(no subject)

[arethinn]

Dear Lazyweb^* (as foxgrrl would say),

Does anyone know why some picture file names on DeviantArt are of obvious and readable forms like Night_fairy_by_Lillucyka.jpg and others come out as hex gibberish like e3e303d151d82974e36b0af65d228309.jpg? (Those happen to be examples from the same user's gallery.)



* -- Actually not that lazy. We've spent about half an hour looking at dA's own help and Googling and trying hex to ASCII or Unicode converters, to no avail. It does not seem to be related to deviation age, length of title, mature filtering, print availability, or anything else we can think of.

Comments

[twopiearr.livejournal.com]

i suspect that the backend changed at some point from reencoding uploads to be machine convenient to keeping filenames intact, but this is just a guess.

[arethinn]

No, as I said in the post, it appears to be unrelated to the age of the deviation. Posts both new and old can go both ways. The example I gave above was not made up, it was copied from a deviation posted in March.

[michiexile]

That's a string of 32 hexadecimal digits, which makes it contain 128 bits, making it the right size for a MD5 or some of the SHA cryptographic hash. These, again, have very few (but some) data collisions, and hence are very often used to get you random strings which stays invariant for some input data.

So, my guess is that if you use a program that'll give you the md5 sum of the gibberishy .jpg file, you'll recognize the output. Possibly.

[arethinn]

You mean run an md5 hash on the file itself and see if it yields that filename? I suppose that would tell me where it was coming from, but to what end would they be doing that in the first place, and why only on some files? (there's not a box to check "encrypt filename" or anything when uploading a deviation.)

[michiexile]

I know several different web apps that do just that to handle "anonymous" files.

[arethinn]

Um, ok, but why would a file uploaded by a logged-in user to their own gallery be "anonymous"?

[michiexile]

I dunno. :-)

[arethinn]

Errr....okay, so, excuse the rude tone as I really intend mere perplexity, but this is an ultimately inapplicable explanation for how arbitrary DeviantArt files would acquire these odd filenames?

[michiexile]

I'm in no way involved in DA, I don't hang out there, and I only very rarely end up even following links there. In your post above, you said:

We've spent about half an hour looking at dA's own help and Googling and trying hex to ASCII or Unicode converters, to no avail. It does not seem to be related to deviation age, length of title, mature filtering, print availability, or anything else we can think of.

and I tried to point out that in high probability there IS a relation to the MD5 fingerprint of the file.

However, I still have no idea WHY DA uses this for some files, I have no idea what makes some files named with MD5 and others with original names. Since I don't use the website, I haven't seen their upload dialogues, and thus couldn't possibly remember subtle parts of those that might end up generating anonymous files; and you seemed to have done a check for whether this is correlated with time, so it doesn't seem to be due to an upgrade in the backend software.

I never did intend to give you a complete solution, since I don't have one to give. I did intend to give you a suggestion of what might cause the "hex gibberish" filenames.

[bluegodjanus.livejournal.com]

Have you tried quantum uncertainty? I find that helps me explain a lot of the world's weirdness.

[arethinn]

Ummm... okay. I was hoping there might be an actual systematic reason.