#!/usr/bin/env python
# - Make this pretty and put it on the Internet
# - Highlight the modules written in C in red.
# - Try and draw an area plot
# - Perform similar queries using Cheese shop / google search results


import pylab
from pylab import *
from operator import itemgetter

f = file('results.csv')
freqs = []
labels = []
for line in f.readlines():
    try:
        if line.strip().startswith("#"):
            continue
        label, freq = line.strip().split(",")
        freq = int(freq)
        labels.append(label)
        freqs.append(freq)
    except:
        print line
        raise

pairs = list(reversed(sorted(zip(freqs, labels), reverse=True)[0:300]))
freqs = map(itemgetter(0), pairs)
labels = map(itemgetter(1), pairs)

name_locations = range(1, len(freqs) + 1)
width = 0.5

barh(name_locations, freqs, height=width, align='center')
yticks(name_locations, labels)

fstep = int(round(10**floor(log10(max(freqs))))) // 2 
ticks = range(0, (max(freqs) + fstep - 1) // fstep * fstep + fstep, fstep)
frequence_labels = ticks
xticks(ticks, frequence_labels)

ylim(0, name_locations[-1]+width*2)

title(r"""Frequency of occurrences of the regular expression
'(from|import)\s<module name>\b' found by google code for
python modules in the standard library - top 300.""", horizontalalignment='center')

xlabel("Frequency")
ylabel("Module")

gca().get_xaxis().tick_top()
gca().get_xaxis().set_label_position('top')

gca().get_yaxis().tick_left()

gcf().set_size_inches(12, 60)
gca().set_position([0.25, 0.04, 0.5, 0.91])

savefig('bar-chart.png')