Way to tell if an uploaded file matches file on server without same file name

[anolan13]

Hello,

 

I need a way where if users upload a file to my server that's the same as another file (both .mp3 files) but don't necessarily have the same filename, the server will know and perform an action based on this.

 

Is there a way for PHP to check certain characteristics in an mp3 file that would make it different from another file or the same as another file regardless of the file name?

 

 

Just a quick example,

These two files are the same but have different file names...

 

mysong1.mp3

thebestsongever.mp3

 

Is there a way for the server to tell the difference or similarity once their both uploaded?

 

Thanks,

[Mchl]

MD5() both files and compare

[gevans]

Mchl, can you jsut md5 the dir/filename to the file?

 

(didn't know this)

[Mchl]

That's what md5() is for:

 

http://www.php.net/downloads.php

[gevans]

I don't know how I completely forgot about that. So it will match if both .mp3 files have the same content?

[gevans]

Wouldn't you have to use md5_file() ??

[Mchl]

No. It will match if both files generate same md5(). There is very low probability, that two different files will generate the same md5() (situation called collision). To reduce risk of collisions, one may use hash with some other hashing algorithm, but I think in this case md5() should be sufficient.

 

The best would be to calculate md5 whenever a file is uploaded and compare it against hashes of previously uploaded files (preferably stored in database).

 

Wouldn't you have to use md5_file() ??

 

That would be actually easier than loading a file into string

 

There's also hash_file

[gevans]

Sounds good, nice to remember.

[Mchl]

Here's a nice site about MD5 usage for identifying files

http://www.win.tue.nl/hashclash/Nostradamus/

[gevans]

Very interesting, and nice use of a ps3

[PFMaBiSmAd]

I would compare file sizes and the md5/checksum. Files that are different sizes cannot be the same. This will further reduce the chance of different files having the same md5/checksum.

[Mchl]

True dat!

 

Compare sizes first, so you can save some CPU cycles calculating hash when they're different

[PFMaBiSmAd]

Actually, the third step if you want to know for certain that files are not the same, if you have matching lengths and md5/checksums, would be to do a byte-by-byte comparison.

[Mchl]

I think I'd just compare hashes generated by some other function... Probability that there's collision of md5 hashes is low already, and probability that both md5 and sha1 (or something) hashes are the same is like... nil[1].

 

And I can hardly see anyone manipulating a file so that he can not upload it

 

 

 

[1] nil is not zero, but close enough