def parse_corpus(proj_path = False,
corpuspath = False,
filelist = False,
corenlppath = False,
operations = False,
only_tokenise = False,
root = False,
stdout = False,
nltk_data_path = False,
memory_mb = 2000,
copula_head = True,
**kwargs):
"""
Create a CoreNLP-parsed and/or NLTK tokenised corpus
"""
import corpkit
import subprocess
from subprocess import PIPE, STDOUT, Popen
import os
import sys
import chardet
from time import localtime, strftime
import time
if not only_tokenise:
if not check_jdk():
print 'Need latest Java.'
return
curdir = os.getcwd()
if nltk_data_path:
if only_tokenise:
import nltk
if nltk_data_path not in nltk.data.path:
nltk.data.path.append(nltk_data_path)
from nltk import word_tokenize as tokenise
# add nltk to path
#td = {}
#from corpkit.other import add_nltk_data_to_nltk_path
#if 'note' in kwargs.keys():
# td['note'] = kwargs['note']
#add_nltk_data_to_nltk_path(**td)
if proj_path is False:
proj_path = os.path.dirname(os.path.abspath(corpuspath.rstrip('/')))
basecp = os.path.basename(corpuspath)
if only_tokenise:
new_corpus_path = os.path.join(proj_path, 'data', '%s-tokenised' % basecp)
else:
new_corpus_path = os.path.join(proj_path, 'data', '%s-parsed' % basecp)
if os.path.join('data', 'data') in new_corpus_path:
new_corpus_path = new_corpus_path.replace(os.path.join('data', 'data'), 'data')
if not os.path.isdir(new_corpus_path):
os.makedirs(new_corpus_path)
else:
fs = os.listdir(new_corpus_path)
if not only_tokenise:
if any([f.endswith('.xml') for f in fs]):
print 'Folder containing xml already exists: "%s-parsed"' % basecp
return False
else:
if any([f.endswith('.txt') for f in fs]):
print 'Folder containing tokens already exists: "%s-tokenised"' % basecp
return False
#javaloc = os.path.join(proj_path, 'corenlp', 'stanford-corenlp-3.6.0.jar:stanford-corenlp-3.6.0-models.jar:xom.jar:joda-time.jar:jollyday.jar:ejml-0.23.jar')
cwd = os.getcwd()
if corenlppath is False:
home = os.path.expanduser("~")
corenlppath = os.path.join(home, 'corenlp')
find_install = [d for d in os.listdir(corenlppath) \
if os.path.isdir(os.path.join(corenlppath, d)) \
and os.path.isfile(os.path.join(corenlppath, d, 'jollyday.jar'))]
if len(find_install) > 0:
corenlppath = os.path.join(corenlppath, find_install[0])
else:
print 'No parser found. Try using the keyword arg "corenlp = <path>".'
