Clean SVG

clean_svg.py

#!/usr/bin/env python

from argparse import ArgumentParser
from io import BytesIO
import re
from xml.etree import ElementTree


NAMESPACED_RE = re.compile(r'^\{(.+)\}.+$')
SVG_NAMESPACE = 'http://www.w3.org/2000/svg'
ALLOWED_NAMESPACES = {
    SVG_NAMESPACE: '',
    'http://www.w3.org/1999/xlink': 'xlink',
}
GROUP_TAG = '{%s}g' % SVG_NAMESPACE


def remove_useless_ids(root):
    all_elements = root.findall('.//')
    all_elements.append(root)
    elements_with_id = root.findall('.//*[@id]')
    elements_with_id.append(root)
    for element_with_id in elements_with_id:
        try:
            id_reference = '#' + element_with_id.attrib['id']
        except KeyError:
            continue
        for element in all_elements:
            if any(id_reference in value
                   for value in element.attrib.values()):
                break
        else:
            del element_with_id.attrib['id']


class UselessTag(Exception):
    pass


def clean_element(element, allowed_namespaces):
    match = NAMESPACED_RE.match(element.tag)
    if match is not None and match.group(1) not in allowed_namespaces:
        raise UselessTag

    for i, child in enumerate(tuple(element)):
        try:
            clean_element(child, allowed_namespaces)
            if child.tag == GROUP_TAG:
                if len(child) == 0:
                    raise UselessTag
                if len(child) == 1 and not child.attrib:
                    element.insert(i, tuple(child)[0])
                    raise UselessTag
        except UselessTag:
            element.remove(child)
    for k in tuple(element.attrib):
        match = NAMESPACED_RE.match(k)
        if match is not None and match.group(1) not in allowed_namespaces:
            del element.attrib[k]


def clean_svg(input_path, output_path=None,
              allowed_namespaces=ALLOWED_NAMESPACES):
    for allowed_namespace, prefix in allowed_namespaces.items():
        ElementTree.register_namespace(prefix, allowed_namespace)
    tree = ElementTree.parse(input_path)
    root = tree.getroot()

    clean_element(root, allowed_namespaces)
    remove_useless_ids(root)

    if output_path is None:
        with BytesIO() as temporary_file:
            tree.write(temporary_file, encoding='utf-8')
            return temporary_file.getvalue().decode()
    tree.write(output_path, encoding='utf-8')
    return 'Result saved in %s' % output_path


if __name__ == '__main__':
    parser = ArgumentParser(
        description='Cleans a SVG from useless tags and attributes.')
    parser.add_argument('input_path', type=str)
    parser.add_argument('output_path', type=str, nargs='?')
    print(clean_svg(**vars(parser.parse_args())))