Plot the 25 most common domains found in Safari's history cache using matplotlib

safarihistory.py

import os
import urllib2
from collections import Counter


#FETCH DATA

files=os.listdir(os.path.expanduser("~/Library/Caches/Metadata/Safari/History"))
webpages=[f for f in files if f.startswith("http")] #https also starts with http
print "Found {} pieces of data".format(len(webpages))
webpages=[f[:-11] for f in webpages] #Remove .webhistory extension
visited=[urllib2.unquote(w) for w in webpages] #Undo HTML URL encoding

#Now we have a list of all the webpages stored in history.
domains=[urllib2.urlparse.urlparse(page).netloc for page in visited] #parse each URL to find the domain it's from.
count=Counter(domains)
tops=count.most_common(25) #Top 25 visited domains



#HISTOGRAM

from matplotlib import pyplot as plt

items,counts=zip(*tops) #List of items, list of times each appears
indices=range(len(tops)) #Index for each item

#Graph the graph
graph=plt.bar(indices, counts, 1, edgecolor="#FF3300", facecolor="#FF7700")

#Label stuff
xvals=[i+0.5 for i in indices] #Where each label will appear
plt.xticks(xvals, items,rotation="vertical") #Label the bars
plt.title("Top 25 Domains") # Title the graph

#Label the tops of the bars
for rect in graph:
    x=rect.get_x()+0.6
    y=rect.get_height() + 50
    label=str(rect.get_height())
    plt.text(x,y,label,ha="center",va="bottom",rotation="vertical")

#Increase spacing
plt.ylim((0.0,4000.0))

#Save an image
plt.savefig("topsites.png",bbox_inches="tight")