return
# if not gui, don't mess with stdout
if stdout is False:
stdout = sys.stdout
if not only_tokenise:
os.chdir(corenlppath)
if root:
root.update_idletasks()
reload(sys)
import os
import time
if memory_mb is False:
memory_mb = 2024
if operations is False:
operations = 'tokenize,ssplit,pos,lemma,ner,parse,dcoref'
if type(operations) == list:
operations = ','.join(operations)
num_files_to_parse = len([l for l in open(filelist, 'r').read().splitlines() if l])
# get corenlp version number
import re
reg = re.compile(r'stanford-corenlp-([0-9].[0-9].[0-9])-javadoc.jar')
fver = next(re.search(reg, s).group(1) for s in os.listdir('.') if re.search(reg, s))
arglist = ['java', '-cp',
'stanford-corenlp-%s.jar:stanford-corenlp-%s-models.jar:xom.jar:joda-time.jar:jollyday.jar:ejml-0.23.jar' % (fver, fver),
'-Xmx%sm' % str(memory_mb),
'edu.stanford.nlp.pipeline.StanfordCoreNLP',
'-annotators',
operations,
'-filelist', filelist,
'-noClobber',
'-outputDirectory', new_corpus_path]
if copula_head:
arglist.append('--parse.flags')
arglist.append(' -makeCopulaHead')
try:
proc = subprocess.Popen(arglist, stdout=sys.stdout)
# maybe a problem with stdout. sacrifice it if need be
except AttributeError:
proc = subprocess.Popen(arglist)
#p = TextProgressBar(num_files_to_parse)
while proc.poll() is None:
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
num_parsed = len([f for f in os.listdir(new_corpus_path) if f.endswith('.xml')])
if num_parsed == 0:
if root:
print '%s: Initialising parser ... ' % (thetime)
if num_parsed > 0 and (num_parsed + 1) <= num_files_to_parse:
if root:
print '%s: Parsing file %d/%d ... ' % (thetime, num_parsed + 1, num_files_to_parse)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set((num_parsed) * 100.0 / num_files_to_parse)
#p.animate(num_parsed - 1, str(num_parsed) + '/' + str(num_files_to_parse))
time.sleep(1)
if root:
root.update()
else:
from nltk import word_tokenize as tokenise
# tokenise each file
import pickle
fs = open(filelist).read().splitlines()
dirs = sorted(list(set([os.path.basename(os.path.dirname(f)) for f in fs])))
if len(dirs) == 0:
one_big_corpus = True
else:
one_big_corpus = False
if any(os.path.isdir(os.path.join(new_corpus_path, d)) for d in dirs):
thetime = strftime("%H:%M:%S", localtime())
print '%s: Directory already exists. Delete it if need be.' % thetime
return False
for d in dirs:
os.makedirs(os.path.join(new_corpus_path, d))
nfiles = len(fs)
thetime = strftime("%H:%M:%S", localtime())
print '%s: Tokenising ... ' % (thetime)
for index, f in enumerate(fs):
data = open(f).read()
enc = chardet.detect(data)
enc_text = unicode(data, enc['encoding'], errors = 'ignore')
tokens = tokenise(enc_text)
thedir = os.path.basename(os.path.dirname(f))
newname = os.path.basename(f).replace('.txt', '-tokenised.p')
if one_big_corpus:
pth = os.path.join(new_corpus_path, newname)
else:
pth = os.path.join(new_corpus_path, thedir, newname)
with open(pth, "wb") as fo:
pickle.dump(tokens, fo)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set((index + 1) * 100.0 / nfiles)
if root:
root.update()
#p.animate(num_files_to_parse)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set(100)
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
print '%s: Parsing finished. Moving parsed files into place ...' % thetime
os.chdir(curdir)
return new_corpus_path
def parse_corpus(proj_path=False,
corpuspath=False,
filelist=False,
corenlppath=False,
operations=False,
only_tokenise=False,
root=False,
stdout=False,
nltk_data_path=False,
memory_mb=2000,
copula_head=True,
multiprocessing=False,
**kwargs
):
"""
Create a CoreNLP-parsed and/or NLTK tokenised corpus
"""
import corpkit
import subprocess
from subprocess import PIPE, STDOUT, Popen
from corpkit.process import get_corenlp_path
import os
import sys
import re
import chardet
from time import localtime, strftime
import time
fileparse = kwargs.get('fileparse', False)
url = 'http://nlp.stanford.edu/software/stanford-corenlp-full-2015-12-09.zip'
if not only_tokenise:
if not check_jdk():
print('Need latest Java.')
