def __init__(self, seed=42):
super(Solver, self).__init__()
self.is_train_task = False
self.morph = pymorphy2.MorphAnalyzer()
self.toktok = ToktokTokenizer()
self.seed = seed
self.init_seed()
def __init__(self, seed=42):
self.seed = seed
self.init_seed()
self.morph = morph
self.toktok = ToktokTokenizer()
self.bert = BertEmbedder()
self.is_loaded = True
def __init__(self):
self.wl = set()
for each in brown.words():
each = each.lower()
if each.isalpha() and (each not in self.wl):
self.wl.add(each)
self.toktok = ToktokTokenizer()
def tokenize(i_file, o_file):
toktok = ToktokTokenizer()
with open(i_file, 'r') as i_f, open(o_file, 'w') as o_f:
for line in tqdm(i_f):
line = line.rstrip('\n')
tokens = toktok.tokenize(line)
print(' '.join(tokens), file=o_f)
def __init__(self, seed=42, data_path = 'data/'):
self.is_train_task = False
self.morph = pymorphy2.MorphAnalyzer()
self.toktok = ToktokTokenizer()
self.seed = seed
self.init_seed()
#self.synonyms = open(os.path.join(data_path, r'synonyms.txt'), 'r', encoding='utf8').readlines()
#self.synonyms = [re.sub('\.','', t.lower().strip('\n')).split(' ') for t in self.synonyms]
#self.synonyms = [[t for t in l if t] for l in self.synonyms]
self.synonyms = open('./data/synmaster.txt', 'r', encoding='utf8').readlines()
self.synonyms = [re.sub('\.','', t.lower().strip('\n')).split('|') for t in self.synonyms if len(t)>5]
self.antonyms = open('./data/antonyms.txt', 'r', encoding='utf8').readlines()
self.antonyms = [re.sub('\.','', t.lower().strip('\n')).split('|') for t in self.antonyms if len(t)>5]
#self.antonyms = open(os.path.join(data_path, r'antonyms.txt'), 'r', encoding='utf8').readlines()
#self.antonyms = [t.strip(' \n').split(' - ') for t in self.antonyms]
self.phraseology = []
self.razgov = []
self.musor = []
with open('./data/word.csv', encoding='utf-8') as f:
for line in f:
line = line.replace("\n","").split('|')
if line[1] == '23':
self.phraseology.append(line[0])
#self.musor.append(line[0])
if line[1] == '25':
self.razgov.append(line[0])
#self.musor.append(line[0])
if line[1] == '99':
self.musor.append(line[0])
def file_to_features(path, word_vocab, window, min_count, total_w):
examples = []
toktok = ToktokTokenizer()
punckt = set(string.punctuation)
try:
with open(path, 'r', encoding='utf8') as f:
for line in f:
for sentence in sent_tokenize(line):
words_1 = toktok.tokenize(sentence)
words_2 = []
for i, word in enumerate(words_1):
word_l = word.lower()
if word_l not in word_vocab:
continue
if word_vocab[word_l] < min_count:
continue
if word in punckt:
continue
frequency = word_vocab[word_l] / total_w
number = 1 - math.sqrt(10e-5 / frequency)
if random.uniform(0, 1) <= number:
continue
words_2.append(word)
max_j = len(words_2)
for i, word in enumerate(words_2):
start = i - window if (i - window) > 0 else 0
to = i + window if (i + window) < max_j else max_j
for j in range(start, to):
if i == j:
continue
target = words_2[j]
examples.append((word, target))
except Exception as error:
print(error)
return examples
def __init__(self, log_dir, from_log=False):
self.log_dir = log_dir
if checkExistenceFile(os.path.join(log_dir,
"PreprocessorConfig.json")):
self.read_config()
self.tok = ToktokTokenizer()
self.parsing_char_ = sha1(b"sally14").hexdigest()
self.fitted = False
if from_log:
self.fitted = True
with open(
os.path.join(self.log_dir, "vocabulary.json"),
"r",
encoding="utf-8",
) as f:
self.vocabulary_ = json.load(f)
with open(
os.path.join(self.log_dir, "WordPhrases.json"),
"r",
encoding="utf-8",
