Mongol

Authors
Justin Quick <>

Version
0.1

pip install mongol==0.1.0

Mongol is a web analytics tool to let you track your site's traffic and generate custom reports. It uses MongoDB as the datastore and Python with pymongo to populate and query the database. It uses standard WSGI to gather analytics information and serve a tracking GIF pixel. Custom analytics data can be tossed at the service and you can then make custom reports to track pretty much anything you want. This project is perfect at answering questions like "What are the top 10 most viewed pages?"

Requirements

Mongol requires pymongo to connect to a MongoDB instance. For the serivce, it is recomended to use with Gevent but it also works with Tornado and even Apache/mod_wsgi. Here is the recomended list of requirements:

sudo pip install pymongo gevent

Install

Setup and install MongoDB. Here is their quickstart guide. Next, download the source code and install the scripts:

git clone git://github.com/justquick/mongol.git
cd mongol
sudo python setup.py install

mongol-serve

mongol-serve [address:port]

Starts the service instance at the optional address and port. Defaults to host at 0.0.0.0:8000. This service hosts the tracking GIF but you can pass any sort of analytics data as GET parameters. The only two required parameters are 1) site which is the domain of your site and 2) url which is the absolute URL of the page you wish to track. site and url will be populated if HTTP_REFERER which the browser sends as the page that the tracking GIF was found on. Any other GET parameters are stored for later use.

mongol-report

mongol-report config-file

Generates JSON reports from the analytics database based on the configuration parameters found in the config-file filename argument. This is best put on a cronjob so that the generated reports are up to date. Report files can be served up by a separate service (ideally lighttpd) so that the analytics service doesnt get clogged up by serving the reports.

Configuration Options

Configuration files contain sections of reports to generate when you run mongol-report. Each section name will generate one report file called section-name.json. Here are the options for each section:

> - output - required string. The output directory to store files in > (eg ~/mongol/reports) > > - time - optional string. Time at which to remove older records > (eg days:2). Mongol is really suited to look at data in the past > X amount of time and removes any records that were recorded before > then. By default it does not do any limiting. > > - callback - optional string. Function name to wrap results with > creating proper JSONP reports. Defaults to results. > > - limit - optional int. Limit the results in the reports. Defaults > to 10. > > - order - optional string. Order the results. Either A for > ascending or D for descending. Defaults to D > > - mapper - optional string. The JS source code of the mapping > function to use in MongoDB's map/reduce capabilies. Defaults to : > > function () { > emit({url:this.url, title:this.title}, this.pageviews); > } > > - reducer - optional string. The JS source code of the reducing > function to use in MongoDB's map/reduce capabilies. Defaults to : > > function (key, values) { > var total = 0; > for (var i = 0; i < values.length; i++) { > total += values[i]; > } > return total; > }

For help with the map/reduce stuff, take a look at MongoDB's MapReduce documentation

TODO

> - Use cookies > - Enable MongoDB clustering support