#!/bin/bash
# wget https://gist.github.com/plutocrat/7636841e480f483a97c0d17568ac0ef1/raw/upload_clean.sh
# Dumps your database out, extracts image paths from it and compares it against a list of files from the filesystem
# Needs wp-cli to dump the database. 
# Only works on Linuxy servers
# It will give you a chance to bail before it deletes anything. 
# If it saves you a few hours, show your appreciation with bitcoin: 35Ws8qgKUCiVaca7vg5NWqkfQozPfzb4my

# Working Dir
WKDIR=zz_temp
# Image Storage Dir
IMGBK=image_backup
# Where is wp-cli
WPCLI=/usr/local/bin/wp-cli

# Make sure we're in the right dir
if [ -e wp-config.php ]
then
    echo "Found wp-config.php. You're in the correct directory."
else
    echo "wp-config not found. Please move this script to the correct directory and try again. "
    exit 1
fi

if [ -e $WPCLI ]
then
    echo "Found wp-cli"
else
    echo "wp-cli not found. Please install and/or update the path in this script. "
    exit 1
fi

mkdir $WKDIR

# Get list of files currently on System
find wp-content/uploads/ -type f | sort | uniq > $WKDIR/filelist.txt

# Dump Database
$WPCLI db export $WKDIR/dump.sql --skip-themes

# Search database for any mention of files in wp-content/uploads
grep -i -o -P "wp-content/uploads/[0-9][0-9][0-9][0-9]/[0-9][0-9]/.*?\.(jpg|png|jpeg|pdf|gif)" $WKDIR/dump.sql | sort | uniq > $WKDIR/dblist.txt
# Delete the dump file as it contains sensitive stuff
rm $WKDIR/dump.sql 

DBCOUNT=$(wc -l $WKDIR/dblist.txt | awk '{print $1}')
FILECOUNT=$(wc -l $WKDIR/filelist.txt | awk '{print $1}')

echo "There are $DBCOUNT files in the database and $FILECOUNT files on the filesystem"
echo "Details of these can be found in $WKDIR/dblist.txt and $WKDIR/filelist.txt"
echo 

comm -12 $WKDIR/dblist.txt $WKDIR/filelist.txt > $WKDIR/images_in_both.txt
comm -23 $WKDIR/dblist.txt $WKDIR/filelist.txt > $WKDIR/images_only_in_db.txt
comm -13 $WKDIR/dblist.txt $WKDIR/filelist.txt > $WKDIR/images_only_in_filesystem.txt

COMMONCOUNT=$(wc -l $WKDIR/images_in_both.txt | awk '{print $1}' )
DBONLYCOUNT=$(wc -l $WKDIR/images_only_in_db.txt | awk '{print $1}' )
FILEONLYCOUNT=$(wc -l $WKDIR/images_only_in_filesystem.txt | awk '{print $1}' )

echo "Comparing the files, there are ..."
echo " - $COMMONCOUNT files in BOTH the database and filesystem"
echo " - $DBONLYCOUNT files are ONLY found in the database. You may want to check your 404 logs for these"
echo " - $FILEONLYCOUNT files are ONLY found in the filesystem, and can probably be removed."
echo "Look in the following files for details of which files are in which group."

ls -hal $WKDIR/images_*

echo
echo "WARNING"
echo "The next step will move all files in $WKDIR/images_only_in_filesystem.txt to $IMGBK"
echo "If you want to do this manually, pres CTRL-C to end this script. Remember to clean up $WKDIR"
echo
read -p "Press RETURN to continue"

mkdir $IMGBK

rsync -a --files-from=$WKDIR/images_only_in_filesystem.txt ./ ./$IMGBK/

IMAGESIZE=$(du -h  image_backup/ --max-depth=0 | awk '{ print $1}' ) 
echo
echo "$IMAGESIZE of images synced into $IMGBK"
echo "Now deleting originals"
echo 

# Delete listed files
echo "Size of uploads folder before is $( du -h  wp-content/uploads/ --max-depth=0 | awk '{ print $1}' )"

while read FILE ;
do 
  rm "$FILE"
done <  $WKDIR/images_only_in_filesystem.txt

echo "Size of uploads folder after is $( du -h  wp-content/uploads/ --max-depth=0 | awk '{ print $1}' )"

# Optional 
echo "If you'd like to regenerate image thumbnails, hit Enter"
echo "The process may take a while. You could run it later yourself with "
echo " wp-cli media regenerate --only-missing --skip-themes"

read -p "If you don't want to do this now, hit CTRL-C to end"

$WPCLI media regenerate --only-missing --skip-themes

echo "Size of uploads folder after thumbnail rebuild is $( du -h  wp-content/uploads/ --max-depth=0 | awk '{ print $1}' )"