Small script that loops through all of the files in your graphs asset folder, then searches your pages and journals files for references to them. If the asset file isn't referenced anywhere, it is deleted.

logseq-asset-clean-up.py

from multiprocessing.connection import wait
from os import listdir
import os
from os.path import isfile, join
from time import time


mygraphpath = "/Users/baileyjennings/Dropbox/Mac/Documents/baileys-mind-new-local"


# Get all the files in the assets folder
# capture file names as a list of strings

assets = []

for f in listdir(mygraphpath + "/assets"):
    if isfile(join(mygraphpath + "/assets", f)):
        assets.append(f)

# search the files text to see if is contain the asset file name
# if it does, list the asset file name in assetswithreference

assetswithreference = []

for asset in assets:
    for journal in listdir(mygraphpath+"/journals"):
        if asset in open(mygraphpath+ "/journals" + "/" + journal).read():
            assetswithreference.append(asset)
    for page in listdir(mygraphpath+"/pages"):
        if asset in open(mygraphpath+"/pages" + "/" + page).read():
            assetswithreference.append(asset)


# remove the assets that are not in the assetswithreference list from the assets folder

deleted_file_count = 0

for asset in assets:
    if asset not in assetswithreference:
        os.remove(mygraphpath + "/assets" + "/" + asset)
        deleted_file_count += 1

print("DONE :D")
print("total_files_deleted: " + str(deleted_file_count))