CODEWITHSUNDEEP
phpsqlmysqlcounter
Enkay
Enkay

Reputation: 1915

mysql query if statement

To keep this simple, let's say I'm making a basic pageview counter in php that stores the counts for each page in a mysql table. The table had 2 colums: PAGE_ID and COUNT.

I added the following code to every page:

$query = "INSERT INTO table VALUES ('$page_id', '1')
         ON duplicate KEY UPDATE COUNT=COUNT+1";

$result = mysqli_query($cxn, $query);

To make sure each person viewing the page only triggers the counter once I added PHP sessions. Basically if you view a page, the page_id gets stored in a session and the counter php code checks that session before triggering the counter. Worked fine in my own testing.

Some pages were getting too many views, I was suspecting duplicates, so I started logging IPs and User agents. It turns out in about 10% of the cases, the IP triggers the counter for the same page 2-3 times in a few minutes.

First question What could be causing the duplicates? The problem seems to happen mostly with IE8 and Safari but I also have at least one instance of it happening with IE7 and IE6. Any known problem with php sessions? Should I use cookies instead?

Part 2: I modified my table so that it now stores the last unix time stamp and the last IP that triggered the counter.

I want to modify my query so that before it runs the "COUNT=COUNT+1" it checks for the following:

If the current IP is the same as the last stored IP for this page {

     check that it's been at least 5 minutes before doing COUNT=COUNT+1

} else { COUNT=COUNT+1; }

Second question How do I write that in a mysql query, while keeping my "ON duplicate KEY" statement?

I understand this statement wouldn't be 100% accurate but until I can figure out why the session thing doesn't seem to be working, this would work. My website is low traffic and I rarely get more than 1 visitors on the same page within a 5 minutes time frame.

Thanks

Upvotes: 0

Views: 875

Answers (4)

docwhat
docwhat

Reputation: 11704

What you're looking for is a CASE statement: http://dev.mysql.com/doc/refman/5.0/en/case-statement.html

An example CASE statement:

SELECT name,
       (CASE WHEN is_happy THEN "Happy!"
        ELSE "sad." END) as happiness 
FROM user_state;

They can be used in UPDATE and INSERT as well.

I suspect that you want to look at the IE stuff, instead. I had similar problems in the past and I ended up having to use javascript to de-bounce clicks. I'm not sure if it was IE users double clicking on things or if it was IE being weird.

Upvotes: 0

Enkay
Enkay

Reputation: 1915

I think I may found a way to fix my session problem. Actually, I switched from sessions to cookies.

A lot of my pages are getting views through an iframe. The iframe was causing a problem with IE browsers. IE was not able to read cookie data from within an iframe unless I add a p3p header.

I added the p3p header and now it seems to be working correctly. I'm so tired of microsoft and their idea of security. I also have no idea what a p3p is but here's what it looks like in php.

header ( "p3p:CP=\"IDC DSP COR ADM DEVi TAIi PSA PSD IVAi IVDi CONi HIS OUR IND CNT\"");

Upvotes: 0

MindStalker
MindStalker

Reputation: 14864

I'd suggest you try to see if you can fix the PHP, but ignoring that. You could store the unixtimestamp divided by 300 (IE, in 5 minute intervals)

$query = "INSERT INTO OtherTable VALUES ('$page_id', '$IP_ADDRESS', (UNIX_TIMESTAMP(NOW())/300))";

If records updated is 0 you don't need to update the page count.

Upvotes: 2

OMG Ponies
OMG Ponies

Reputation: 332661

1. What could be causing the duplicates?

Some pages were getting too many views, I was suspecting duplicates, so I started logging IPs and User agents. It turns out in about 10% of the cases, the IP triggers the counter for the same page 2-3 times in a few minutes.

There's no way to know if the host requesting the page sits behind a NAT router - to you the request would have the same IP but in reality is a different host. A cookie or session would help you isolate on a per workstation basis, though I wonder when you would write the info to the database.

2. How do I write that in a mysql query, while keeping my "ON duplicate KEY" statement?

I don't see the need for the COUNT = COUNT + 1, because SQL has a COUNT function:

  SELECT page_id,
         COUNT(*) 'num_hits'
    FROM ZZZ_NETWORK
   WHERE page_id = ?
GROUP BY page_id

If you used the following structure for ZZZ_NETWORK:

  • page_id, primary key
  • ip_address, primary key
  • timestamp, primary key

...you wouldn't have to be concerned with duplicate key handling, and it would allow you to know how many hits you got in a specific day, week, month/etc. Including the timestamp would ensure there could never be duplicates.

After a month I'd have thousands of rows in my table if I use one row per page view.

Drive space is dirt cheap, and having a timestamp so you can datamine would be invaluable for reporting. But you don't have to keep all the data either - you could archive it by dumping the data to file so you could retrieve it if necessary.

I don't understand how it would ensure I would never get a duplicate.

A timestamp includes both date (Dec 25/09) and time (07:00:00 AM). Some datetime data types get down to fractions of a second. This makes it close to impossible to have the same date & time for a given page_id with a given IP address - I can't click the refresh button fast enough, even if I wanted to. So records could never be duplicates, because the last of the three columns would have a different value every time (no pun intended).

Upvotes: 1

Related Questions