Download Counting with Apache and PHP
Collect statistics on the popularity of your downloads with Apache's mod_rewrite and PHP.
How some of the others do it
Some sites present you an URI like
http://www.example.com/download.php?file=/foo/bar,
which is less than perfect. Being a Windows user (well sometimes)
I expect that when I add a file to my download manager I'll see the
filename in the list - unfortunately the result from adding a file from
such a link is the meaningless download.php in your
file list.
While the method mentioned above is easy to implement it is not the best way to do it. We want visitors to see the real filename as the URI not as a query string. So what we'll do internally is exactly the same as in the above example but this time the visitor will see the real filename.
How to do we do it
Our URIs will be in the form of
http://www.example.com/foo/bar
which makes more sense, doesn't it?
What do we need
- Apache compiled with mod_rewrite
(off by default, you need to add
--enable-module=rewriteto your configure line) - PHP as the scripting engine
- A database, like MySQL, to store the download data
Apache's configuration
Options +FollowSymLinks RewriteEngine On RewriteBase /foobar/ RewriteRule download/send.php - [L] RewriteRule download/(.+..+)$ download/send.php?file=$1 [L]
You can put this block of code in a .htaccess file
in /foobar/, your downloads should be in
/foobar/download/ - these are the directories
accessible by the paths mentioned from your webserver.
What we do is switch on FollowSymLinks which is
required by mod_rewrite, if you don't need other options
you can remove the +. We turn on the rewrite engine
which is off by default, then set the base location for the URI rewrites.
The next two lines are our rewrite rules, if the request is for
send.php (our script that counts downloads) we don't modify
it, if we get a request for download/foo.bar that will become
download/send.php?file=foo.bar for Apache (the rewrite base
is prepended). The rule processed only filenames with extensions.
The download counter
<?php
$file = isset($_GET['file']) ? trim($_GET['file']) : '';
if (!$file) {
die("Error");
}
if ( substr_count($file, '..') > 0 or substr($file, 0, 1) == '/' ) {
die("Invalid filename.");
}
$path = dirname($_SERVER['PATH_TRANSLATED']) . '/' . $file;
if ( !file_exists($path) ) {
die("File not found: $file");
}
$ext = explode('.', $file);
if ( sizeof($ext) < 2 ) {
die("Invalid filename: should have extension");
}
We do some checking first, you should never
display files to the visitors that they have requested without checking
for unwanted characters like ../ or / at
the start of the filename.
The $path's value is set to the directory containing
our script + the filename requested. We then check if the file really exists
and if it has an extension.
$ext = $ext[sizeof($ext)-1];
switch ($ext) {
case 'tgz' :
$type = 'application/x-gzip';
break;
case 'php' :
$type = 'text/html';
break;
default :
$type = 'text/plain';
}
header("Content-type: $type");
By default PHP sends a content type header of text/html
to the browser so we need to modify it when this is not right. You should
add more extension -> MIME type pairs if you serve different file types.
We've sent text/html for PHP files because we want to present
them syntax highlighted - our script is something like a download counter
+ PHP file browser.
Because we send a HTTP header there should be nothing sent to the browser before the last line of the block above. Output buffering can be used as a way around this but it's not needed here.
$fd = fopen ($path, "r");
$code = fread ($fd, filesize($path));
fclose ($fd);
switch ($ext) {
case 'php' :
?>
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
"http://www.w3.org/TR/1999/REC-html401-19991224/loose.dtd">
<html>
<head>
<title><?php echo $file?> syntax highlighted</title>
</head>
<body>
<?php
highlight_string($code);
?>
</body>
</html>
<?php
break;
default :
echo $code;
}
This is the part which sends the file to the browser or presents the highlighted PHP file. The functions used are binary safe so you can send every type of files not only text.
require_once('../../config.php');
$db = db_connect();
$sql = "UPDATE download_file SET count=count+1 WHERE file = '$file' ";
$db->query($sql);
?>
And the final one, which actually counts the download. Include a file
with our database info first - as you can see if opens a file which is
out of the webserver root which makes it pretty safe. We connect to
the database with our predefined function db_connect()
which returns a PEAR::DB instance.
The query updated the download_file table, which
holds the download files and the times they have been downloaded.
We increase the count field by one for our file.
Note: Never make the mistake to first SELECT
the count value, increment it in PHP and then write it to the database.
Delicious
Digg
Reddit
Technorati
Comments
Useful addition to database operation
I've added a few code. It controls and inserts new row if file name not exist in table. If exist increments count value. Hope it helps some one.
I will also implement get additional info about visitor too.
Code snippet
------------------
$addrecord ="INSERT INTO download_file (count_id, count, file) VALUES (NULL, 1, '$file')";
$check ="SELECT file, counter FROM download_file WHERE file= '$file'";
$update_it = "UPDATE download_file SET counter=counter+1 WHERE file = '$file' ";
$result = my_own_mysql_query_routine($check);
if (mysql_num_rows($result))
{
my_own_mysql_query_routine($update_it);
}
else
{
my_own_mysql_query_routine($addrecord);
}
-------------------
end snippet
Take care
Use INSERT IGNORE
I think it would be faster like:
INSERT IGNORE INTO download_file ...
UPDATE download_file SET ...
It doesn't involve 2 way communication with the database and a check in between.
ok
good work
Windows and binary files
A small note, on Windows use
$fd = fopen ($path, "rb");
to open the files in binary mode. Note the added b.
First, thanks for this information! I have just modified the above for use on my own site. I made a few changes such as only counting zip files, and logging the datetime each counted file is downloaded. However, I ran into a problem where it would occasionally count the file twice.
While using Firefox (and possibly others), if you click on the link it counts it once; however if you right-click and select 'Save Link As' the download count gets incremented by two. What I found was that the browser was sending a HEAD request before the GET request.
My solution was to change my code as follows:
$PPDB->query("INSERT download_log VALUES (null,'$file',now(),'$_SERVER[REMOTE_ADDR]')");
$fsize = filesize($path);
header("Content-Type: application/zip");
header("Content-Length: $fsize");
$fd = fopen($path,"r");
$code = fread($fd,$fsize);
fclose($fd);
echo $code;
became
if ($_SERVER['REQUEST_METHOD']=='GET') {
$PPDB->query("INSERT download_log VALUES (null,'$file',now(),'$_SERVER[REMOTE_ADDR]')");
}
$fsize = filesize($path);
header("Content-Type: application/zip");
header("Content-Length: $fsize");
if ($_SERVER['REQUEST_METHOD']=='GET') {
$fd = fopen($path,"r");
$code = fread($fd,$fsize);
fclose($fd);
echo $code;
}
which appears to happily ignore the initial HEAD request (apart from sending the header information, of course...)
As you will notice, I also opted to include the content-length header; this enables the downloading mechanism to provide estimates of completion time, etc...