Simple Bayesian classifier

import sys

from text.classifiers import NaiveBayesClassifier


# Utility for coloring
def col(line, color):
    return '\033[%sm%s\033[0m' % (color, line)

# Quick util for pretty command line output
def ln(metric, value, space=40):
    color = 31 if value is 'bullshit' else 32
    s = '{0: <' + str(space) + '}'
    print s.format('- %s' % metric) + col(value, color)

train = []

# Train dat shit
def trainer(content, directory):

    if '-d' in sys.argv: print os.listdir(directory)

    for file in os.listdir(directory):
        if file == '.DS_Store': continue
        if '-d' in sys.argv: print file

        with open('%s/%s' % (directory, file)) as f: r = f.read()
        content.append( (r, directory) )

    return content

train = trainer(train, 'legit')
train = trainer(train, 'bullshit')

cl = NaiveBayesClassifier(train)

print 'Accuracy: %s' % col(cl.accuracy(train), 93)
print 'Features:'
print cl.show_informative_features(40)

# Run ze trials
for file in os.listdir('trials'):
    if file == '.DS_Store': continue
    with open('trials/' + file) as f: r = f.read()
    ln(file, cl.classify(r))