def generate(sentences=SENTENCES_DEFAULT):
# print(repr(request.args['s']))
sentences=int(sentences)
chain = ScanResults()
chain.add_files(CORPUS_GLOB)
chain.train(n=TRAIN_N_DEFAULT, paragraphs=TRAIN_PARA_DEFAULT)
if chain.errors:
return render_template('error.html', errors=[repr(item) for item in chain.errors])
return render_template('output.html', output=chain.sentences(n=sentences))
from scanresults import ScanResults
import sys
from argparse import ArgumentParser
parser = ArgumentParser(description='Generate markov text from a corpus')
parser.add_argument('-t','--train', default=2, type=int, metavar='train_n',
help="The size of the Markov state prefix to train the corpus on")
parser.add_argument('-s','--sentences', default=25, type=int, metavar='sentences',
help="The number of sentences to print")
parser.add_argument('corpus_path', nargs="+",
help="Paths to corpus files to train on. Globs are permitted")
args = parser.parse_args()
sr = ScanResults()
sr.add_files(*args.corpus_path)
sr.train(args.train)
out = sr.sentences(args.sentences)
if out is not None:
print(out)