return
curdir = os.getcwd()
note = kwargs.get('note', False)
if nltk_data_path:
if only_tokenise:
import nltk
if nltk_data_path not in nltk.data.path:
nltk.data.path.append(nltk_data_path)
from nltk import word_tokenize as tokenise
if proj_path is False:
proj_path = os.path.dirname(os.path.abspath(corpuspath.rstrip('/')))
basecp = os.path.basename(corpuspath)
if fileparse:
new_corpus_path = os.path.dirname(corpuspath)
else:
if only_tokenise:
new_corpus_path = os.path.join(proj_path, 'data', '%s-tokenised' % basecp)
else:
new_corpus_path = os.path.join(proj_path, 'data', '%s-parsed' % basecp)
# todo:
# this is not stable
if os.path.join('data', 'data') in new_corpus_path:
new_corpus_path = new_corpus_path.replace(os.path.join('data', 'data'), 'data')
# this caused errors when multiprocessing
# it used to be isdir, but supposedly there was a file there
# i don't see how it's possible ...
if not os.path.exists(new_corpus_path):
os.makedirs(new_corpus_path)
else:
if not os.path.isfile(new_corpus_path):
fs = os.listdir(new_corpus_path)
if not multiprocessing:
if not only_tokenise:
if any([f.endswith('.xml') for f in fs]):
print('Folder containing xml already exists: "%s-parsed"' % basecp)
return False
else:
if any([f.endswith('.p') for f in fs]):
print('Folder containing tokens already exists: "%s-tokenised"' % basecp)
return False
corenlppath = get_corenlp_path(corenlppath)
if not corenlppath:
cnlp_dir = os.path.join(os.path.expanduser("~"), 'corenlp')
corenlppath, fpath = download_large_file(cnlp_dir, url,
root=root,
note=note,
actually_download=True,
custom_corenlp_dir=corenlppath)
if corenlppath is None and fpath is None:
import shutil
shutil.rmtree(new_corpus_path)
shutil.rmtree(new_corpus_path.replace('-parsed', ''))
os.remove(new_corpus_path.replace('-parsed', '-filelist.txt'))
raise ValueError('CoreNLP needed to parse texts.')
extract_cnlp(fpath)
import glob
globpath = os.path.join(corenlppath, 'stanford-corenlp*')
corenlppath = [i for i in glob.glob(globpath) if os.path.isdir(i)]
if corenlppath:
corenlppath = corenlppath[-1]
else:
raise ValueError('CoreNLP installation failed for some reason. Try manual download.')
# if not gui, don't mess with stdout
if stdout is False:
stdout = sys.stdout
if not only_tokenise:
os.chdir(corenlppath)
if root:
root.update_idletasks()
reload(sys)
if memory_mb is False:
memory_mb = 2024
if operations is False:
operations = 'tokenize,ssplit,pos,lemma,parse,ner,dcoref'
if isinstance(operations, list):
operations = ','.join([i.lower() for i in operations])
with open(filelist, 'r') as fo:
dat = fo.read()
num_files_to_parse = len([l for l in dat.splitlines() if l])
# get corenlp version number
reg = re.compile(r'stanford-corenlp-([0-9].[0-9].[0-9])-javadoc.jar')
fver = next(re.search(reg, s).group(1) for s in os.listdir('.') if re.search(reg, s))
if fver == '3.6.0':
extra_jar = 'slf4j-api.jar:slf4j-simple.jar:'
else:
extra_jar = ''
arglist = ['java', '-cp',
'stanford-corenlp-%s.jar:stanford-corenlp-%s-models.jar:xom.jar:joda-time.jar:%sjollyday.jar:ejml-0.23.jar' % (fver, fver, extra_jar),
'-Xmx%sm' % str(memory_mb),
'edu.stanford.nlp.pipeline.StanfordCoreNLP',
'-annotators',
operations,
'-filelist', filelist,
'-noClobber',
'-outputExtension', '.xml',
'-outputDirectory', new_corpus_path]
if copula_head:
arglist.append('--parse.flags')
