agiliq / django-graphos

Django charting made *really* easy.
http://agiliq.com/demo/graphos
BSD 2-Clause "Simplified" License
441 stars 98 forks source link

Graphos

Build Status

Graphos is a Django app to normalize data to create beautiful charts. It provides a JS agnostic way to work with charts and allows seamless and quick switching between different chart providers.

Supported Backends:

Charting API Supported

Chart types supported

Flot

Google Charts

YUI

Morris.js

Highcharts

(You will need to buy a license if you use highcharts for commerical use)

C3.js

Matplotlib

With Graphos, switching from google's LineChart to yui LineChart can be done within minutes. So would be the case in switching from yui AreaChart to morris AreaChart.

Running demo project locally

The installed demo app shows the various suported chart types.

In case you want to use mongo data while charting, you must have mongodb properly setup and pymongo installed. Make sure mongo server is running.

mongod --dbpath ~/data/db

Mongo setup is optional and is not needed to get running with demo project.

Overview of Plot generation

Generating a plot requires two things. A DataSource object and a Chart object.

In your view, you do something like this:

from graphos.sources.simple import SimpleDataSource
from graphos.renderers.gchart import LineChart

data =  [
        ['Year', 'Sales', 'Expenses'],
        [2004, 1000, 400],
        [2005, 1170, 460],
        [2006, 660, 1120],
        [2007, 1030, 540]
    ]
# DataSource object
data_source = SimpleDataSource(data=data)
# Chart object
chart = LineChart(data_source)
context = {'chart': chart}
return render(request, 'yourtemplate.html', context)

And then in the template:

{{ chart.as_html }}

In this example we are planning to use Google chart, as is evident from the import statement in the view, we import gchart.LineChart. So we must also include the google chart javascript in our template.

<script type="text/javascript" src="https://www.google.com/jsapi"></script>
<script type="text/javascript">
    google.load("visualization", "1", {packages:["corechart"]});
</script>

So the template would look like

<script type="text/javascript" src="https://www.google.com/jsapi"></script>
<script type="text/javascript">
    google.load("visualization", "1", {packages:["corechart"]});
</script>

{{ chart.as_html }}

If we want to use yui LineChart instead of google LineChart, our view would have:

from graphos.renderers.yui import LineChart
chart = LineChart(data_source)

And our template would inclue yui javascript and it would look like:

<script src="http://yui.yahooapis.com/3.10.0/build/yui/yui-min.js"></script>
{{ chart.as_html }}

See, how easy it was to switch from gchart to yui. You did not have to write or change a single line of javascript to switch from gchart to yui. All that was taken care of by as_html() of the chart object.

DataSources

SimpleDataSource

This should be used if you want to generate a chart from Python list.

from graphos.sources.simple import SimpleDataSource
data = SimpleDataSource(data=data)

Data could be:

data = [
       ['Year', 'Sales', 'Expenses', 'Items Sold', 'Net Profit'],
       ['2004', 1000, 400, 100, 600],
       ['2005', 1170, 460, 120, 710],
       ['2006', 660, 1120, 50, -460],
       ['2007', 1030, 540, 100, 490],
       ]

or it could be

data = [
       ['Year', 'Sales', 'Expenses'],
       ['2004', 1000, 400],
       ['2005', 1170, 460],
       ['2006', 660, 1120],
       ['2007', 1030, 540],
       ]

or it could be

data = [
       ['Year', 'Sales', 'Expenses'],
       ['2004', 1000, 400],
       ['2005', 1170, 460],
       ]

You got the idea.

data has to be a list of lists. First row of data tells the headers. First element of each list elementis the x axis.

This data essentially tells that in year 2004, sales was 1000 units and expense was 400 units. And in year 2005, sales was 1170 units and expense was 460 units.

ModelDataSource

This should be used if you want to generate a chart from a Django queryset.

from graphos.sources.model import ModelDataSource
queryset = Account.objects.all()
data_source = ModelDataSource(queryset,
                              fields=['year', 'sales'])

This assumes that there is a Django model called Account which has fields year and sales. And you plan to plot year on x axis and sales on y axis.

Or you could say

data_source = ModelDataSource(queryset,
                              fields=['year', 'sales', 'expenses'])

This would plot the yearly sale and yearly expense

CSVDataSource

This should be used if you want to generate a chart from a CSV file.

from graphos.sources.csv_file import CSVDataSource
csv_file = open("hello.csv")
data_source = CSVDataSource(csv_file)

MongoDataSource

TODO

Charts

We have following charts

Most of the chart providers support LineChart, BarChart, ColumnChart and PieChart, and it is very easy to switch from specific chart type of one provider to other. eg: It is super quick to switch from gchart LineChart to flot LineChart.

