Check for imagens not used or referenced in HTML or CSS files in a site project

check-used-images.py

# Recursively check for images in a directory and compare with referenced images in CSS and HTML files
# in tha folder, generating two CSV files:
# One with used images and one with not referenced ones
# (does not look into JS files)

import glob
import csv

# Create the list of file names
pasta = "src/"						#root of the project source files
pastaIMG = "src/img/"			#root of images folder
buscaPNG = "**/*.png"
buscaJPG = "**/*.jpg"
buscaGIF = "**/*.gif"

pngs = glob.glob(pastaIMG + buscaPNG, recursive = True)
jpgs = glob.glob(pastaIMG + buscaJPG, recursive = True)
gifs = glob.glob(pastaIMG + buscaGIF, recursive = True)
htmls = glob.glob(pasta + "**/*.html", recursive = True)
csss = glob.glob(pasta + "**/*.css", recursive = True)

imagens = pngs + jpgs + gifs # join all image lists into one
imagens[:] = [i[4:] for i in imagens] # remove o /src from the path, must be changed for other names
fontes = htmls + csss


# Search files looking for references for the image
imgsUsadas = []
for arquivo in fontes:
    fonte = open(arquivo, "r")
    for l in fonte:
        for img in imagens:
            if img in l:
                imgsUsadas.append(img)
                imagens.remove(img)

# Write lists to csv files 
arqImgsUsadas = "site-used-images.csv"
arqImgsNaoUsadas = "site-images-NOT-referenced.csv"

with open(arqImgsUsadas, "w") as output:
    writer = csv.writer(output)
    for val in imgsUsadas:
        writer.writerow([val])

with open(arqImgsNaoUsadas, "w") as output:
    writer = csv.writer(output)
    for val in imagens:
        writer.writerow([val])