arglist.append(' -makeCopulaHead')
try:
proc = subprocess.Popen(arglist, stdout=sys.stdout)
# maybe a problem with stdout. sacrifice it if need be
except:
proc = subprocess.Popen(arglist)
#p = TextProgressBar(num_files_to_parse)
while proc.poll() is None:
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
if not fileparse:
num_parsed = len([f for f in os.listdir(new_corpus_path) if f.endswith('.xml')])
if num_parsed == 0:
if root:
print('%s: Initialising parser ... ' % (thetime))
if num_parsed > 0 and (num_parsed + 1) <= num_files_to_parse:
if root:
print('%s: Parsing file %d/%d ... ' % \
(thetime, num_parsed + 1, num_files_to_parse))
if kwargs.get('note'):
kwargs['note'].progvar.set((num_parsed) * 100.0 / num_files_to_parse)
#p.animate(num_parsed - 1, str(num_parsed) + '/' + str(num_files_to_parse))
time.sleep(1)
if root:
root.update()
else:
from nltk import word_tokenize as tokenise
# tokenise each file
import cPickle as pickle
fs = open(filelist).read().splitlines()
dirs = sorted(list(set([os.path.basename(os.path.dirname(f)) for f in fs])))
one_big_corpus = len(dirs) == 0
if any(os.path.isdir(os.path.join(new_corpus_path, d)) for d in dirs):
thetime = strftime("%H:%M:%S", localtime())
print('%s: Directory already exists. Delete it if need be.' % thetime)
return False
for d in dirs:
os.makedirs(os.path.join(new_corpus_path, d))
nfiles = len(fs)
thetime = strftime("%H:%M:%S", localtime())
print('%s: Tokenising ... ' % (thetime))
for index, f in enumerate(fs):
with open(f, 'r') as fo:
data = fo.read()
enc = chardet.detect(data)
enc_text = data.decode(enc['encoding'], errors='ignore')
tokens = tokenise(enc_text)
thedir = os.path.basename(os.path.dirname(f))
newname = os.path.basename(f).replace('.txt', '-tokenised.p')
if one_big_corpus:
pth = os.path.join(new_corpus_path, newname)
else:
pth = os.path.join(new_corpus_path, thedir, newname)
with open(pth, "wb") as fo:
pickle.dump(tokens, fo)
if kwargs.get('note'):
kwargs['note'].progvar.set((index + 1) * 100.0 / nfiles)
if root:
root.update()
#p.animate(num_files_to_parse)
if kwargs.get('note'):
kwargs['note'].progvar.set(100)
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
print('%s: Parsing finished. Moving parsed files into place ...' % thetime)
os.chdir(curdir)
return new_corpus_path
import os, json
from nltk import word_tokenize as tokenise
## os.getcwd() returns the current working directory
## os.path.dirname() takes a path and returns the directory above it
corpus_filename = os.path.dirname(os.getcwd()) + '/corpus.json'
## load the corpus
corpus = json.load(open(corpus_filename))
index = {}
## number of training examples in corpus
m = len(corpus)
for i in range(m):
tokens = tokenise(corpus[i]['title']) + tokenise(corpus[i]['text'])
for token in tokens:
token_lower = token.lower()
if token_lower not in index:
index[token_lower] = [i]
else:
index[token_lower].append(i)
with open('index.json', 'w') as op:
op.write(json.dumps(index))
def parse_corpus(proj_path, corpuspath, filelist, corenlppath = False, operations = False,
only_tokenise = False, root = False, stdout = False, **kwargs):
import corpkit
import subprocess
from subprocess import PIPE, STDOUT, Popen
import os
import sys
import chardet
from time import localtime, strftime
import time
if not only_tokenise:
if not check_jdk():
print 'Need latest Java.'