) as f:
p = json.load(f)
self.phrasewords_ = {
i.replace("_", self.parsing_char_): p[i]
for i in p.keys()
}
def __init__(self, seed=42, data_path='data/'):
self.is_train_task = False
self.morph = pymorphy2.MorphAnalyzer()
self.toktok = ToktokTokenizer()
self.seed = seed
self.init_seed()
self.synonyms = open(os.path.join(data_path, r'synonyms.txt'),
'r',
encoding='utf8').readlines()
self.synonyms = [
re.sub('\.', '',
t.lower().strip('\n')).split(' ') for t in self.synonyms
]
self.synonyms = [[t for t in l if t] for l in self.synonyms]
self.antonyms = open(os.path.join(data_path, r'antonyms.txt'),
'r',
encoding='utf8').readlines()
self.antonyms = [t.strip(' \n').split(' - ') for t in self.antonyms]
self.phraseology = open(os.path.join(data_path, r'phraseologs.txt'),
'r',
encoding='utf8').readlines()
self.phraseology = [[
l for l in self.lemmatize(l) if l not in
['\n', ' ', '...', '', ',', '-', '.', '?', r' (', r'/']
] for l in self.phraseology]
def build_vocabs(directory_path, min_count):
"""Build the word and char counter vocabularies"""
toktok = ToktokTokenizer()
word_vocab = Counter()
char_vocab = Counter()
char_vocab.update(['{', '}'])
filenames = os.listdir(directory_path)
filepaths = [os.path.join(directory_path, e) for e in filenames]
for i, filepath in enumerate(filepaths):
if i % 100 == 0:
print('Reading file number {}'.format(i), end="\r")
with open(filepath, 'r', encoding='utf8') as f:
try:
line = f.read()
if 'numbers_' in filepath:
tmp = toktok.tokenize(line.lower())
for i in range(min_count):
word_vocab.update(tmp)
else:
word_vocab.update(word_tokenize(line.lower()))
char_vocab.update(line)
except Exception as error:
print('Error with file: {}'.format(filepath))
print(error)
return word_vocab, char_vocab
def extract_wiki_fdict():
f_count = 0
# for each wiki table, get header name, and corresponding content
f = open(wiki_path, 'r')
f_dest = open(wiki_fdict_path, 'w')
toktok = ToktokTokenizer()
tid = 0
pool = mp.Pool()
for line in f:
tid += 1
t = json.loads(line)
if not check_format(t):
continue
try:
# header process
header_iter = iter(t['tableHeaders'][-1])
header_span = []
header_content = dict()
header_bows = dict()
header_idx = 0
for each_header in header_iter:
html_desc = each_header['tdHtmlString']
span = int(html_desc.split('colspan="')[1].split('"')[0])
header_span.append((each_header['text'], span))
header_content[header_idx] = []
header_bows[header_idx] = []
header_idx += 1
if span != 1:
for skip_num in range(span - 1):
next(header_iter)
# content process
for row in t['tableData']:
global_col_index = 0
header_idx = 0
for header, span in header_span:
for idx in range(span):
if row[global_col_index]['text'] != '':
header_content[header_idx].append(
row[global_col_index]['text'])
header_bows[header_idx].extend(
toktok.tokenize(row[global_col_index]['text']))
global_col_index += 1
header_idx += 1
except:
continue
#combine header and features
cols_features = pool.map(gov_data.get_col_features,
list(header_content.values()))
all_col_features = list(
zip([each[0] for each in header_span], cols_features))
for i in range(len(all_col_features)):
if all_col_features[i][1]:
all_col_features[i][1]['content'] = header_bows[i]
all_col_features = list(filter(lambda x: x[1], all_col_features))
f_dest.write(json.dumps({tid: all_col_features}, cls=MyEncoder) + '\n')
print("finishing {0}".format(f_count))
f_count += 1
def loss_char(sentence, position):
toktok = ToktokTokenizer()
if sentence[position] in " ,./;'[]\<>?:{}!@#$% ^&*()":
