# imgfile = StringIO(binary_jpeg)
# img = Image.open(imgfile)
# img = img.convert('L')
# print str(list(img.getdata()))
# -------------------------------------
# 4- transform first image JPEG binary to RGB matrix and output
# pymongo_spark.activate()
# bsonFileRdd = sc.BSONFileRDD(sys.argv[1])
# first_item = bsonFileRdd.take(1)
# rdd_first_item = sc.parallelize(first_item)
# rdd_result = rdd_first_item.map(lambda x: ( str(x['_id'])+"_1", str(x['category_id']), str(list(Image.open(StringIO(x['imgs'][0]['picture'])).convert('L').getdata()))))
# rdd_result.saveAsTextFile(sys.argv[2])
# -------------------------------------
# 5- transform all bson, only 1st image of each product
# pymongo_spark.activate()
# bsonFileRdd = sc.BSONFileRDD(sys.argv[1])
# rdd_result = bsonFileRdd.map(lambda x: ( str(x['_id'])+"_1", str(x['category_id']), str(list(Image.open(StringIO(x['imgs'][0]['picture'])).convert('L').getdata()))))
# rdd_result.saveAsTextFile(sys.argv[2])
# -------------------------------------
# 6- transform all bson, all images of each product
pymongo_spark.activate()
bsonFileRdd = sc.BSONFileRDD(sys.argv[1])
rdd_result = bsonFileRdd.flatMap(
lambda x: [(str(x['_id']) + "_" + str(ind), str(x['category_id']),
jpeg_binary_to_grayscale_array(y['picture']))
for ind, y in enumerate(x['imgs'])])
rdd_result.saveAsTextFile(sys.argv[2])
