According to notes in attached issue, pie charts, grand total and shares do not make much sense for multivalue statistics. This commit replaces these visualisations with simple horizontal bar charts with absolute counts, shares in data table are calculated against event total. I have stumbled upon hideous gotcha when working with NVD3 library. For some reason authors choose 'values' as a dictionary subkey containing data series. This however collides with built-in values javascript method. I have spent quite some time trying to figure out why am I not able generate the data structure for bar chart programmatically. (Redmine issue: #6308)

Custom CSS, JS and image files were moved to application`s static folder. (Redmine issue: #3443)

Context search action groups are now more specific for searching from event dashboards and report detail, the dt_from and dt_to parameters are now pre-filled with available values. This patch also fixes the long lasting bug in DateTime search form fields. (Redmine issue: #5341,#4518)

(Redmine issue: #4486)

Chart toplists for dashboards are now calculated in the frontend in Javacript instead of in background in Python. This approach will enable users to access the full information, which is necessary in some use cases. (Redmine issue: #4486)

The current implementation is flawed and is holding back the release of the next version. I am rejecting the code and will revisit the idea sometime in the future. (Redmine issue: #4321)

This commit introduces new group of timescatter visualisations. It is still work in progress and it is being deployed prematurely for demonstartion purposes. The chart capabilies are being demonstrated on reporting dashboard to show the number of reports being generated in different time of day and days. Many time consuming features were not yet implemented like exporting to CSV and rendering dataset table, I would not be wise to spend time implementing feature, that will be rejected in the end. (Redmine issue: #4321)

NVD3 charting library has a drawback, that it does not render correctly charts, that are initially hidden from the user, for example in tab panels. These charts are not re-rendered on tab show and have to be updated manually. Previous solution used simple mechanism of emiting window.resize event, which was captured by all charts that were currently on the page. This approach could be very slow on some views with many charts and it was very inefficient, because all chart were updated regardless of their visibility to the user. New approach is much smarter, because it updates only those charts, that are necessary. There is still room for improvement though, it does not work optimally for nested tabs. This drawback will be fixed in some of the following commits. (Redmine issue: #4321)

After previous commits the database status view module was broken, because it was still using old charting code. This is fixed in this commit. Additionally, this commit introduces new ability to define custom value label and formatter for dict based pie datasets. The implementation of this feature was necessary to enable displaying values in tables as size in bytes instead of simple count (which was necessary for datasets showing sizes of database tables). (Redmine issue: #4321)

Certain key categories (IPs, abuses, ...) in pie chart dataset tables have associated context search actions that provide user with ability to quickly search different parts of the application. (Redmine issue: #4321)

(Redmine issue: #4321)

It has special meaning, it does not belong to any data series. (Redmine issue: #4321)

Added missing pie charts containing totals to all timeline charts. Hidden the abuses panel from external processing statistics display. (Redmine issue: #4321)

(Redmine issue: #4321)

Improved all related Jinja2 macros. Moved as much functionality as possible to plain JavaScript code to reduce the amount of code that was being parametrized and generated from Jinja2 templates. (Redmine issue: #4321)

Improved all related Jinja2 macros. Moved as much functionality as possible to plain JavaScript code to reduce the amount of code that was being parametrized and generated from Jinja2 templates. (Redmine issue: #4321)

Instead of coloring the call background insert small circle icon with color corresponding with related dataset. (Redmine issue: #4321)

In previous implementation the pie charts and their apropriate dataset tables were generated separately. Now the table is also generated by D3 and all of this is neatly encapsulated in Jinja2 macros similarly to rendering of timeline charts. (Redmine issue: #4321)

(Redmine issue: #4321)

Minor JS code optimalizations and documentation and style improvements. (Redmine issue: #4321)

(Redmine issue: #4321)

Table footer now contains following information: cnt, min, max, sum, avg, median for each of the data series in the table dataset. (Redmine issue: #4321)

The timeline chart dataset tables now have localized header titles and numbers in the table are localized with Globalize. (Redmine issue: #4321)

Implemented prototype of collapsible dataset view and downloadable dataset option for timeline charts.

All timeline charts in reporting dashboard and timeline module now support displaying of chart datasets in tables within collabsible sidebar. Additionaly the chart dataset can now be downloaded as CSV file for easy import into table processor software like LibreOffice. (Redmine issue: #4321)

Previously the JS code for charts was generated from Jinja2 templates. New approach uses chart factory functions in separate JS file and much more simple Jinja2 templates. (Redmine issue: #4321)

Unnecessary optimalization. (Redmine issue: #4321,#3443)

Improved the calculations of time step size for timeline charts. Improved the format of rendered time stamp for chart time scale. (Redmine issue: #4321)