return
# add nltk to path
td = {}
from corpkit.other import add_nltk_data_to_nltk_path
if 'note' in kwargs.keys():
td['note'] = kwargs['note']
add_nltk_data_to_nltk_path(**td)
basecp = os.path.basename(corpuspath)
if only_tokenise:
new_corpus_path = os.path.join(proj_path, 'data', '%s-tokenised' % basecp)
else:
new_corpus_path = os.path.join(proj_path, 'data', '%s-parsed' % basecp)
if not os.path.isdir(new_corpus_path):
os.makedirs(new_corpus_path)
else:
fs = os.listdir(new_corpus_path)
if not only_tokenise:
if any([f.endswith('.xml') for f in fs]):
print 'Folder containing xml already exists: "%s-parsed"' % basecp
return False
else:
if any([f.endswith('.txt') for f in fs]):
print 'Folder containing tokens already exists: "%s-tokenised"' % basecp
return False
#javaloc = os.path.join(proj_path, 'corenlp', 'stanford-corenlp-3.5.2.jar:stanford-corenlp-3.5.2-models.jar:xom.jar:joda-time.jar:jollyday.jar:ejml-0.23.jar')
cwd = os.getcwd()
if corenlppath is False:
home = os.path.expanduser("~")
corenlppath = os.path.join(home, 'corenlp')
find_install = [d for d in os.listdir(corenlppath) \
if os.path.isdir(os.path.join(corenlppath, d)) \
and os.path.isfile(os.path.join(corenlppath, d, 'jollyday.jar'))]
if len(find_install) > 0:
corenlppath = os.path.join(corenlppath, find_install[0])
else:
print 'No parser found.'
return
if not only_tokenise:
os.chdir(corenlppath)
root.update_idletasks()
reload(sys)
import os
import time
if operations is False:
operations = 'tokenize,ssplit,pos,lemma,ner,parse,dcoref'
num_files_to_parse = len([l for l in open(filelist, 'r').read().splitlines() if l])
proc = subprocess.Popen(['java', '-cp',
'stanford-corenlp-3.5.2.jar:stanford-corenlp-3.5.2-models.jar:xom.jar:joda-time.jar:jollyday.jar:ejml-0.23.jar',
'-Xmx2g',
'edu.stanford.nlp.pipeline.StanfordCoreNLP',
'-annotators',
operations,
'-filelist', filelist,
'-noClobber',
'-outputDirectory', new_corpus_path,
'--parse.flags', ' -makeCopulaHead'], stdout=sys.stdout)
#p = TextProgressBar(num_files_to_parse)
while proc.poll() is None:
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
num_parsed = len([f for f in os.listdir(new_corpus_path) if f.endswith('.xml')])
if num_parsed == 0:
print '%s: Initialising parser ... ' % (thetime)
if num_parsed > 0 and num_parsed <= num_files_to_parse:
print '%s: Parsing file %d/%d ... ' % (thetime, num_parsed + 1, num_files_to_parse)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set((num_parsed) * 100.0 / num_files_to_parse)
#p.animate(num_parsed - 1, str(num_parsed) + '/' + str(num_files_to_parse))
if root:
root.update()
time.sleep(1)
else:
# tokenise each file
from nltk import word_tokenize as tokenise
import pickle
fs = open(filelist).read().splitlines()
dirs = sorted(list(set([os.path.basename(os.path.dirname(f)) for f in fs])))
if len(dirs) == 0:
one_big_corpus = True
else:
one_big_corpus = False
if any(os.path.isdir(os.path.join(new_corpus_path, d)) for d in dirs):
thetime = strftime("%H:%M:%S", localtime())
print '%s: Directory already exists. Delete it if need be.' % thetime
return
for d in dirs:
os.makedirs(os.path.join(new_corpus_path, d))
nfiles = len(fs)
thetime = strftime("%H:%M:%S", localtime())
print '%s: Tokenising ... ' % (thetime)
for index, f in enumerate(fs):
data = open(f).read()
enc = chardet.detect(data)
enc_text = unicode(data, enc['encoding'], errors = 'ignore')
tokens = tokenise(enc_text)
thedir = os.path.basename(os.path.dirname(f))
newname = os.path.basename(f).replace('.txt', '-tokenised.p')
if one_big_corpus:
pth = os.path.join(new_corpus_path, newname)
else:
pth = os.path.join(new_corpus_path, thedir, newname)
with open(pth, "wb") as fo:
pickle.dump(tokens, fo)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set((index + 1) * 100.0 / nfiles)
if root:
root.update()
#p.animate(num_files_to_parse)
if 'note' in kwargs.keys():
kwargs['note'].progvar.set(100)
sys.stdout = stdout
print 'Parsing finished. Moving parsed files into place ...'
