#script to convert features from SVM_light format to RF format (R randomForest package)
#created Joseph Luttrell 2/21/17
import sys
import os

svmFeats = sys.argv[1] #feature file in svm_light format
outputFeats = sys.argv[2] #output filename and path
numFeats = int(sys.argv[3]) #number of feature columns

svmFeatsFile = open(svmFeats, 'r')
outputFeatsFile = None
if not os.path.isfile(outputFeats):
	outputFeatsFile = open(outputFeats, 'w')
else:
	print ("output file exists, exiting")
	exit(1)

headerLine = 'Target,'

for i in range(1, int(numFeats)+1):
	if i == int(numFeats):
		headerLine += 'f' + str(i)
		break

	headerLine += 'f' + str(i) + ','

outputFeatsFile.write(headerLine + '\n')

for line in svmFeatsFile:
	items = line.split()
	vals = [] #will hold just the feature values
	targetVal = items[0]
	for item in items:
		if ':' in item:
			vals.append(item[(item.find(':'))+1:]) #get just the value from the featureLabel:value pair that svm format uses
		if item == '#':
			break

	valLine = targetVal + ','
	for i in range(0, len(vals)):
		if i == len(vals)-1:
			valLine += vals[i]
			break

		valLine += vals[i] + ','

	outputFeatsFile.write(valLine + '\n')