import xml.etree.cElementTree as ET


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

    def clearxml(self):
        xmltree = ET.parse(self.xmlpath)
        # root ==  list of records
        root = xmltree.getroot()

        # create new data
        newTreeRoot = ET.Element('data')

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

        newTree = ET.ElementTree(newTreeRoot)
        return newTree