HN Explorer

Ask HN: Best Python graphing and reporting libraries for web app

46 points by TJensen ↗ HN
I'm building a web site that will include multiple graphs and printable reports. The graphs are pretty simple line graphs (perhaps with multiple data sources). Ideally, the graphic library would be able to do some smoothing of the data.

The reports are also fairly simple (lists of data with said graphs).

What do you think the best graphing and reporting libraries (for the planned usage) out there are?

10 comments

[ 3.1 ms ] story [ 42.5 ms ] thread
cloug
Google Chart is a good choice : various graph types & facility of use. You may want to use a wrapper like pygooglechart
gourneau
I am a fan of the Open Flash Chart 2 project: http://teethgrinder.co.uk/open-flash-chart-2/

As the name implies, they are flash charts, what except JSON data. There is a wonderful Python lib that makes it very easy to use OFC in a Pythonic way: http://btbytes.github.com/pyofc2/

thesethings
I second this recommendation. If you want to see it in action (though with PHP), check out piwik.org (the open source (Google-esque) analytics program.
nailer
Agreed. PyOFC2's graphs have better interactivity than the other options - being able to see values at a specific point where the mouse cursor is is quite liked by my users. There are two distribution sites: one has far more sample code.
mace
Pycha (PYthon CHArts) is pretty good. It use the Cairo vector graphics library.

http://www.lorenzogil.com/projects/pycha/

loginx
I second this. Graphing libs fall into 2 camps:

- Flash and JS charts. Those are interactive and most of the processing takes place on the client side, but they cannot be embedded into email reports or PDF documents or anything else, and are very difficult for the average end-user to archive.

- Server-side generated images. Those are not interactive, but obviously can be embeded into anything and easily archived, so tend to be more flexible and require no extra dependencies, such as a fast JS engine or a plugin.

I've been working with both types for way too long, and I eventually settled on PyCha. I wrote a presets package for my apps that allow me to very quickly change every aspect of charts in bulk, and I was floored when I ran my charting benchmarks. It's also one of the few libraries that scale to a very high number of datapoints (provided you remove the line shadow).