View >Actions >Pages spammed by search engines and local copies


(thoro) #1

Is there a way to mask certain pages or folders from the view >Actions >Pages?

In my installation this view is completely spammed with entries like:

/search?q=cache:1aBc2DEf3gh4:www.test.com (various entries)
/translate_c?hl=pl&langpair=en|pl&http: (various entries)
/cache.aspx?q=“test”&d=11324516661761 (various entries)
/showthread.php?p=123441 (various entries)
/file: (folder with subfolders and pages)
/page url not defined

I plan to use Piwik as an inhouse web analytics software and >Actions >Pages will be our main tool.
I’d like to get the data as clean as possible for my colleagues.

What I did now is that I added an ‘if ( window.location.hostname == ‘www.mysite.com’ )’ to the
Piwik JavaScript on my site but I think the entries are from pages that were cached/saved before
so it would be better to handle this on my side.

I searched this forum but found no way to mask them either before or after they get into Piwik’s database.
Is this already possible or planned for the future?

Cheers

Thorsten


(vipsoft) #2

You can strip these using JavaScript before calling the Piwik tracking code, or wait for http://dev.piwik.org/trac/ticket/41 to be implemented.


(thoro) #3

Thanks for the hint! I modified /js/piwik.js:

/*
 * Log the page view / visit
 */

function logPageView() {
    if ( (configHostsAlias == 'www.mydomain1.com') || (configHostsAlias == 'www.mydomain2.com') ) {
        var request = getRequest();
        request += '&action_name=' + escapeWrapper(configTitle); // refs #530;

        // encode custom data
        if (isDefined(configCustomData)) {
            request += '&data=' + escapeWrapper(stringify(configCustomData));
        }

        request += executePluginMethod('log');
        getImage(request, configTrackerPause);
    }
}

After compressing the file using “yuicompressor-2.4.2” I uploaded it to ‘/piwik.js’ on my webserver and flushed my Piwik database.

A crude approach but now I get clean results in >Actions >Pages.