def get_chunk_of_corpus(corpus):
"""
Returns a randomly selected piece of a poem from a text file.
:param corpus: text file
:return: some number of successive lines from the text file
"""
test = ''
lines = []
f = codecs.open(corpus, 'r', 'utf-8')
#Go to the randomly selected line
random_place = random.randint(1, helpers.file_length(corpus) - 10)
for i in range(random_place):
f.readline()
while len(lines) < 3:
test = f.readline()
#Go to the beginning of the next verse
while not test.isspace():
test = f.readline()
if test == '':
break
while test.isspace():
test = f.readline()
if test == '':
break
while not test.isspace() and len(lines) < 6:
lines.append(test.strip())
test = f.readline()
if test == '':
break
sent_end = ('.', '!', '?', ':', ';')
if not test.isspace():
if not lines[-1].endswith(sent_end):
while not test.isspace() and\
not lines[-1].endswith(sent_end):
lines.append(test.strip())
test = f.readline()
if test == '':
break
f.close
# Insert a dot to the end
punctuation = [',', ':', ';']
last = lines[-1]
if last[-1].isalpha():
last = last + '.'
elif last[-1] in punctuation:
last = last[:len(last) - 1] + '.'
lines[-1] = last
string = '\n'.join(lines)
return string
def get_chunk_of_corpus(corpus):
"""
Returns a randomly selected piece of a poem from a text file.
:param corpus: text file
:return: some number of successive lines from the text file
"""
test = ''
lines = []
f = codecs.open(corpus, 'r', 'utf-8')
#Go to the randomly selected line
random_place = random.randint(1, helpers.file_length(corpus)-10)
for i in range(random_place):
f.readline()
while len(lines) < 3:
test = f.readline()
#Go to the beginning of the next verse
while not test.isspace():
test = f.readline()
if test == '':
break
while test.isspace():
test = f.readline()
if test == '':
break
while not test.isspace() and len(lines) < 6:
lines.append(test.strip())
test = f.readline()
if test == '':
break
sent_end = ('.', '!', '?', ':', ';')
if not test.isspace():
if not lines[-1].endswith(sent_end):
while not test.isspace() and\
not lines[-1].endswith(sent_end):
lines.append(test.strip())
test = f.readline()
if test == '':
break
f.close
# Insert a dot to the end
punctuation = [',', ':', ';']
last = lines[-1]
if last[-1].isalpha():
last = last + '.'
elif last[-1] in punctuation:
last = last[:len(last)-1] + '.'
lines[-1] = last
string = '\n'.join(lines)
return string
def get_verse_of_corpus(corpus, num_words, num_lines):
"""
Reads some text from a file and formats it.
The formatted text has the given number of lines and the given
number of words per line.
>>> verse = get_verse_of_corpus('../apparatus/poetry/english_poems.txt', 4, 4)
>>> len(verse.split()) == 16
True
:param corpus: text file
:param num_words: number of words per line
:param num_lines: number of lines
:return: random text with the given number of lines and given number
of words per line
"""
f = codecs.open(corpus, 'r', 'utf-8')
#Go to the randomly selected line
random_place = random.randint(1, helpers.file_length(corpus) - 10)
for i in range(random_place):
f.readline()
test = ''
lines = []
words = []
i = 0
test = f.readline()
words = test.split()
while len(lines) < num_lines:
line = []
while len(line) < num_words:
if i == len(words):
i = 0
test = f.readline()
if test == '':
break
words = test.split()
for j in range(i, len(words)):
if len(line) < num_words:
line.append(words[j])
i = i + 1
else:
break
lines.append(line)
f.close
string = ''
for line in lines:
string = string + ' '.join(line) + '\n'
return string
def get_verse_of_corpus(corpus, num_words, num_lines):
"""
Reads some text from a file and formats it.
The formatted text has the given number of lines and the given
number of words per line.
>>> verse = get_verse_of_corpus('../apparatus/poetry/english_poems.txt', 4, 4)
>>> len(verse.split()) == 16
True
:param corpus: text file
:param num_words: number of words per line
:param num_lines: number of lines
:return: random text with the given number of lines and given number
of words per line
"""
f = codecs.open(corpus, 'r', 'utf-8')
#Go to the randomly selected line
random_place = random.randint(1, helpers.file_length(corpus)-10)
for i in range(random_place):
f.readline()
test = ''
lines = []
words = []
i = 0
test = f.readline()
words = test.split()
while len(lines) < num_lines:
line = []
while len(line) < num_words:
if i == len(words):
i = 0
test = f.readline()
if test == '':
break
words = test.split()
for j in range(i, len(words)):
if len(line) < num_words:
line.append(words[j])
i = i+1
else:
break
lines.append(line)
f.close
string = ''
for line in lines:
string = string + ' '.join(line) + '\n'
return string