os.chdir(proj_path)
return new_corpus_path
def parse_corpus(proj_path=False,
corpuspath=False,
filelist=False,
corenlppath=False,
operations=False,
only_tokenise=False,
root=False,
stdout=False,
nltk_data_path=False,
memory_mb=2000,
copula_head=True,
**kwargs):
"""
Create a CoreNLP-parsed and/or NLTK tokenised corpus
"""
import corpkit
import subprocess
from subprocess import PIPE, STDOUT, Popen
import os
import sys
import chardet
from time import localtime, strftime
import time
if not only_tokenise:
if not check_jdk():
print('Need latest Java.')
return
curdir = os.getcwd()
if nltk_data_path:
if only_tokenise:
import nltk
if nltk_data_path not in nltk.data.path:
nltk.data.path.append(nltk_data_path)
from nltk import word_tokenize as tokenise
# add nltk to path
#td = {}
#from other import add_nltk_data_to_nltk_path
#if 'note' in kwargs.keys():
# td['note'] = kwargs['note']
#add_nltk_data_to_nltk_path(**td)
if proj_path is False:
proj_path = os.path.dirname(os.path.abspath(corpuspath.rstrip('/')))
basecp = os.path.basename(corpuspath)
if only_tokenise:
new_corpus_path = os.path.join(proj_path, 'data',
'%s-tokenised' % basecp)
else:
new_corpus_path = os.path.join(proj_path, 'data', '%s-parsed' % basecp)
if os.path.join('data', 'data') in new_corpus_path:
new_corpus_path = new_corpus_path.replace(os.path.join('data', 'data'),
'data')
if not os.path.isdir(new_corpus_path):
os.makedirs(new_corpus_path)
else:
fs = os.listdir(new_corpus_path)
if not only_tokenise:
if any([f.endswith('.xml') for f in fs]):
print('Folder containing xml already exists: "%s-parsed"' %
basecp)
return False
else:
if any([f.endswith('.txt') for f in fs]):
print(
'Folder containing tokens already exists: "%s-tokenised"' %
basecp)
return False
#javaloc = os.path.join(proj_path, 'corenlp', 'stanford-corenlp-3.6.0.jar:stanford-corenlp-3.6.0-models.jar:xom.jar:joda-time.jar:jollyday.jar:ejml-0.23.jar')
cwd = os.getcwd()
if corenlppath is False:
home = os.path.expanduser("~")
corenlppath = os.path.join(home, 'corenlp')
find_install = [d for d in os.listdir(corenlppath) \
if os.path.isdir(os.path.join(corenlppath, d)) \
and os.path.isfile(os.path.join(corenlppath, d, 'jollyday.jar'))]
if len(find_install) > 0:
corenlppath = os.path.join(corenlppath, find_install[0])
else:
print(
'No parser found. Try using the keyword arg "corenlp = <path>", or moving your corenlp folder to ~/corenlp/stanford-corenlp-full ...'
