Using BeautifulSoup to remove Adobe entity tags (and width/height attributes) from SVG

remove_adobe_crud.py

'''
When rendering SVG files exported from Adobe Illustrator using Phantom JS, 
the injected Adobe ENTITY tags can intefere with the render. This removes
the tags and also removes width/height attributes from the SVG tag.
'''


SVG_FILE = 'Some-adobe-illustrator-export.svg'

# By default, BeautifulSoup will ignore the <!ENTITY> array when parsing the file
# so we don't need to delete the array. It's not entirely safe, but works OK as a hack.
def sanitizeAdobeSVG(file):
    from bs4 import BeautifulSoup
    svg = BeautifulSoup(file, "xml")
    svg_tag = svg.find('svg')
    if 'width' in svg_tag.attrs:
        del svg_tag['width']
    if 'height' in svg_tag.attrs:
        del svg_tag['height']
    return svg

# Get the SVG XML
f = open(SVG_FILE, 'r')
svg_xml = f.read()

# Check for the tell-tale Adobe ENTITY tags
if '<!ENTITY' in svg_xml and 'ns.adobe.com' in svg_xml:
    print('Found Adobe-specific data')
    f = open(SVG_FILE, 'r')
    sanitized_svg = sanitizeAdobeSVG(f)
    f.close()
    svg_out = sanitized_svg.prettify("utf-8")
    with open('output.svg', 'wb') as file: # new file, but you can overwrite old one
        file.write(svg_out)
else:
    f.close()