return sentence
if sentence[position] == " ":
return sentence
if sentence[position] in toktok.tokenize(sentence):
return sentence
return sentence[:position] + sentence[position + 1:]
def RemoveWords_by_tag(text):
remove_tag_list = ['JJ', 'JJR', 'JJS', 'RBR', 'RBS']
token = ToktokTokenizer()
words = token.tokenize(text)
words_tagged = nltk.pos_tag(words)
filtered = untag([
w for w in words_tagged if not w[1] in remove_tag_list
]) # Filtre les mots qui n'appartiennt pas à la catégorie à supprimer
return ' '.join(map(str, filtered))
def extract_wiki_features(wiki_feature_path, wiki_bow_path):
f_count = 0
# for each wiki table, get header name, and corresponding content
f = open(wiki_path, 'r')
f_dest = open(wiki_feature_path, 'w')
f_bow = open(wiki_bow_path, 'w')
toktok = ToktokTokenizer()
for line in f:
t = json.loads(line)
if not check_format(t):
continue
try:
# header process
header_iter = iter(t['tableHeaders'][-1])
header_span = []
header_content = dict()
header_bows = dict()
header_idx = 0
for each_header in header_iter:
html_desc = each_header['tdHtmlString']
span = int(html_desc.split('colspan="')[1].split('"')[0])
header_span.append((each_header['text'], span))
header_content[header_idx] = []
header_bows[header_idx] = []
header_idx += 1
if span != 1:
for skip_num in range(span - 1):
next(header_iter)
# content process
for row in t['tableData']:
global_col_index = 0
header_idx = 0
for header, span in header_span:
for idx in range(span):
if row[global_col_index]['text'] != '':
header_content[header_idx].append(
row[global_col_index]['text'])
header_bows[header_idx].extend(
toktok.tokenize(row[global_col_index]['text']))
global_col_index += 1
header_idx += 1
except:
continue
#combine header and features
for col, f_dict, bows in zip([each[0] for each in header_span],
map(get_col_features,
header_content.values()),
header_bows.values()):
if f_dict:
f_dict['_id'] = t['_id']
f_dest.write(json.dumps({col: f_dict}) + '\n')
f_bow.write(json.dumps({col: bows}) + '\n')
print("finishing {0}".format(f_count))
f_count += 1
def extract_gov_fdict(all_resources,
fdict_path=gov_data_fdict_path,
tid_type='cat_id',
restrict_resource=False):
#extracting features:
#table_id;label,curated_features,content;label,curated_features...
f = open(fdict_path, 'w')
#all_resources = gov_data.read_resources()
all_resources = gov_data.wrong_csv(all_resources)
all_resources = list(filter(lambda x: x.status, all_resources))
if restrict_resource:
all_resources = gov_data.select_resources(all_resources,
fsize=50,
rs_ct=len(all_resources))
pool = mp.Pool()
total = len(all_resources)
count = 0
toktok = ToktokTokenizer()
for resource in all_resources:
print("processing {0}-th resource".format(count))
for each_data in resource.data_files:
try:
if tid_type == 'cat_id':
tid = resource.rs_id + ':' + each_data.df_id
elif tid_type == 'path':
tid = resource.path + '/' + each_data.df_id
d_path = each_data.path + '/data.csv'
df = pd.read_csv(d_path,
delimiter=',',
quotechar='"',
dtype=str,
na_filter=True)
cols = df.columns
contents = [
df[each_col].dropna().tolist() for each_col in cols
]
print("extract content finished")
cols_features = pool.map(gov_data.get_col_features, contents)
all_col_features = list(zip(cols, cols_features))
for i in range(len(all_col_features)):
if all_col_features[i][1]:
all_col_features[i][1]['content'] = toktok.tokenize(
' '.join(contents[i]))
all_col_features = list(
filter(lambda x: x[1], all_col_features))
f.write(
