#!/usr/bin/env python
# version December 28, 2005
import sys,cPickle
# You will need to have the data file "latest.mdf", which is available for download
# at the tutorial. Or grab the fresh one at:
# http://www.mesonet.org/data/public/mesonet/latest/latest.mdf
# On Linux, you could run this script as: makePickle.py latest.mdf
# Normally, the sys.argv list contains command line arguements,
# which do not work in the IDLE of Windows.  You could UNCOMMENT this next line to
# simulate that command line arguments have been entered:
# sys.argv=['bogus','latest.mdf',2,None]

#set defaults:
fileopen=False
filename=None
sep=None
skiplines=2

print "sys.argv[0] always contains the name of the script:",sys.argv[0]

#try to get a filename from the command line:
try:
	filename=sys.argv[1]
except:
	print "filename was not entered from command line, so"
	filename=raw_input("enter filename=>")

#allow specification of number of lines to skip in the data file:
try:
	skiplines=int(sys.argv[2])
except:
	print "assuming",skiplines,"lines to be skipped"

#allow specification of a separator in the CSV file: 
try:
	sep=sys.argv[3]
except:
	print "The seperator in the CSV file will be assumed to be white space"

#try to open the file:
while not fileopen:
	try:
		inf=open(filename,'r')
		fileopen=True
	except:
		print "trouble opening",filename,",try another name ..."
		filename=raw_input("enter filename=>")

#process the CSV file into a dictionary of dictionaries:
all=inf.readlines()[skiplines:]
columntitles=all.pop(0)
ct=columntitles.split()
ct.pop(0)
h={}  
for line in all:
	if len(line)<1:
		print "end of file"
		break
	v=line.split()
	key=v.pop(0)
	h[key]={}
	assert len(v)==len(ct)
	for i in range(len(ct)):
		h[key][ct[i]]=float(v[i])
		
#Done making the dictionary of dictionaries, now inspect it:
mainkeys=h.keys()
print "\n here are the keys to the main dictionary:",mainkeys
samplekey=mainkeys[0]
subkeys=h[samplekey].keys()
print "\n using key",samplekey,", we find a dictionary with keys:", subkeys
subkey=subkeys[0]
print "\n using subkey",subkey,", here is a sample value:",h[samplekey][subkey] 

#now write a pickle file of the dictionary: 
oufn=filename+'.pickle'
print "\n will attempt to pickle this dictionary of dictionaries in",oufn
try:
	outfile=open(oufn,'w')
	cPickle.dump(h,outfile)
except:
	print "what a shame, no pickle file could be written"
print "All done!"