[PYTHON] Display matplotlib diagrams in a web application

Right now, I think that the JavaScript library will probably make cool graphs, but no, I'm not sure, but I still want to use the familiar matplotlib in web applications.

However, since matplotlib draws using X11, if you think that it may be difficult to use in a web application, that is not the case.

Create image data using AGG (Anti-Grain Geometry) for the back end.

things to do

Draw a graph created with matplotlib in a flask web application. The reason for using flask is because it is convenient, so I think that you can do the same if you do the same without using it.

Also, do the following two patterns.

• png Return data in response
• Create an image file as a temporary file

Return png data in response

Make a chart

For the time being, I need a chart to display something, so

• x-axis: numbers from 1 to 284
• y-axis: x is randomly multiplied by 436 to 875

I will make a graph.

These numbers don't seem to have a deep meaning, but they don't. The title of the graph should be "IMINASHI GRAPH".

fig, ax = matplotlib.pyplot.subplots()
ax.set_title(u'IMINASHI GRAPH')
x_ax = range(1, 284)
y_ax = [x * random.randint(436, 875) for x in x_ax]
ax.plot(x_ax, y_ax)

Like this.

Display in web app

The processing flow is like this.

1. Create png data for charts
2. Write the data to the buffer of cStringIO.StringIO
3. Return data as image content in response

Maybe it's faster to look at the code.

@app.route('/graph1')
def graph1():
    import matplotlib.pyplot
    from matplotlib.backends.backend_agg import FigureCanvasAgg
    import cStringIO
    import random

    fig, ax = matplotlib.pyplot.subplots()
    ax.set_title(u'IMINASHI GRAPH')
    x_ax = range(1, 284)
    y_ax = [x * random.randint(436, 875) for x in x_ax]
    ax.plot(x_ax, y_ax)

    canvas = FigureCanvasAgg(fig)
    buf = cStringIO.StringIO()
    canvas.print_png(buf)
    data = buf.getvalue()

    response = make_response(data)
    response.headers['Content-Type'] = 'image/png'
    response.headers['Content-Length'] = len(data)
    return response

Now, if you access it from a web browser, you will see the image.

Create an image file as a temporary file

Another one is this one.

The process is like this.

1. Export the chart as a png file
2. Return png file in response
3. Delete the png file

The interesting one would be 3. Considering simultaneous access from multiple people, I think it is better to use a random character string for the file name. However, in that case, the number of files will increase steadily, so you have to delete them properly after displaying the files.

Use the same graph as before. The title of the graph should be "IMINASHI GRAPH 2".

@app.route('/graph2')
def graph2():
    import matplotlib.pyplot
    from matplotlib.backends.backend_agg import FigureCanvasAgg
    import random
    import string
    import os

    class TempImage(object):

        def __init__(self, file_name):
            self.file_name = file_name

        def create_png(self):
            fig, ax = matplotlib.pyplot.subplots()
            ax.set_title(u'IMINASHI GRAPH 2')
            x_ax = range(1, 284)
            y_ax = [x * random.randint(436, 875) for x in x_ax]
            ax.plot(x_ax, y_ax)

            canvas = FigureCanvasAgg(fig)
            canvas.print_figure(self.file_name)

        def __enter__(self):
            return self

        def __exit__(self, exc_type, exc_value, traceback):
            os.remove(self.file_name)

    chars = string.digits + string.letters
    img_name = ''.join(random.choice(chars) for i in xrange(64)) + '.png'

    with TempImage(img_name) as img:
        img.create_png()
        return send_file(img_name, mimetype='image/png')

There are several possible ways to delete it after returning a response, but in this example we use the context manager to destroy it. This is the most Python-like image.

It was displayed. It may be easier to use if it is made into a file.