wp_uploads_purge.py

'''
@Author: Naren
@Date: 4/16/2014

This script can be used to delete uploads longer than specified period in years for a wordpress multisite. 
Note: Because uploads or wordpress blogs itself can be migrated, the timestamp of an upload wont necessarily reflect the date of creation.
Further the uploads are organized in such a way as below

year
   month
        date

Hence this script!

Algorithm:

Get current and previous year
Set the path to the uploads directory of wordpress: /wp-content/blogs.dir
Walk through the directory to find all subdirectories and put it in a list
Remove all entries that contains previous and current year and put it in final list
rm -rf all dirs in the final list

'''

import os, sys
from datetime import date

def main():
	if len(sys.argv) < 2:
		print 'Usage: ' + sys.argv[0] + ' <number of years>'
		print 'ex: ' + sys.argv[0] + ' 2 <-- To delete files older than 2 years'
		print 'ex: ' + sys.argv[0] + ' 3 <-- To delete files older than 3 years'
		return
		

	years = int(sys.argv[1])
	
	remove = []
	for i in range(0,years):
		remove.append('files/' + str(date.today().year - i))
	
	path = '.' # run the script in the blogs.dir directory itself
	
	# Directory walk to get all directories in the uploads folder
	def iterdirs(path):
		for root, dirs, files in os.walk(path, topdown=False):
			for d in dirs:
				yield os.path.join(root, d)
	
	files = []
	for d in iterdirs(path):
		if 'files' in d:
			files.append(d)
		
	newlist = []
	
	
	newlist = [f for f in files if not any(r in f for r in remove)]
	
	directoryListToRemove = []
	for a in newlist:
		mystr = a[-4:]
		if mystr.isdigit():
			directoryListToRemove.append(a)
	
	size = ''
	for a in directoryListToRemove:
		size = size + ' ' + a + ' '
	size = size.strip()

	if not directoryListToRemove:
		print 'No uploads older than ' + sys.argv[1] + ' years is found'
		return
		
	os.system('du -csh ' + size + ' | sort -h')
	option = raw_input('Above is the list of folders going to be removed, Do you wish to continue? [y|n]: ')
	
	if 'y' in option:
		for a in directoryListToRemove:
			sys.stdout.write('\nExecuting ' + 'rm -rf ' + a)
			os.system('rm -rf ' + a)
		print '\nAll commands executed successfully'
	else:
		print 'Folders are not deleted'

if __name__ == "__main__":
    main()