Find unused Script references in Unity - using meta GUIDs + naïve Regular Expressions

find_unused_references.py

import os, sys
import unityparser
import re

def get_files(folder_path):
    return [x for f in os.walk(folder_path) for x in map(lambda x: f"{f[0]}{os.path.sep}{x}", f[2])]

scripts = {d.data[0]['guid']: d for d in [unityparser.UnityDocument.load_yaml(path) for path in get_files(os.path.join('Assets', 'Scripts')) if path.endswith(".meta")]}
assets_paths = [path for path in get_files('Assets') if path.endswith('.unity') or path.endswith('.prefab') or path.endswith('.asset')]

i = 0
total = len(assets_paths)
found = set()

for path in assets_paths:
    i += 1
    if i % 10 == 0:
        print(f"{i} / {total} ({float(i) / total * 100}%)")
    try:
        doc = unityparser.UnityDocument.load_yaml(path)
    except UnicodeDecodeError:
        # some files aren't yamls...
        continue
    monos = doc.filter(class_names=('MonoBehaviour',))
    for mono in monos:
        guid = mono.m_Script['guid'] 
        if guid in scripts:
            print(f"Found dependency: {doc.file_path}->{scripts[guid].file_path}")
            found.add(guid)

not_found = set(scripts.keys()).difference(found)

# find (naive) references in all script files
names = list()

for guid in not_found:
    path = scripts[guid].file_path
    if not path.endswith(".cs.meta"):
        continue
        
    script_name = os.path.basename(path[:-len(".cs.meta")])
    names.append((guid, script_name))
    
names_re = [(guid, re.compile(f"(^|\W){name}($|\W)")) for guid, name in names]
    
# scan through scripts only
for script_guid, meta in scripts.items():
    if not meta.file_path.endswith(".cs.meta"):
        continue
    
    script_path = meta.file_path[:-len(".meta")]
    contents = open(script_path, "r").read()
    for guid, compiled in names_re:
        if script_guid == guid:
            continue
        if len(compiled.findall(contents)):
            print(f"Found naive reference: {script_path}->{compiled}")
            not_found.discard(guid)

sorted_not_found = sorted(not_found, key=lambda x: scripts[x].file_path)
for guid in sorted_not_found:
    path = scripts[guid].file_path
    if not path.endswith(".cs.meta"):
        continue
    
    script_path = path[:-len(".meta")]
    print(guid, script_path)