from pylab import *
import re, datetime


fig = figure()
fig.subplots_adjust(bottom=0.2)

day_zero = datetime.date(2000, 1, 1)

data_lines = open('out_comments_time.txt').read().splitlines() + open('out_posting_time.txt').read().splitlines()
data = []

for l in data_lines:
    l = l.strip()
    if not l:
        continue
    m = re.match('(xxxx)-(xx)-(xx) (xx):(xx):(xx)'.replace('x', '\\d'), l)
    assert m, repr(l)
    g = lambda i: int(m.group(i))
    if g(1) < 1500:
        continue
    dt = (datetime.date(g(1), g(2), g(3)) - day_zero)
    data.append((
        (dt.seconds + dt.days * 24 * 3600) / (365.25 * 24 * 60 * 60),
        (60*60*g(4) + 60*g(5) + g(6)) / 3600.0
    ))
data.sort()

print min(data), max(data)


grid()
plot([a[0] for a in data], [a[1] for a in data],'bo', alpha=0.3)

show()