Python3 script to find and move unused plots in latex project

find-unused-latex-plots.py

#! /usr/bin/env python 
#coding=utf8

import re, os

path = '/home/slisakov/Dropbox/documents/science/thesis/'
plots = '/home/slisakov/Dropbox/documents/science/thesis/plots/'

used_plots = []      # present in tex
commented_plots = [] # present in tex, commented
all_plots =[]        # present in tex (commented and uncommented)
unused_plots = []    # not present in tex
real_plots = []      # physically existing files in plots/ dir

# Create [all_plots], [used_plots], [commented_plots]
for f in os.listdir(path):
    if re.search(r'\.tex$', f):
        with open(path + f) as f:
            for line in f:
                if re.search(r'.*%.*\includegraphics', line):
                    g = re.sub(r'.*{(.*)}', r'\1', line).strip()
                    commented_plots.append(g)
                    all_plots.append(g)
                if re.search(r'^[^%]*\includegraphics', line):
                    g = re.sub(r'.*{(.*)}', r'\1', line).strip()
                    used_plots.append(g)
                    all_plots.append(g)


# Create [real_plots]
for f in os.listdir(plots):
    g = os.path.splitext(f)[0]
    real_plots.append(g)

# Check consistency
if len(commented_plots) + len(used_plots) != len(all_plots):
    raise Exception('Check your script sucker, it is not working properly')

# Create [real_plots]
for rp in real_plots:
    if rp not in used_plots and rp not in commented_plots:
        unused_plots.append(rp)

#print( len(unused_plots), unused_plots)

# Create plots/unused dir if doesn't exist
if len(unused_plots) > 0:
    os.makedirs(plots + 'unused/', exist_ok=True)

# Move [unused_plots] to plots/unused/, DON'T rewrite files
for f in os.listdir(plots):
    if os.path.isfile(plots+f):
        if os.path.splitext(f)[0] in unused_plots:
            if not os.path.isfile(plots + 'unused/' + f):
                os.rename(plots + f, plots + 'unused/' + f)
                print('file',f,'is moved to unused/ dir')
            else:
                print('file',f,'already exists. File ___IS NOT___ moved')