Shopify 2 WooCommerce import CSV

shopify2woocommerce.py

# Usage: python3 shopify2woocommerce.py https://ABC.myshopify.com
# or for interactive mode, just: python3 shopify2woocommerce.py

import json
import os.path
import requests
from collections import OrderedDict
import pandas as pd
from sys import argv, exit

# Flags, accumulators
interactiveMode = False
collector = []

# create folders if they don't exist
if not os.path.exists('jsons'):
	os.makedirs('jsons')

##############################
# Functions:
def downloadFile(x,y):
	if os.path.exists(y) :
		# print('skipping '+ y + ' as its already been downloaded.')
		return
	print('\nDownloading '+x)

	try:
		r = requests.get(x, allow_redirects=True)

	except requests.exceptions.RequestException as e:  # from https://stackoverflow.com/a/16511493/4355695
		print('\nInvalid URL.')
		if interactiveMode: interactiveExit()
		else: exit()

	
	with open(y, 'wb') as f:
		f.write(r.content)
	print( 'Saved as '+ y )

def interactiveExit():
	print('\n\nPress any key to exit.\n')
	a = input()
	exit()


#############################
# Main Program Start:


# taking first argument as URL, or asking user to provide URL
if len(argv) > 1 :
	shopifyURL = argv[1]
else:
	print('\n\nEnter a shopify URL, like https://abc.myshopify.com:')
	shopifyURL = input()
	interactiveMode = True


downloadFile( shopifyURL + '/collections.json', 'jsons/collections.json' )

collections = json.load(open('jsons/collections.json')).get('collections',[]);

#########################################
# Looping through each collection

print('\n\nShopify API to WooCommerce-import-CSV converter.\n\nStarting to loop through collections')
for collection in collections:
	handle = collection.get('handle')
	category = collection.get('title')

	countCheck = collection.get('products_count')
	# https://ecoexist-enterprises.myshopify.com/collections/cloth-bags-15-x-17/products.json
	productJsonURL = 'https://ecoexist-enterprises.myshopify.com/collections/' + handle + '/products.json'
	productJson = 'jsons/' + handle + '.json'
	
	if countCheck: downloadFile( productJsonURL, productJson )
	else: continue

	products = json.load(open(productJson)).get('products',[])

	print( '\nCATEGORY: ' + category + ' has ' + str(len(products)) + ' products.')

	#########################################
	# Looping through each product entry in the collection
	for product in products:
		common = OrderedDict()
		numVariants = len( product.get('variants',[]))

		common['SKU'] = product.get('handle')

		#########################################
		# duplicates check, and if so, then just add the collection to categories and skip to next loop.
		repeatingSKU = False
		for x in collector:
			if x['SKU'] == common['SKU']:
				x['Categories'] += ', ' + category
				categorycount = len( x['Categories'].split(', ') )
				print( 'handle=' + x['SKU'] + ' encountered again, it now has ' + str(categorycount) + ' Categories.')
				repeatingSKU = True
				break
		if repeatingSKU:
			continue



		common['Name'] = product.get('title')
		common['Description'] = product.get('body_html').replace('\n','')
		common['Categories'] = category
		common['Tags'] = ', '.join( product.get('tags') )
		# Images:
		imageURLs = [ x.get('src') for x in product.get('images',[]) ] 
		common['Images'] = ', '.join(imageURLs)

		numOptions = len( product.get('options',[])[0].get('values'))
		if numOptions > 1: print('OPTIONS for product ' + common['SKU'] +' : ' + str(numOptions))

		#########################################

		# assumptions
		common['Is featured?'] = 0
		common['Stock'] = ''
		common['Backorders allowed?'] = 0
		common['Sold individually?'] = 0
		common['Length (in)'] = ''
		common['Width (in)'] = ''
		common['Height (in)'] = ''
		common['Allow customer reviews?'] = 0
		common['Stock'] = ''


		# defaults
		common['Published'] = 1
		common['Visibility in catalog'] = 'visible'

		row = OrderedDict(common) # making a copy.. see if this works
		
		#########################################
		# now, here we check for variants
		# firstly, if it's  a simple product with no variants, it'll have ONE variant in shopify.
		if (numVariants < 2):
			
			row['Type'] = 'simple'

			# print( product.get('variants')[0].get('taxable') )
			taxStatus =  product.get('variants')[0].get('taxable')
			# row['Tax status'] = ('taxable' if taxStatus else 'not taxable')
			# skipping tax status.. woocommerce didn't accept "not taxable"
			
			row['In stock?'] = 1 if product.get('variants')[0].get('available') else 0
			grams = product.get('variants')[0].get('grams') if ( row['Type'] == 'simple' ) else 0
			#row['Weight (lbs)'] = 0.00220462 * grams if ( grams > 0 ) else ''
			# not doing weight, importer isn't taking it.
				
			row['Regular price'] = product.get('variants')[0].get('price') if (numVariants == 1) else ''
			collector.append(row)
		

		#########################################
		# next, if there's variants then processing each variant and adding as a new entry to the table
		else: 
			print( 'VARIATIONS for product ' + row['SKU'] + ' : ' + str(numVariants) )
			
			# first, let's complete the parent row. Most fields will be left blank here as their variants step in.
			row['Type'] = 'variable'

			row['Attribute 1 name'] = product.get('options',[])[0].get('name')
			row['Attribute 1 value(s)'] = ', '.join( product.get('options',[])[0].get('values') )

			collector.append(row)

			# next, make a new row for each variation
			for variant in product.get('variants',[]):
				row2 = OrderedDict(common)
				row2['Type'] = 'variation'
				row2['SKU'] = common['SKU'] + '-' + variant.get('title').lower()
				# have to over-write the parent's SKU with a new one. This was what was causing a duplicating issue for all variants. Shopify didn't give 

				row2['Name'] = common['Name'] + ' - ' + variant.get('title')
				# have to pre-pend the parent's title to variant's title to get full title. 
			
				row2['In stock?'] = 1 if variant.get('available') else 0
				grams = variant.get('grams')
				#row2['Weight (lbs)'] = 0.00220462 * grams if ( grams > 0 ) else ''
				# not doing weight, importer isn't taking it.

				row2['Regular price'] = variant.get('price') if (numVariants == 1) else ''

				# if there's a featured image, add at beginning of images list
				if variant.get('featured_image'):
					featured = variant.get('featured_image').get('src')
					if featured in imageURLs:
						# if the featured image is there in the list, remove it so it can be inserted at beginning without repeating.
						imageURLs.remove(featured)
					imageURLs.insert(0,featured)
					common['Images'] = ', '.join(imageURLs)
				collector.append(row2)
		
df = pd.DataFrame(collector)
df.to_csv('woocommerce-import.csv', index=False)

print('\n\nProcessed ' + shopifyURL + '. Total ' + str(len(df)) + ' products found.')
print('Created woocommerce-import.csv in the same folder where you ran this script.')

if interactiveMode:
	interactiveExit()