text_model.make_sentence() always outputs none, even with a script running it for an infinite number of tries.

[Id405]

import markovify

with open("data", encoding="utf-8") as f:
    text = f.read()
text_model = markovify.Text(text, state_size=7)

txt = []
while len(txt) in range(0,5):
        txt.append(text_model.make_sentence(tries=100))
        txt = [x for x in txt if x is not None]

print(txt)

This script just hangs on my screen. Reinstalling the library did nothing for me, trying different corpus did not work either.

[theIsaacLim]

you need to specify the length of the statement
ie. text_model.make_sentence(140, tries=100)

[jsvine]

@8o7wer's code above seems to work fine on large corpora. E.g., the sherlock.txt test text. The hitch seems to be the specification of state_size = 7. That's a fairly large state size, and will only be able to generate new sentences with fairly large corpora. (I.e., those which contain many examples of the same seven-word sequences.)

If you reduce the state_size to a smaller number — e.g., 2 or 3 — it should work fine on most corpora.