import xml.etree.cElementTree as ET


class INLXmlParser:
    def __init__(self, lst, path):
        self.whitelist = lst
        self.xmlpath = path


    #expects to get a record as ElementTree
    def clearxml(self, record):
        # root ==  list of tags
        root = record.getroot()

        # scan the datafields in the records and copy to the new one only the tags in the whitelist
        # create new record
        newRecord = ET.Element('record')
        for field in root:
            fieldtag = field.attrib.get('tag')
            if fieldtag in self.whitelist:
                tempTag = fieldtag
                # tag 700 and 400 are the same
                if tempTag == '700':
                    tempTag = '400'
                for data in field:
                    newFieldTag = tempTag
                    newFieldTag +='.'
                    newFieldTag += data.attrib.get('code')
                    newTag = ET.SubElement(newRecord, 'datafield', {'tag': newFieldTag})
                    newTag.text = data.text

        newRecordTree = ET.ElementTree(newRecord)
        return newRecordTree