prdct_mstr.py

def all_same(items):
	return all(x == items[0] for x in items)

import openpyxl
import csv
wb = openpyxl.load_workbook('output_beautify.xlsx')

#wb.get_sheet_names()

sheet = wb.get_sheet_by_name('Sheet1')

row_count = sheet.max_row

col_count = sheet.max_column

#print row_count

#print col_count
f = open('output_master.csv','w')

# Add Title row on first line
for x in range(1,col_count+1):
	cell_title = str(sheet.cell(row=1, column=x).value)
	f.write(cell_title)
	
	if (x == col_count):
		f.write("\n")
	else:
		f.write(", ")
# end Title row

#initialize old_unique_identifier variable to check if it is a new product or product varient

old_unique_identifier = ""
smallArray = []
bigArray = []
colorArray = []
sizeArray = []
for i in range(2, row_count + 1):
	#select a unique idetifier for the product
	cur_unique_identifier = str(sheet.cell(row=i, column=4).value)
	#print ("cur = " + cur_unique_identifier)
	
	
	if(cur_unique_identifier != old_unique_identifier):
		#print("This is a new line")
		
		if bigArray:
			#print(bigArray)
			
			#clear size and color array
			colorArray = []
			sizeArray = []
			
			# fix the options as coloumn 12th (11th index) and 14th(13th index) of exel. So find variant and options
			#Color Array
			for j in range(len(bigArray)):
				colorArray.append(bigArray[j][11])
			
			#print(colorArray)
			#remove duplicate in array
			NewcolorArray = list(set(colorArray))
			
			#Size Array
			for j in range(len(bigArray)):
				sizeArray.append(bigArray[j][13])
				
			#print(sizeArray)
			#remove duplicate in array
			NewsizeArray = list(set(sizeArray))
		
			firstArray = bigArray[0]
			
			for q in range(len(firstArray)):
				texttoinsert = str(firstArray[q])
				f.write(texttoinsert + ",")
			
			
			#write features
			f.write('"' + ','.join(map(str, NewsizeArray)) + '",')
			f.write('"' + ','.join(map(str, NewcolorArray)) + '"')
			f.write("\n")
		
		#clear arrays
		bigArray = []
		smallArray = []

		for m in range(1, col_count + 1):
			#print(sheet.cell(row=1, column=m).value , sheet.cell(row=i, column=m).value)
			
			string = str(sheet.cell(row=i, column=m).value)
			smallArray.append(string)
		
	else:
		smallArray = []
		varien = False
		#print("Not a new line")
		
		for m in range(1, col_count + 1):
			#print(sheet.cell(row=1, column=m).value , sheet.cell(row=i, column=m).value)
			
			string = str(sheet.cell(row=i, column=m).value)
			smallArray.append(string)
			
			'''
		if (m == col_count):
			f.write(string + "\n")
		else:
			f.write(string + ", ")
	'''
	if smallArray:
		bigArray.append(smallArray)
	
	# Move cur_unique_identifier to old_unique_identifier
	old_unique_identifier = cur_unique_identifier
	#print("old = " + old_unique_identifier)
	
f.close()

'''

f = open('myfile','w')
f.write('hi there\n') # python will convert \n to os.linesep
f.close()

'''