def test_double_word_not_found(self):
# It should look for a match to both a & c first,
# and when it doesn't find anything, should return the match for c
mc = MarkovChain(self.corpus, 2)
mc.fit()
self.assertEqual(mc.next_word(['a','c']),'a')
self.assertEqual(mc.next_word(['alef','gimel']),'alef')
class MCMarkov():
def __init__(self, corpus, n_order=1, reverse=True):
if reverse:
self.corpus = [line[::-1] for line in corpus]
else:
self.corpus = corpus
self.markovchain = MarkovChain(self.corpus, n_order)
self.markovchain.fit()
self.starting_words = [line[0] for line in self.corpus]
def test_single_word(self):
mc = MarkovChain(self.corpus, 1)
mc.fit()
# First line
self.assertEqual(mc.next_word(['a']),'b')
self.assertEqual(mc.next_word(['b']),'c')
self.assertEqual(mc.next_word(['c']),'a')
# Second line
self.assertEqual(mc.next_word(['alef']),'bet')
self.assertEqual(mc.next_word(['bet']),'gimel')
self.assertEqual(mc.next_word(['gimel']),'alef')
def test_single_word(self):
mc = MarkovChain(self.corpus, 1)
mc.fit()
# First line
self.assertEqual(mc.next_word(["a"]), "b")
self.assertEqual(mc.next_word(["b"]), "c")
self.assertEqual(mc.next_word(["c"]), "a")
# Second line
self.assertEqual(mc.next_word(["alef"]), "bet")
self.assertEqual(mc.next_word(["bet"]), "gimel")
self.assertEqual(mc.next_word(["gimel"]), "alef")
def test_double_word_found(self):
# It should look for a match to both a & b first,
mc = MarkovChain(self.corpus, 2)
mc.fit()
self.assertEqual(mc.next_word(['a','b']),'c')
self.assertEqual(mc.next_word(['b','c']),'a')
self.assertEqual(mc.next_word(['alef','bet']),'gimel')
self.assertEqual(mc.next_word(['bet','gimel']),'alef')
def test_double_word_found(self):
# It should look for a match to both a & b first,
mc = MarkovChain(self.corpus, 2)
mc.fit()
self.assertEqual(mc.next_word(["a", "b"]), "c")
self.assertEqual(mc.next_word(["b", "c"]), "a")
self.assertEqual(mc.next_word(["alef", "bet"]), "gimel")
self.assertEqual(mc.next_word(["bet", "gimel"]), "alef")
def __init__(self, corpus, n_order=1, reverse=True):
if reverse:
self.corpus = [line[::-1] for line in corpus]
else:
self.corpus = corpus
self.reverse = reverse
self.markovchain = MarkovChain(self.corpus, n_order)
self.markovchain.fit()
self.starting_words = [line[0] for line in self.corpus if line]
self.rhymedict = {}
for word in self.starting_words:
keys = rhymesyls(word)
for key in keys:
if key in self.rhymedict:
self.rhymedict[key].append(word)
else:
self.rhymedict[key] = [word]
self.words_to_rhyme = {}
for key in self.rhymedict.keys():
if len(set(self.rhymedict[key]))>1:
self.words_to_rhyme[key] = self.rhymedict[key]
class MCMarkov():
def __init__(self, corpus, n_order=1, reverse=True):
if reverse:
self.corpus = [line[::-1] for line in corpus]
else:
self.corpus = corpus
self.reverse = reverse
self.markovchain = MarkovChain(self.corpus, n_order)
self.markovchain.fit()
self.starting_words = [line[0] for line in self.corpus if line]
self.rhymedict = {}
for word in self.starting_words:
keys = rhymesyls(word)
for key in keys:
if key in self.rhymedict:
self.rhymedict[key].append(word)
else:
self.rhymedict[key] = [word]
self.words_to_rhyme = {}
for key in self.rhymedict.keys():
if len(set(self.rhymedict[key]))>1:
self.words_to_rhyme[key] = self.rhymedict[key]
def create_line(self, startingword, syllable_count): # To be changed to syllable count
line = [startingword]
remaining_syllable_count = syllable_count - nsyl(startingword)
i = 0
while remaining_syllable_count > 0:
if i == 0:
word = self.markovchain.next_word([startingword])
else:
word = self.markovchain.next_word([word])
if word == '\n': # If previous word has no following words, backup and start again
prevword = line.pop()
remaining_syllable_count += nsyl(prevword)
if nsyl(word) <= remaining_syllable_count:
line.append(word)
remaining_syllable_count -= nsyl(word)
i += 1
if self.reverse:
line = line[::-1]
return line
def create_song(self, couplet_count, syllable_count, syllable_map = None):
if syllable_map is not None:
couplet_count = len(syllable_map)
song = []
for i in range(0,couplet_count):
if syllable_map is not None:
line_map = syllable_map[i]
line1_syllable_count = line_map[0]
line2_syllable_count = line_map[1]
else:
line1_syllable_count = syllable_count
line2_syllable_count = syllable_count
rhymewords = set(self.words_to_rhyme[random.choice(self.words_to_rhyme.keys())])
thisrhymes = random.sample(rhymewords,2)
startingword1 = thisrhymes[0]
startingword2 = thisrhymes[1]
line1 = self.create_line(startingword1, line1_syllable_count)
line2 = self.create_line(startingword2, line2_syllable_count)
couplet = [line1, line2]
song.append(couplet)
return song
from markov.markov import MarkovChain
import pickle
corpus = pickle.load(open( "eminemcorpus.pickle", "rb" ) )
mc = MarkovChain(corpus, 4)
print "Fitting..."
mc.fit()
print "Fit complete!"
generated_lyrics = []
word = None
print('starting lyric generation!')
for i in range(0,50):
print i
if word:
# NOTE: this argument has to a be a list
word = mc.next_word([word])
print word
else:
word = mc.next_word()
print word
generated_lyrics.append(word)
print(generated_lyrics)
def test_no_match(self):
#TODO: Make it so the markov chain always generates something?
mc = MarkovChain(self.corpus, 1)
mc.fit()
self.assertRaises(ValueError, mc.next_word, ['d'])
def test_no_word(self):
# If no word is passed, markov should return something
# TODO: Test this is actually random?
mc = MarkovChain(self.corpus, 1)
mc.fit()
self.assertNotEqual(mc.next_word([]),None)
