How to decrypt images

[asanakoy]

Hi,

Google art stores images in encrypted format. I think the haven't changed the encryption algorithm till today.
For example image.
I have seen in your core.py some functions that decrypt images. Could you please explain how this decryption works?

[expenses]

I've been working on this as well.

Try decrypterFragment. It takes the bytes of the image, decrypts them and then returns them.

Use it like this:

from extractionGoogleArtProject_Unix import decrypterFragment

input = "encrypted.jpg"
output = "decrypted.jpg"

fragment = open(input, "rb").read()
fragment = decrypterFragment(fragment)
open(output, "wb").write(fragment)

Google art stores images in encrypted format. I think the haven't changed the encryption algorithm till today.

I'm not sure what you mean by this, but I'm not sure if it'll work for all images.

I don't really know how the decryption works, I think @EmelyanenkoK just copied it right out of the minified javascript of the website - that's why the code is written so unobviously.

As far as I've been able to tell so far, the files aren't really encrypted - it's just that 4 bytes have been added at the start and the ends and their huffman tables have been messed up. Use a hex editor like Hex Fiend to compare the files:

If you're really interested, here are the jpeg marker bytes.

I presume that this function just undoes those changes. I'll try digging into it to see if I can understand it. Email me at expenses@airmail.cc or post in this thread I guess if you want to know more!

[expenses]

Update: I've made an attempt to understand the decrypt function and make it more explicit, see my fork here: https://github.com/expenses/GAPDownloader/blob/master/core.py#L240

[asanakoy]

@expenses This is awesome! Thanks for a great effort! Your function makes perfect sense now.

[expenses]

@asanakoy No problem! I'm still working on it, mostly to understand it and see if I can speed it somehow.

[asanakoy]

@expenses if you are interested in scraping the data from googleart on a large scale take a look at my scraper https://github.com/asanakoy/googleart_scraper.
But I donwload small images, not zoomed in, and a lot of metadata.

[EmelyanenkoK]

@expenses
I just translate obfuscated js-code from GAP to python, so it is indeed very messy. However in 24hours after I publish code on github, google have changed something in data encryption.
Since I only needed a few images at that moment and I already got it, I abandoned this project.