classfinder.py

import json
import re
from operator import itemgetter
import requests

# finds the most similar list
# master = []
# candidates = [{}, {}, {},...]
def best_similarity(master, candidates):
    counts = [0] * len(candidates)
    for candidate_index, candidate in enumerate(candidates):
        for item in candidate:
            if item in master:
                counts[candidate_index] += 1

        # makes sure that candidates with a similar amount of items but less matches have some advantage
        # over candidates with more items and more matches
        # can be visualised as y = -x^2 + 100, where x = 0 is a perfect match
        counts[candidate_index] *= (-((len(candidate) - len(master)) ** 2) + 100)

    candidate_scores = zip(counts, candidates)
    # returns a map
    return max(candidate_scores, key=itemgetter(0))[1]

def main():

    old = requests.get("https://discord.com/assets/454210598d8beed381ab.js")._content.decode("utf8")
    new = requests.get("https://canary.discord.com/assets/782c119d5f878e08f08e.js")._content.decode("utf8")
    
    # jsonifies the source files
    replacements = [[r"\n", ""], [r"\},\d+.*?s=", ","], [r"\}{2,}.*", "}]"], [r"^.*\{,", "["], [r"(?<=\{|,)\b", "\""], [":", "\":"],["\"+", "\""]]
    for r in replacements:
        old = re.sub(r[0], r[1], old)
        new = re.sub(r[0], r[1], new)
    old = json.loads(old)
    new = json.loads(new)

    output = open("output.md", "w+")
    
    for element_map in old:
        match = best_similarity(element_map, new)

        for element in element_map:

            # checks if element contains a number as the first character, or a parenthesis
            # -> gets rid of elements such as "calc(...)" or "24px"
            if element in match and re.search(r"^\d|\(|\)", match[element]):
                print(f"Discarding {element}")
            else:
                try:
                    # splits elements with multiple classes into multiple lines
                    for substring in zip(element_map[element].split(' '), match[element].split(' ')):
                        output.write(f"{substring[0]} = {substring[1]}\n")
                    print(f"Found {element_map[element]}")
                except:
                    print(f"Couldn't find {element}")

    print("Finished")
    output.close()

main()

Example files that this could be used with:
https://cdn.discordapp.com/attachments/898335083944099880/929688339936866314/new_classes.json
https://cdn.discordapp.com/attachments/898335083944099880/929688340150779925/old_classes.json
Edit: additional files are no longer needed

how use

@Errorcrafter download the py file, open cmd, pip install requests

wait while it installs

python classfinder.py

enjoy.