)
return
# if not gui, don't mess with stdout
if stdout is False:
stdout = sys.stdout
if not only_tokenise:
os.chdir(corenlppath)
if root:
root.update_idletasks()
reload(sys)
import os
import time
if memory_mb is False:
memory_mb = 2024
if operations is False:
operations = 'tokenize,ssplit,pos,lemma,ner,parse,dcoref'
if type(operations) == list:
operations = ','.join(operations)
num_files_to_parse = len(
[l for l in open(filelist, 'r').read().splitlines() if l])
# get corenlp version number
import re
reg = re.compile(r'stanford-corenlp-([0-9].[0-9].[0-9])-javadoc.jar')
fver = next(
re.search(reg, s).group(1) for s in os.listdir('.')
if re.search(reg, s))
if fver == '3.6.0':
extra_jar = 'slf4j-api.jar:slf4j-simple.jar:'
else:
extra_jar = ''
arglist = [
'java', '-cp',
'stanford-corenlp-%s.jar:stanford-corenlp-%s-models.jar:xom.jar:joda-time.jar:%sjollyday.jar:ejml-0.23.jar'
% (fver, fver, extra_jar),
'-Xmx%sm' % str(memory_mb),
'edu.stanford.nlp.pipeline.StanfordCoreNLP', '-annotators',
operations, '-filelist', filelist, '-noClobber',
'-outputExtension', '.xml', '-outputDirectory', new_corpus_path
]
if copula_head:
arglist.append('--parse.flags')
arglist.append(' -makeCopulaHead')
try:
proc = subprocess.Popen(arglist, stdout=sys.stdout)
# maybe a problem with stdout. sacrifice it if need be
except:
proc = subprocess.Popen(arglist)
#p = TextProgressBar(num_files_to_parse)
while proc.poll() is None:
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
num_parsed = len(
[f for f in os.listdir(new_corpus_path) if f.endswith('.xml')])
if num_parsed == 0:
if root:
print('%s: Initialising parser ... ' % (thetime))
if num_parsed > 0 and (num_parsed + 1) <= num_files_to_parse:
if root:
print('%s: Parsing file %d/%d ... ' %
(thetime, num_parsed + 1, num_files_to_parse))
if 'note' in list(kwargs.keys()):
kwargs['note'].progvar.set(
(num_parsed) * 100.0 / num_files_to_parse)
#p.animate(num_parsed - 1, str(num_parsed) + '/' + str(num_files_to_parse))
time.sleep(1)
if root:
root.update()
else:
from nltk import word_tokenize as tokenise
# tokenise each file
import pickle
fs = open(filelist).read().splitlines()
dirs = sorted(
list(set([os.path.basename(os.path.dirname(f)) for f in fs])))
if len(dirs) == 0:
one_big_corpus = True
else:
one_big_corpus = False
if any(os.path.isdir(os.path.join(new_corpus_path, d)) for d in dirs):
thetime = strftime("%H:%M:%S", localtime())
print('%s: Directory already exists. Delete it if need be.' %
thetime)
return False
for d in dirs:
os.makedirs(os.path.join(new_corpus_path, d))
nfiles = len(fs)
thetime = strftime("%H:%M:%S", localtime())
print('%s: Tokenising ... ' % (thetime))
for index, f in enumerate(fs):
data = open(f).read()
enc = chardet.detect(data)
enc_text = str(data, enc['encoding'], errors='ignore')
tokens = tokenise(enc_text)
thedir = os.path.basename(os.path.dirname(f))
newname = os.path.basename(f).replace('.txt', '-tokenised.p')
if one_big_corpus:
pth = os.path.join(new_corpus_path, newname)
else:
pth = os.path.join(new_corpus_path, thedir, newname)
with open(pth, "wb") as fo:
pickle.dump(tokens, fo)
if 'note' in list(kwargs.keys()):
kwargs['note'].progvar.set((index + 1) * 100.0 / nfiles)
if root:
root.update()
#p.animate(num_files_to_parse)
if 'note' in list(kwargs.keys()):
kwargs['note'].progvar.set(100)
sys.stdout = stdout
thetime = strftime("%H:%M:%S", localtime())
print('%s: Parsing finished. Moving parsed files into place ...' % thetime)
os.chdir(curdir)
return new_corpus_path