json.dumps({tid: all_col_features}, cls=MyEncoder) + '\n')
except Exception as e:
print(e)
count += 1
print("finish {0} out of {1}".format(count, total))
f.close()
return all_resources
def index(request):
global invertedIndex
global jsonData
output_links = []
searchTermsReq = request.GET.get('term', '')
print(searchTermsReq)
tokenizer = ToktokTokenizer()
searchTerms = tokenizer.tokenize(searchTermsReq)
print(searchTerms)
response = {}
output_data = defaultdict(int)
output_links = []
for token in searchTerms:
token = token.lower()
if invertedIndex[token]['idf'] > 0.25 and len(token) > 1:
print('Looking through high for: ' + token)
for docFilePath in invertedIndex[token]['high']:
tfidf = invertedIndex[token]['high'][docFilePath]
output_data[docFilePath] += tfidf
if (len(output_data) < 10):
for token in searchTerms:
token = token.lower()
if invertedIndex[token]['idf'] > 0.25 and len(token) > 1:
print('Looking through low for: ' + token)
for docFilePath in invertedIndex[token]['low']:
tfidf = invertedIndex[token]['low'][docFilePath]
output_data[docFilePath] += tfidf
output_data = sorted(output_data.items(), key=itemgetter(1), reverse=True)
for docFilePath, tfidf in output_data[:10]:
output_links.append((jsonData[docFilePath], tfidf))
output_links.sort(key=itemgetter(1), reverse=True)
response['term'] = searchTermsReq
response['results'] = output_links
response['totalURLs'] = len(output_data)
response['uniqueTokens'] = len(invertedIndex)
response['totalDocuments'] = len(jsonData)
return JsonResponse(response)
def __init__(self):
self.morph = morph
self.mystem = Mystem()
self.tokenizer = ToktokTokenizer()
self.w2v = Word2vecProcessor()
self.synonyms = None
self.antonyms = None
self.phraseology = None
self.phraseologisms = None
self.prep_synon = None
self.set_f = None
self.verbs_dict = None
self.chasti_rechi = None
self.set_f_2 = None
def write_in_file():
"""
"""
# Get the files
list_of_files = os.listdir(TRAINING_FOLDER_PATH)
number_of_file = len(list_of_files)
# Initialise the lemmatizer
lemmatizer = nltk.WordNetLemmatizer()
# Initialise the tokenizer
tokenizer = ToktokTokenizer()
tokenizer.AMPERCENT = re.compile('& '), '& '
tokenizer.TOKTOK_REGEXES = [
(regex, sub) if sub != '& ' else (re.compile('& '), '& ')
for (regex, sub) in ToktokTokenizer.TOKTOK_REGEXES
]
toktok = tokenizer.tokenize
total_text = pd.Series([])
# loop in the files
for i in range(0, 11):
file_name = list_of_files[i]
print(i)
# open the files
with open(os.path.join(TRAINING_FOLDER_PATH, file_name),
'r',
encoding="utf8") as text:
string_text = text.read()
splitted = toktok(string_text)
# Lemmatize
lemmatized = [lemmatizer.lemmatize(t) for t in splitted]
tokens = pd.Series(lemmatized)
# Take off random punctuation
# All the numbers under the same name
tokens.loc[tokens.apply(lambda x: x.isnumeric())] = "NUMBER"
total_text = total_text.append(tokens)
# Write in a file
txtfilename = "training_text_file/" + str(i + 1) + "yo.txt"
with io.open(txtfilename, "w", encoding="utf-8") as f:
for item in total_text:
f.write("%s " % item)
def error_generator(utterance):
toktok = ToktokTokenizer()
length = len(utterance)
nb = nb_of_errors_in_utterance(length) + 1
utterance = utterance + " "
for i in range(nb):
length = len(utterance) - 1
position = np.random.choice(range(length), p=(length) * [1 / (length)])
l = len(toktok.tokenize(utterance))
utterance_old = utterance
nb = np.random.randint(1, 5)
utterance = functions[nb](utterance, position)
return utterance
def clean_archive_data(folder):
toktok = ToktokTokenizer()
