Own page in outlinks


(kwizzles) #1

Hi, i lately started to use piwik and am really happy with it

only problem i have is that i got my own page in the outlinks from time to time, its strange, it looks like this:

<row>
<label>http://mypage.com</label>
<full_url>0</full_url>
<nb_visits>82</nb_visits>
<nb_uniq_visitors>82</nb_uniq_visitors>
<nb_hits>87</nb_hits>
<entry_nb_unique_visitor>3</entry_nb_unique_visitor>
<entry_nb_visits>3</entry_nb_visits>
<entry_nb_actions>3</entry_nb_actions>
<entry_sum_visit_length>30</entry_sum_visit_length>
<entry_bounce_count>3</entry_bounce_count>
<exit_nb_unique_visitor>3</exit_nb_unique_visitor>
<exit_nb_visits>3</exit_nb_visits>
<exit_bounce_count>3</exit_bounce_count>
<sum_time_spent>110</sum_time_spent>
<url>0</url>
<idsubdatatable>13</idsubdatatable>
</row>
−
<row>
<label> http://mypage...</label>
<full_url>http://mypage.com</full_url>
<nb_visits>36</nb_visits>
<nb_uniq_visitors>36</nb_uniq_visitors>
<nb_hits>36</nb_hits>
<exit_nb_unique_visitor>1</exit_nb_unique_visitor>
<exit_nb_visits>1</exit_nb_visits>
<exit_bounce_count>0</exit_bounce_count>
<sum_time_spent>40</sum_time_spent>
<url>http://mypage.com</url>
</row>
<row>
<label>http://MyPage.com</label>
<full_url>0</full_url>
<nb_visits>4</nb_visits>
<nb_uniq_visitors>4</nb_uniq_visitors>
<nb_hits>5</nb_hits>
<exit_nb_unique_visitor>1</exit_nb_unique_visitor>
<exit_nb_visits>1</exit_nb_visits>
<exit_bounce_count>0</exit_bounce_count>
<sum_time_spent>5</sum_time_spent>
<url>0</url>
<idsubdatatable>14</idsubdatatable>
</row>

one of the item has different cases, MyPage.com instead of mypage.com but that shouldnt matter and isnt the onlyproblem
what i dont understand is why there is myspace.com in there, has it something to do with the 0?

i dont really have an idea where to look in the code and would be happy if somebody had some hints


(kwizzles) #2

i read a bit through code and the DB

what i found is that the mypage.com is already in piwik_log_action with type = 2, i think this means it is outlink and that it is already
in the database so i have to check at the place data is inserted
not where it is read and viewed, any hints?


(vipsoft) #3

You may need to use piwik_hosts_alias to exclude intra-links. This is needed when visitors reach your site through multiple names (eg www.mypage.com and mypage.com), and/or you use a mixture of those names internally on your anchor elements.


(kwizzles) #4

i think thats it, i thought that is done at the site settings in the admin panel

shame that i didnt read http://piwik.org/docs/javascript-tracking/

thanks a lot for the help


(kwizzles) #5

ok, still doesnt work

but it is indeed going from www.mypage.com to mypage.com
and vice versa
i added

piwik_hosts_alias = [ "*.mypage.com", "mypage.com", "www.mypage.com"

no it looks like

...
piwik_action_name = '';

piwik_hosts_alias = [ “*.mypage.com”, “mypage.com”, “www.mypage.com


piwik_idsite = 1;
piwik_url = pkBaseURL + "piwik.php";
piwik_log(piwik_action_name, piwik_idsite, piwik_url);
</script>

...

but still it gets shown in outlinks

did i miss anything?


(Matthieu Aubry) #6

in the next version, outlinks on domains that you entered for your website in the admin, won’t be counted in Piwik. This was a limitation in existing version that we are fixing.
Also thanks for the report, it seems the outlinks wasn’t properly strtolowered, I will check that.

Also, like vipsoft suggest, the piwik_alias_hosts should work to not count links on other domains.

finally there is also http://dev.piwik.org/trac/ticket/579 that you mentionned


(vipsoft) #7

kwizzles: It should be sufficient to have:

piwik_hosts_alias = ['*.example.com'];

This only applies to new visits. It won’t correct older entries in your database.

matthieu: Yes, this is a behavioural difference on Safari. Other browers return a lowercase hostname. This is fixed in the new piwik.js.


(kwizzles) #8

[quote=vipsoft @ Mar 24 2009, 11:40 PM]kwizzles: It should be sufficient to have:

piwik_hosts_alias = ['*.example.com'];

This only applies to new visits. It won’t correct older entries in your database.

matthieu: Yes, this is a behavioural difference on Safari. Other browers return a lowercase hostname. This is fixed in the new piwik.js.[/quote]

i of course checked for new recoirds only
will try to debug it the next days

thanks for your input!


(kwizzles) #9

hi, first thanks for the help!

i got it now working, mostly …

maybe it would be good to tell the user somewhere
to use lowercase in the piwik_hosts_alias variable or
add a toLowerCase() at line 193 …

192             for (var i=0; i < _pk_hosts_alias.length; i++) {
193                     alias = _pk_hosts_alias[i];

i am not sure about mixed cases in the hostnames
i think, since dns is case insensitive, everything should be lowercased
i’d like to have the statistics for mypage.com MyPage.com added together

i also don’t know where to implement that, in .js
lowercasing the host part of document.location.href ?

do it in .js because its actually fixing a browser issue might be an argument

in php part i am not sure where, maybe in Tracker/Action.php
generateInfo() ?


(vipsoft) #10

This is now fixed in SVN: http://dev.piwik.org/trac/changeset/1019


(kwizzles) #11

yes i saw that

but what about preventing mixed case hostnames from safari browsers showing up, i dont see something handling that in svn current piwik.js?


(vipsoft) #12

Thanks. Fixed in SVN.


(kwizzles) #13

just updated to latest.zip
and now i got mypage.com as well as MyPage.com in outlinks again
i got www.mypage.com and mypage.com in piwik settings page for the site
and using javascript alias “*.mypage.com”

i checked piwik.js and the lowercasing seems to be there … so i dont really get whats the problem

any idea?


(vipsoft) #14

As before, the change doesn’t modify outlink data collected prior to the upgrade. If you have some new insight into the cause, we’d appreciate it.