MIME type of an image?

DmitryK

Hello there PHPeers 🙂.

Had a bit of a sleepless night and thought I'll ask the question on here in case the solution is right in front of me.

I would like the check the MIME type of a particular image. However, the image is being pulled by a PHP script out of a database, so that my image tag is like so:

<img src="image.php?name=imagename" id="image1" alt="I'm an image yay!" />

The trouble, of course, is that opening up the file using fopen() or getimagesize() doesn't produce the right header, and the filename under the src attribute doesn't list the image extension. In my mind the way around it I came up with before I collapsed to sleep in front of a laptop is to somehow make image.php run through first so we have a header on hands and pass the values onto my current script.

Am I thinking in the right direction or do I need more caffeine? Thanks in advance!

DmitryK

Now that I think about it, mixing the two scripts seems like the only way around this. I mean, to this script, image.php?name=imagename is nothing but a path. It won't be an image until the actual function within image.php is run. Hence, I need to find a way to execute the script before I can catch its output. I can feel the answer's near, anyone have any tips?

Thanks!

NogDog

Something in the image.php script needs to determine the image type the send the appropriate mime type via a Content-Type header() command. If you are storing the images directly in the database, then my recommendation would be to also determine the mime-type when storing the images, and saving it in its own DB column. Then when you query the DB for the image, you can also get the mime-type in the same query and set the header() accordingly.

DmitryK

Hey Nog,

Yeah, that's exactly how the image.php script works - it first adds a header and then pulls out the appropriate DB entry to output the actual image. Which works great in a browser window, but the PHP functions see the entire image.php?name=imagename string as a path and think it's incorrect.

Buh! Shtupid machines =)

DmitryK

I could actually do it by including the image.php file into my main script, and to image.php adding a few lines which will allow me to feed the path directly into the function versus doing it via the GET method like it is done right now. Still, I wanted to try achieving the result without doing any modifications to image.php.

Anyffing?

NogDog

Not sure I follow you. The display script will have to be a stand-alone item, as it needs to be called by a separate HTTP request from the browser after it parses the <img> tag and is ready to request the image file from the web server.

Frankly, my recommendation would be to not store the images in the database itself. Instead, I would store the images as files in some directory (possibly outside of the web root directory if you only want them to be accessible via your image.php script). All you would store in the database is the file name. This would then accomplish a few things for you: you'd be able to apply the getimagesize() function to the files, and it would probably decrease page load time as you would not have to obtain all that image data from the database: a simple readfile() would do the trick.

DmitryK

The display script is indeed a standalone item. I was thinking of including it into the main script just for the purposes of testing the current image for the MIME type. And I'm dealing with an existing database so I can't move the images elsewhere.

Creating a field in the database makes the most sense though, as the scripts that access the data won't have to worry about guessing what they're receiving. I'll see if I can try that.

Weedpacket

DmitryK wrote:
getimagesize() doesn't produce the right header,

getimagesize() requires a filename, of course, not a blob. I mention this for context later.

In general I'd go with NogDog's suggestion of having the file separate and using the database to store various metadata about the file (since, as well as for the reasons he stated, there's not a lot you can do with a blob in a database).

But there is another option: write a stream class that wraps blob-related database stuff and presents it as something that can be used like a file. Then you've got a filename (more accurately, a URL - something like "db://images/187" or however you want to structure it), and then you can use getimagesize() and suchlike directly on blobs.

DmitryK

Thanks for the advice guys.

Yeah.. I figured I'll have to do something like this. The trouble of course is that with a database I can't run any of the mime checking functions as they work with physical files. So I have to pull the image from the database, make a file out of it, detect the header, do stuff with that, and then pull the same image out of the database again to actually display it within the browser. Not very efficient when you have other better things you could spend server time on. Modifying the database is a better option indeed it seems. It might take longer initially but in the long run it will save a lot of headache.

By the way, I found a way to obtain the image using my image.php?name=imagename scheme. Normally if you feed something like "image.php?name=imagename" into getimagesize() it will not find it to be a valid file, for obvious reasons. All you do is run the script from the console, pass the needed parameters as arguments, and then convert the $argv array into the appropriate $_GET entries. That way the image.php script requires very little modification. Eg.:

// grab the output and feed the arguments in
$image = shell_exec("php image.php name=imagename");

// inside of image.php: the passed arguments are part of the superglobal $argv array. 'name=imagename' is $argv[1]. $argv[0] is the script name.

if ($argv)
   for ($i=1;$i<count($argv);$i++)
   {
       $it = split("=",$argv[$i]); // split our string into appropriate $_GET array entries
       $_GET[$it[0]] = $it[1];
   }

Will probably come in handy somewhere down the line or maybe someone else might find it useful. Gotta love programming - a learning experience every time!

I'm such a nerd... :rolleyes:

Weedpacket

getimagesize('http://www.yourserver.com/image.php?name=imagename');

but it seems excessive to use an entire webserver when just a class will do 🙂

DmitryK

Weedpacket wrote:
Or
getimagesize('http://www.yourserver.com/image.php?name=imagename');
but it seems excessive to use an entire webserver when just a class will do 🙂

Trouble is, to get my hands on the blob of the image I have to pass on arguments via URL. And since I'm not doing it directly by directing the browser to the script the only way to pass those to image.php seems to be using the console.

Weedpacket

That line I wrote does pass arguments via URL.... have another look.

DmitryK

Hmm.. I have Not tried passing a URL before, just an absolute path. That makes sense then.

Let me try it!

DmitryK

Ah.. URL file access was disabled on my server 🙁.

DmitryK

Ok, I managed to retrieve the contents with CURL, though that still means I have to query the database twice for the same image, plus I have to find a way to deal with the actual blurb since getimagesize() does not work on file contents. I have PHP5 installed on the server and no PECL so I can't run fileinfo. Its deprecated predecessor mime_content_type() gives me an undefined function warning too. Urgh!

Either way, having an extra database field would be far more efficient than pulling the same binary data from the DB two times in a row. However I want to get to the bottom of this and learn from the experience 🙂.

Must sleep now...