How to determine uploaded file is ascii/binary?

shelbytll

Hi,
how do I determine whether a file being uploaded is an ascii file or binary file? is it by the file extension? but there are so many file extension. can't possibly want me to check all file extension right?

epp_b

Generally, it is easier to determine whether or not a file is ASCII, instead of vice-versa. If it's not, it's obviously binary. Don't use the file extension, instead use the MIME content-type:

$filetype = $_FILES['filename']['type'];

if (ereg("text", $filetype)) {
   echo "ASCII File";
}
else {
   echo "Binary File";
}

The assigned value of $filetype for any ASCII file will be text/sometype (where "sometype" is something like plain, or html, or any valid text file type). If $filetype does not begin with "text", then the file is then Binary.

Note that filename must be an uploaded file that has been passed from a form in the last instance.

shelbytll

How do I use MIME content type? Is it what u have suggested earlier?

Mordecai

$_FILES['filename']['type']
See the ['type']? That means it will return the mimetype.

This is a built in PHP thing, but mimetypes are used all the time (not just PHP).

Hope that helps.

shelbytll

Hi,

met a problem with using MIME type. I realise alot of document gave me same MIME type. That is application/octet-stream.
Such as photoshop, excel, .cfg and .ini file.

I would like to know that will the MIME type string be the same even if on Linux platform?

Weedpacket

Most programs that need to determine whether or not a given file is ASCII or binary tend to examine the file itself rather than the filename (which could have who knows what extension) or MIME type (which might not even be present).

Generally, a file where most of the characters are between chr(32) and chr(127), with a scattered handful of control characters chr(<32) for things like linefeeds and tabs, are considered ASCII; too many weird control characters, or characters chr(>127) tend to be the sign of a binary file.

One trick that isn't foolproof (not a lot of heuristics are) but fairly good, is to find occurrences of chr(13) and chr(10) in the file, since these are the carriage return and line feed characters. A Mac text file would use chr(13) and not chr(10), a Unix text file would use chr(10) and not chr(13), and a Windows textfile would use chr(13).chr(10). So one can guess that a file that breaks all three rules is a binary.