More Examples

Using SimpleDataSource with gchart LineChart

View
data =  [
        ['Year', 'Sales', 'Expenses'],
        [2004, 1000, 400],
        [2005, 1170, 460],
        [2006, 660, 1120],
        [2007, 1030, 540]
    ]
from graphos.sources.simple import SimpleDataSource
from graphos.renderers.gchart import LineChart
chart = LineChart(SimpleDataSource(data=data))
Template
<script type="text/javascript" src="https://www.google.com/jsapi"></script>
<script type="text/javascript">
    google.load("visualization", "1", {packages:["corechart"]});
</script>

{{ chart.as_html }}

Using SimpleDataSource with yui LineChart

View
data =  [
        ['Year', 'Sales', 'Expenses'],
        [2004, 1000, 400],
        [2005, 1170, 460],
        [2006, 660, 1120],
        [2007, 1030, 540]
    ]
from graphos.sources.simple import SimpleDataSource
from graphos.renderers.yui import LineChart
chart = LineChart(SimpleDataSource(data=data))
Template
<script src="http://yui.yahooapis.com/3.10.0/build/yui/yui-min.js"></script>
{{ chart.as_html }}

Using SimpleDataSource with yui BarChart

View
data =  [
        ['Year', 'Sales', 'Expenses'],
        [2004, 1000, 400],
        [2005, 1170, 460],
        [2006, 660, 1120],
        [2007, 1030, 540]
    ]
from graphos.sources.simple import SimpleDataSource
from graphos.renderers.yui import BarChart
chart = BarChart(SimpleDataSource(data=data))
Template
<script src="http://yui.yahooapis.com/3.10.0/build/yui/yui-min.js"></script>
{{ chart.as_html }}

Using SimpleDataSource with gchart BarChart

View
data =  [
        ['Year', 'Sales', 'Expenses'],
        [2004, 1000, 400],
        [2005, 1170, 460],
        [2006, 660, 1120],
        [2007, 1030, 540]
    ]
from graphos.sources.simple import SimpleDataSource
from graphos.renderers.gchart import BarChart
chart = BarChart(SimpleDataSource(data=data))
Template
<script type="text/javascript" src="https://www.google.com/jsapi"></script>
<script type="text/javascript">
    google.load("visualization", "1", {packages:["corechart"]});
</script>
{{ chart.as_html }}

Options

Your rendered chart is contained in a div.

Setting id of chart containing div

You might want to do additional jquery or javascript operations with your chart containing div. In such case you might want to set an id on the div. You can do this while instantiating the chart element.

chart = gchart.LineChart(html_id='gchart_div')

Setting width and height of chart containing div

You can control the width and height of chart containing div while instantiating the chart element.

chart = gchart.LineChart(simple_data_source, height=100, width=100)

Chart specific options

Different chart providers give different options to customise the chart.

Google chart api allows setting title for the rendered chart, see Gchart documentation, using title attribute. You can accomplish this by adding a keyword argument called options while instantiating the chart element.

chart = gchart.LineChart(simple_data_source, height=100, width=100, options={'title': 'Sales growth'})

Google pie chart allows making the chart as 3 dimensional. You can accomplish this by using keyword argument options.

pie_chart = gchart.PieChart(simple_data_source, options={'is3D': True})

Morris.js allows options like lineWidth, smooth etc. You can find more here. You can accomplish this by using options.

chart = morris.LineChart(simple_data_source, options={'lineWidth': 50, 'smooth': False})

Installation

pip install django-graphos

Compatibility

Graphos is compatible with Python 2.7 and Python 3.3+

available on pypi

Handling non serializable fields

You need to override get_data() of existing DataSource and convert datetime field into something which could be serialized.

Assuming you are using a Python list as data, then you need to do:

from graphos.sources.simple import SimpleDataSource
class MySimpleDataSource(SimpleDataSource):
    def get_data(self):
        data = super(MySimpleDataSource, self).get_data()
        header = data[0]
        data_without_header = data[1:]
        for row in data_without_header:
            # Assuming first column contains datetime
            row[0] = row[0].year
        data_without_header.insert(0, header)
        return data_without_header

And data has

d1 = datetime(2015, 7, 8, 1, 1)
d2 = datetime(2016, 7, 8, 3, 1)

data1 = [
         ['Year', 'Sales', 'Expenses', 'Items Sold', 'Net Profit'],
         [d1, 1000, 400, 100, 600],
         [d2, 1170, 460, 120, 310],
 ]

chart = flot.LineChart(MySimpleDataSource(data=data1))

If you are planning to use queryset with ModelDataSource, then you would create following class

from graphos.sources.model import ModelDataSource
class MyModelDataSource(ModelDataSource):
    def get_data(self):
        data = super(MyModelDataSource, self).get_data()
        header = data[0]
        data_without_header = data[1:]
        for row in data_without_header:
            # Assuming second column contains datetime
            row[1] = row[1].year
        data_without_header.insert(0, header)
        return data_without_header

And you would use this class like:

queryset = Account.objects.all()
chart = flot.LineChart(MyModelDataSource(queryset=queryset, fields=['sales', 'datetime_field','expenses']))

Creating new DataSource

A DataSource is a class which has these three methods.

get_data
get_header
get_first_column

get_header is used by a Renderer to create the labels. get_first_column is used to set the x axis labels get_data is used to get the data for display. It should always return a nested list. Eg:

[
    ['Year', 'Sales', 'Expenses'],
    [2004, 1000, 400],
    [2005, 1170, 460],
    [2006, 660, 1120],
    [2007, 1030, 540]
]

If you create a class extending SimpleDataSource, and implement get_data. You get get_header and get_first_column for free.

Creating new Renderer

A renderer is a class which takes a DataSource and can convert it to the html to display.

The only required method on a Renderer is as_html. This will convert the data to a format which can display the chart.

Generally you will convert the data to json and pass it to the template which you return.

License

BSD