bintime.py

TIMEUNIT_MICROSECOND = 0
TIMEUNIT_SECOND = 1
TIMEUNIT_MINUTE = 2
TIMEUNIT_HOUR = 3
TIMEUNIT_DAY = 4
TIMEUNIT_MONTH = 5
TIMEUNIT_YEAR = 6

def round_datetime(timeunit):
    '''Round the given time downward to some unit of time.  For example:
    'round_datetime(TIMEUNIT_YEAR)(2015-12-12)' will yield 2015-01-01.'''
    def inner(t):
        if timeunit > TIMEUNIT_MICROSECOND:
            t = t.replace(microsecond=0)
        if timeunit > TIMEUNIT_SECOND:
            t = t.replace(second=0)
        if timeunit > TIMEUNIT_MINUTE:
            t = t.replace(minute=0)
        if timeunit > TIMEUNIT_HOUR:
            t = t.replace(hour=0)
        if timeunit > TIMEUNIT_DAY:
            t = t.replace(day=1)
        if timeunit > TIMEUNIT_MONTH:
            t = t.replace(month=1)
        return t
    return inner

def bin_events(events, rounder):
    '''Read a sequence of temporal events and bin them according to the
    rounding function provided by 'rounder'.

    The 'events' argument should a list of pairs '(time, weight)'.  If each
    event is counted equally, then weight should be just 1.

    The 'rounder' argument must a function that rounds the time to some
    discrete scale.
    '''
    bins = {}
    for time, weight in events:
        binned_time = rounder(time)
        if binned_time not in bins:
            bins[binned_time] = 0
        bins[binned_time] += weight
    return sorted(bins.items(), key=(lambda x: x[0]))

def transpose(lists):
    return zip(*lists)

import csv, datetime
import matplotlib.pyplot as plt

with open("downloads/chat.csv", "rt") as f:
    reader = csv.reader(f)
    next(reader)
    fmt = "%Y-%m-%dT%H:%M:%S"
    events = [(datetime.datetime.strptime(r[3], fmt), 1) for r in reader]

xs, ys = transpose(bin_events(events, round_datetime(TIMEUNIT_DAY)))
plt.fill_between(xs, 0, ys, color="#9b19e6", linewidth=.2)
plt.show()