sanatize.py

data = [
   {
      "reason":{
         "inventory_event_reason":"10",
         "reason_input_36":"see david's email",
         "reason_input_33":"",
         "reason_input_34":"Credit",
         "reason_input_30":"Wayne\u2019s Candies",
         "reason_input_31":"",
         "reason_input_35":"",
         "reason_input_29":"",
         "reason_input_32":""
      },
      "items":[
         {
            "ITEMID":"112996",
            "NAME":"TSC CROWN PEANUT LOG 3OZ 12CT-429",
            "LOCATION":"TN",
            "ONHAND":"224",
            "box_cost":"7.2",
            "item_status":"ACTIVE",
            "ADJUSTMENTTYPE":"adjustment",
            "ADJUSTMENT":"180"
         }
      ]
   },
   "jmatlock",
   "10",
   "Jeffrey Matlock <[email protected]>"
]

def sanatize( data ):
	def _sanatizeStr( dirty_string ):
		chars = {
            u'\u201c': '"', # RIGHT DOUBLE QUOTATION MARK
            u'\u201d': '"', # RIGHT DOUBLE QUOTATION MARK
            u'\u2018': "'", # LEFT SINGLE QUOTATION MARK
            u'\u2019': "'", # RIGHT SINGLE QUOTATION MARK
            u'\u2014': "-", # EM DASH
            u'\u2013': "-", # EM DASH
            u'\u02DC': "~", # SMALL TILDE
            u'\u201A': "'", # SINGLE LOW-9 QUOTATION MARK
            u'\u201E': '"'  # DOUBLE LOW-9 QUOTATION MARK
        }

        dirty_string = dirty_string.translate(None, string.punctuation).upper()

        for k, v in chars.iteritems():
            dirty_string = dirty_string.replace(k, v)

        return dirty_string

	def _sanatizeList( listData ):
		for subkey, subvalue in enumerate(listData):
			#print("Subkey %s, Subvalue %s" % (subkey, subvalue))
			if isinstance( subvalue, str ):
				listData[subkey] = _sanatizeStr( subvalue )
			else:
				listData[subkey] = sanatize( subvalue )

		return listData

	def _sanatizeDict( dictData ):
		for key, value in dictData.items():
			#print("Parsing key %s, value %s" % (key, value))
			if isinstance( value, str ):
				dictData[key] = _sanatizeStr( value )
			else:
				dictData[key] = sanatize( value )

		return dictData


	if isinstance( data, list ):
		return _sanatizeList( data )

	if isinstance( data, dict ):
		return _sanatizeDict( data )

	if isinstance( data, str ):
		return _sanatizeStr( data )

	return data

print "Before:"
print data

data = sanatize( data )

print "After:"
print data