download-svgs.py

import hashlib
import os
import shutil
from collections import defaultdict
from glob import glob
from subprocess import check_call

import pyquery


def md5(s):
    return hashlib.md5(s.encode('utf-8')).hexdigest()[:7]


shutil.rmtree('downloaded')
os.makedirs('downloaded')

doc = pyquery.PyQuery('https://developer.mozilla.org/en-US/')
counts = defaultdict(list)
for svg in doc('svg').items():
    name = str(repr(svg))[2:-2]
    html = svg.outer_html()
    name += '_' + md5(html)
    if name not in counts:
        fn = 'downloaded/' + name + '.svg'
        with open(fn, 'w') as f:
            f.write(html)
            f.write('\n')
        check_call(("svgo {} -o {}".format(fn, fn.replace('.svg','.min.svg'))).split())

    counts[name].append(len(html))

for x in glob('downloaded/*.min.svg'):
    check_call(("zopfli -i100 {}".format(x)).split())
    check_call(("brotli -9 {}".format(x)).split())

def kb(b):
    return '{:.1f}KB'.format(b / 1024)

# print(counts)
print("TOTAL BYTES BEFORE ({}): {:,} ({})".format(
    sum(len(x) for x in counts.values()),
    sum(sum(x) for x in counts.values()),
    kb(sum(sum(x) for x in counts.values()))
))

after = {
    x: os.stat(x).st_size
    for x in glob('downloaded/*.min.svg')
}
# print(after)
print("TOTAL BYTES AFTER ({}): {:,} ({})".format(
    len(after),
    sum(after.values()),
    kb(sum(after.values()))
))


gzipped = {
    x: os.stat(x).st_size
    for x in glob('downloaded/*.min.svg.gz')
}
# print(gzipped)
print("TOTAL BYTES ZOPFLI ({}): {:,} ({})".format(
    len(gzipped),
    sum(gzipped.values()),
    kb(sum(gzipped.values()))
))

brotlied = {
    x: os.stat(x).st_size
    for x in glob('downloaded/*.min.svg.br')
}
# print(brotlied)
print("TOTAL BYTES BROTLI ({}): {:,} ({})".format(
    len(brotlied),
    sum(brotlied.values()),
    kb(sum(brotlied.values()))
))

Context; this blog post: https://www.peterbe.com/plog/optimize-inlined-svg-on-developer.mozilla.org