kruidvat.py

#!/usr/bin/env python3

import sys

from lxml import etree

def show_help():
    print("{} INPUT > OUTPUT".format(sys.argv[0]))


def parse_command():
    if len(sys.argv) < 2:
        show_help()
        sys.exit(0)

    convert(
        sys.argv[1]
    )



def convert(file_name):
    with open(file_name) as input_file:
        tree = etree.parse(input_file)
        root = tree.getroot()
        for element in root:
            is_chemist = False
            is_kruidvat = False
            ref = None
            for tag in element:
                key = tag.get('k')
                value = tag.get('v')
                if key == 'shop' and value == 'chemist':
                    is_chemist = True
                    continue
                if key == 'brand' and value == 'Kruidvat':
                    is_kruidvat = True
                    continue
                if key == 'ref' and value is not None:
                    ref = value.lstrip('0')
                    continue

            if is_chemist and is_kruidvat and ref is not None:
                for tag in element:
                    key = tag.get('k')
                    if key in ('website', 'brand:website', 'brand:wikipedia'):
                        element.remove(tag)
                
                fix_tags(element, ref)

        xml = etree.tostring(tree,
                             encoding="utf8",
                             xml_declaration=True,
                             pretty_print=True)
        print(bytes.decode(xml))
    
def fix_tags(element, ref):
    website_tag = etree.Element("tag", attrib={'k':'website', 'v':f'https://www.kruidvat.nl/store/drogisterij-kruidvat-{ref}'})
    brand_website_tag = etree.Element("tag", attrib={'k':'brand:website', 'v':f'https://www.kruidvat.nl'})
    brand_wikipedia_tag = etree.Element("tag", attrib={'k':'brand:wikipedia', 'v':f'nl:Kruidvat'})
    element.append(website_tag)
    element.append(brand_website_tag)
    element.append(brand_wikipedia_tag)

    element.set('action', 'modify')


if __name__ == '__main__':
    parse_command()