merge

Show top datasets cross-publisher. Drop-down for the publisher. Browser version numbers filtered on download, so you get this version in the CSV too - for privacy. single_popular_dataset now copes when not much data, and can return the figures so DGU can reskin it in its own repo. Notes about usage stats centralised to notes.html.

Changes to support % of bounces from /

New single entry random dataset helper

Added support for a popular datasets widget.

The widget will look in the ga_publishers for the 30 top publishers
and then randomly choose one from that list to show the top 10 datasets
of all time

Removing unused file

Changed the icons in the minor nav to simpler icons

Adding back a link to the publisher

Rename of all time to all months

Adding new social network referral information

Changes related to presentation, and new tabs.

- Changed nearly all values to percentages of the total
- Added non-version info for OS and browsers
- Made tabs into dropdowns for Browser/OS

Remove unnecessary footnote marker (*)

Made sure the csv is served as an attachment

Changes to enable proper download

Moved minor nav into util file and implemented

Fixes to show All Time data

Added a class for spacing to More link

Added a link to the publisher statistics

Changed to provide a lambda to check if the extension is installed

Added a helper for generating a list of all time popular datasets

Nicer 404 message

Make sure we 404 for non-existent groups

Human readable times

Trimming long float numbers

Change the rules for top datasets

Tidying up

Changing the content disposition, just in case

Changing the download url for csv

Minor change to sidebar

Removing cruft

Debugging

reordering maps

Fix to date fetching

Uncommenting needed code

Template fix

Minor template tweaks

Fixing setup

Fixing a typo in the instructions

Reinstating a sidebar with related info.

Adding site analytics at /data/analytics

- Allows for download as CSV
- Also refactored the publisher urls to allow index page and a read for a specific publisher

Remembered to actually limit to the top 10

Provides a top-10 most viewed datasets.

Fixing up setup.py and making sure routes exist for site and publisher reports

Fixed the publishr information (to enable leaderboard) to show the number
of children publishers and the total for all sub-publishers

Implemented further analytics by fetching more specific items.

Now fetches:
[x] Pages per visit
[x] Average visit duration
[x] % new visits
[x] Bounce rate
[x] Browser distribution
[x] IE version distribution (can work out from Browser stats)
[x] OS distribution
[x] Language distribution
[x] Location distribution
[x] Mobile device distribution
[x] Top twitter referrals

Removed some unused commented out code

Remove prettyprinting unless DEBUG is set in env and add a new
test for actually making the calls.