create streamgraphs from delicious history

posts.py

#!/usr/bin/env python

import lxml.etree
import xml.utils.iso8601
from datetime import datetime, timedelta
from graphication import FileOutput, Series, SeriesSet, Label, SimpleScale, css, default_css as style
from graphication.scales.date import AutoWeekDateScale, AutoDateScale
from graphication.wavegraph import WaveGraph
"""
  curl https://user:[email protected]/v1/posts/all > posts.xml
  ./posts.py 
"""
def group_period(list, threshold=10):
  grouped = {}
  for time, tags in list:
    year_week = "%s-%s" % (time.isocalendar()[0], time.isocalendar()[1])
    for t in tags:
      if not t in grouped: grouped[t] = {}
      if not year_week in grouped[t]: grouped[t][year_week] = 0
      grouped[t][year_week] += 1


  sorted = [(t, grouped[t]) for t in grouped]
  sorted.sort(lambda x,y: cmp(len(y[1]), len(x[1])))
  return sorted[0:threshold]

def parse(file='posts.xml'):
  return [[datetime.fromtimestamp(xml.utils.iso8601.parse(p.get('time'))), p.get('tag').split()] for p in lxml.etree.XML(open(file).read())]

def colorgen(colours):
  while 1:
    for c in colours: yield c


def get_series(grouped):
  colour_loop = colorgen(
       ("#4D8963",
       "#69A583",
       "#E1B378",
       "#E0CC97",
       "#EC799A",
       "#9F0251")
  )
  series_set = SeriesSet()

  for (t, list) in grouped:
    data = dict([(datetime.strptime(k+'0', "%Y-%W%w") ,v) for (k,v) in list.iteritems()])
    series_set.add_series(Series(t, data, colour_loop.next()))

  return series_set


def create_graph(filename="tags", threshold=15):
  series_set = get_series(group_period(parse(), threshold=threshold))
  scale = AutoWeekDateScale(series_set)
  output = FileOutput()
  output.add_item(WaveGraph(series_set, scale, None, label_curves=True), x=0, y=30, width=20*len(series_set.keys()), height=500)
  output.write("png", ("%s.png" % filename))
  output.write("pdf", ("%s.pdf" % filename))
                                                   
create_graph(threshold=10)