Python: NLTK bigram and trigram filter example

with open('_sorted.csv','r') as f:
    lines = f.read().replace('\t',',')

lines = lines.split('\n')
lines = [l.split(',')[12:] for l in lines]
lines = [item for l in lines for item in l if ("domain.com" not in item)]
lines = " ".join(lines)
lines = lines.split(' ')

#---------------------------------
# requires nltk v3.0
#-------------------------------
import nltk
from nltk.collocations import BigramAssocMeasures, TrigramAssocMeasures, BigramCollocationFinder, TrigramCollocationFinder, QuadgramCollocationFinder
from nltk.util import ngrams
#from nltk.metrics.association import QuadgramAssocMeasures
#n = 4
#quadgram = ngrams(lines, n) huge set that can be counted and sorted works for any N size
# abstract_measure = nltk.collocations.AbstractCollocation???


bigram_measures = BigramAssocMeasures()
trigram_measures = TrigramAssocMeasures()
#quadgram_measures = QuadgramAssocMeasures()


finder2 = BigramCollocationFinder.from_words(lines)
#finder21 = BigramCollocationFinder.from_words(nltk.corpus.brown.tagged_words('ca01', tagset='universal')) # finds tagged words
# r21 = finder21.nbest(bigram_measures.pmi, 5)
finder3 = TrigramCollocationFinder.from_words(lines)
finder4 = QuadgramCollocationFinder.from_words(lines)

finder2.apply_freq_filter(5)
finder3.apply_freq_filter(5)
finder4.apply_freq_filter(5)
# r2 = sorted(finder2.ngram_fd.items(), key=lambda t: (-t[1], t[0]))[:-1] # (("w1","w2"),int(freq))

r2 = finder2.nbest(bigram_measures.pmi, 100)
r3 = finder3.nbest(trigram_measures.pmi, 100)
r4 = sorted(finder4.ngram_fd.items(), key=lambda t: (-t[1], t[0]))[:-1]
#$r4 = finder4.nbest(quadgram_measures.pmi, 100) ???
# r2 = [item for item in r2 if any(item not in _ for _ in r3)]
# r3 = [item for item in r3 if any(item not in _ for _ in r4)]

lines = []
for item in r2:
    lines.append(' '.join([str(x) for x in item]+['\n']))
for item in r3:
    lines.append(' '.join([str(x) for x in item]+['\n']))
for item in r4:
    lines.append(' '.join([str(x) for x in item]+['\n']))
with open('token_results.txt','w+') as f:
    for l in lines:
        f.write(l.replace("', '","\t").lstrip("('").replace("')", "\t"))