Number of new workflows submitted per year in the myExperiment platform

myExperiment.py

# Fetch data with:
# wget -O myExperiment.html "http://www.myexperiment.org/workflows?num=10000" 
# Resulted plot: http://imgur.com/a/7Zzlu 

import re
import datetime

import matplotlib.pyplot as plt

import seaborn

from bs4 import BeautifulSoup

fn = "myExperiment.html"

with open(fn) as f:
	content = f.read()
	soup = BeautifulSoup(content, "lxml")


data =[]
entries = soup.find_all(class_="main_panel")
for e in entries:

	created = None
	updated = None

	found = re.search(r"Created: ([\d]+)\-([\d]+)\-([\d]+)",    e.find_all('p')[1].text.split('\n')[1])
	if found:
		created_year = int(found.group(1))
		created_month = int(found.group(2))
		created_day = int(found.group(3))

		#print created_year, created_month, created_day

		created = datetime.datetime(created_year, created_month,  created_day, 0, 0, 0) 	

	found = re.search(r"Last updated: ([\d]+)\-([\d]+)\-([\d]+)",    e.find_all('p')[1].text.split('\n')[2])
	if found:
		updated_year = int(found.group(1))
		updated_month = int(found.group(2))
		updated_day = int(found.group(3))

		#print updated_year, updated_month, updated_day
		updated = datetime.datetime(updated_year, updated_month,  updated_day, 0, 0, 0) 	

	if created and updated:
		data.append((created, updated))
	elif created:
		data.append((created, None))
	else:
		print e
		a=1/0

epoch = datetime.datetime.utcfromtimestamp(0)

fig, ax = plt.subplots()

if True:
	to_plot = {}
	for d in data:
		year = d[0].year
		to_plot[year] = to_plot.get(year, 0) + 1

	data_2 = sorted(list(to_plot.iteritems()))
	print data_2
	ax.plot([x[0] for x in data_2], [x[1] for x in data_2], color="black")

	ticks = ax.get_xticks().tolist()
	new_ticks = [str(int(x)) for x in ticks]
	ax.set_xticklabels(new_ticks)
	ax.set_ylabel('Number of new Workflows')

	plt.show()