Recover original images from nextcloud previews

When you find yourself in a situation where you only have downloaded previews from your nextcloud server (e.g. via rightclick -> save as...), this repo will help you recover the originals behind those previews.

1. Prepare data

We'll match the preview images over their hashsums, so the first thing we need is two lists that map md5sum -> path for the selection of previews and all available previews on the server.

Desired preview hashes

In a directory containing all the desired previews, execute.

find . -name "*.jpg" -exec md5sum {} > desired-hashes.list \;

All preview hashes

On your server, navigate to your preview directory. Find your nextcloud data, locate your preview directory in one of the app_... dirs. Inside the preview directory, execute:

find . -name "*max*" -exec md5sum {} > all-hashes.list \;

Copy all-hashes.list over to your main machine.

FileId -> Path mapping

Get the fileId -> filePath mapping from your nextcloud DB. I am using postgres in docker, so it looked like this for me:

docker exec docker_db_1 psql -U nextcloud -c "COPY (SELECT fileid,path FROM filecache) TO '/tmp/fileid-path-mapping.csv' (FORMAT CSV)"
docker cp docker_db_1 /tmp/fileid-path-mapping.csv .`

Copy the fileid-path-mapping.csv file over to your main machine.

2. Join it all together

On your main machine, gather the generated files. Then execute

npm install
node index.mjs all-hashes.list desired-hashes.list fileid-path-mapping.csv

You will now have a mapped-everything.awkcompatible. Copy that to your server.

3. Collect all the originals!

On your server, copy all originals together. You need to know which user collected the previews. The users directory will be your USER_NEXTCLOUD_ROOT. For me, it's USER_NEXTCLOUD_ROOT=nextcloud/{username}/. Mind the trailing dir separator.

Execute:

mkdir collected
cat mapped-everything.awkcompatible| awk -F ':' -v q="'" -v rd="$USER_NEXTCLOUD_ROOT" '{ print "cp " q rd$2q" "q"collected/"$1q | "/bin/bash" }'
tar cfv collected-originals.tar collected/*

The originals are now in the collected-originals.tar archive. Missing files should be reported from cp.

Cleanup

Some files and directories have been created. Clean them up.