simple visualisation of twofactorauth.org data with python

data_scraping_and_vis.py

import yaml

# load the sections descriptions
sections = open("sections.yml", "r")
docs = yaml.load(sections)
doc_list = [i['id'] for i in docs]
sections.close()

# read the single files
data = {}
for section in doc_list:
    section_file = open(section+'.yml', "r")
    section_data = yaml.load(section_file)
    data[section] = section_data['websites']
    section_file.close()

# collect the data
plot_data = {}
plot_data['len']={}
plot_data['yes']={}
plot_data_yes = {}
for k,v in data.iteritems():
    plot_data['len'][k]=len(v)
    plot_data['yes'][k]=sum(( i['tfa'] for i in v ))

import matplotlib
from matplotlib import pyplot as plt
%matplotlib

plt.bar(range(len(plot_data['len'])), plot_data['len'].values(), align='center',color='red',width=1, lw=0)
plt.bar(range(len(plot_data['yes'])), plot_data['yes'].values(), align='center',color='green',width=1,lw=0)
plt.gca().yaxis.set_major_locator(plt.NullLocator())
plt.xticks(range(len(plot_data['len'])), [i[0:] for i in plot_data['len'].keys()], rotation=90);
plt.xlim([-0.5,len(plot_data['len'])-0.5])
plt.tight_layout()

plt.savefig('yes_no.png')


plt.bar(range(len(plot_data['len'])), [1]*len(plot_data['len']), align='center',color='red',width=1, lw=0)
plt.bar(range(len(plot_data['yes'])),  [y/float(l) for l,y in zip(plot_data['len'].values(),plot_data['yes'].values())], align='center',color='green',width=1,lw=0)
plt.gca().yaxis.set_major_locator(plt.NullLocator())
plt.xticks(range(len(plot_data['len'])), [i[0:] for i in plot_data['len'].keys()], rotation=90);
plt.xlim([-0.5,len(plot_data['len'])-0.5])
plt.tight_layout()

plt.savefig('yes_no_perc.png')