplotlib.pyplot.title("Seasonal snowfall in Boston")
matplotlib.pyplot.show()