if not os.path.exists(f"{folder}-cleaned"):
os.makedirs(f"{folder}-cleaned")
for count, file in enumerate(os.listdir(f"{folder}")):
if count % 1000 == 0:
print(count)
file_data = open(f"{folder}/{file}", "r").read()
try:
text_newspaper = toktok.tokenize(fulltext(file_data))
text_newspaper_cleaned = clean(" ".join(text_newspaper))
with open(f"{folder}-cleaned/{file}", "w") as output:
output.write(text_newspaper_cleaned)
except: # pylint: disable=W0702
print(f"error with {file}", file=sys.stderr)
def preprocess(data):
X, Y = [], []
toktok = ToktokTokenizer()
for index, review in data.iterrows():
if (index+1) % 100000 == 0:
print(index+1)
# words = nltk.word_tokenize(review['text'])
tokens = toktok.tokenize(review['text'].lower())
X.append(tokens)
# X.append(nltk.word_tokenize(review['text']))
Y.append(int(review['stars'] - 1))
# if len(Y) == 10000:
# break
df_new = pd.DataFrame({'text': X, 'stars': Y})
return df_new
def load_tokenizer(lang):
if lang == "ko":
from konlpy.tag import Mecab
tokenizer = Mecab()
elif lang == "ja":
import Mykytea
opt = "-model jp-0.4.7-1.mod"
tokenizer = Mykytea.Mykytea(opt)
elif lang == "zh_cn":
import Mykytea
opt = "-model ctb-0.4.0-1.mod"
tokenizer = Mykytea.Mykytea(opt)
elif lang == "zh_tw":
import jieba
tokenizer = jieba
elif lang == "vi":
from pyvi import ViTokenizer
tokenizer = ViTokenizer
elif lang == "th":
from pythainlp.tokenize import word_tokenize
tokenizer = word_tokenize
elif lang == "ar":
import pyarabic.araby as araby
tokenizer = araby
# elif lang=="en":
# from nltk import word_tokenize
# tokenizer = word_tokenize
else:
from nltk.tokenize import ToktokTokenizer
tokenizer = ToktokTokenizer()
return tokenizer
def makeModel():
#sentences = webtext.raw()+brown.raw()+reuters.raw()
sentences = webtext.raw() + reuters.raw()
# Tokenize the sentences
try: # Use the default NLTK tokenizer.
from nltk import word_tokenize, sent_tokenize
# Testing whether it works.
# Sometimes it doesn't work on some machines because of setup issues.
word_tokenize(
sent_tokenize("This is a foobar sentence. Yes it is.")[0])
except: # Use a naive sentence tokenizer and toktok.
import re
from nltk.tokenize import ToktokTokenizer
# See https://stackoverflow.com/a/25736515/610569
sent_tokenize = lambda x: re.split(r'(?<=[^A-Z].[.?]) +(?=[A-Z])', x)
# Use the toktok tokenizer that requires no dependencies.
toktok = ToktokTokenizer()
word_tokenize = word_tokenize = toktok.tokenize
tokenized_text = [
list(map(str.lower, word_tokenize(sent)))
for sent in sent_tokenize(sentences)
]
# Make it ready for making 3 grams
n = 5
train_data, padded_sents = padded_everygram_pipeline(n, tokenized_text)
model = MLE(n) # Lets train a 3-grams model, previously we set n=3
model.fit(train_data, padded_sents)
#print(model.vocab)
return model
def tokenize_nltk(self, text):
toktok = ToktokTokenizer()
# tokens =[toktok.tokenize(sent) for sent in sent_tokenize(text)]
tokens = nltk.word_tokenize(text)
stems = [stemmer.stem(t) for t in tokens]
# print("Number of tokens:", len(tokens))
return stems
def prepare_data(args):
UNK_TOKEN = "<unk>"
PAD_TOKEN = "<pad>"
SOS_TOKEN = "<s>"
EOS_TOKEN = "</s>"
if args.tokenize:
toktok = ToktokTokenizer()
SRC = data.Field(unk_token=UNK_TOKEN, pad_token=PAD_TOKEN, init_token=None, eos_token=EOS_TOKEN,
lower=args.lower, tokenize=toktok.tokenize)
TGT = data.Field(unk_token=UNK_TOKEN, pad_token=PAD_TOKEN, init_token=SOS_TOKEN, eos_token=EOS_TOKEN,
lower=args.lower, tokenize=toktok.tokenize)
else:
SRC = data.Field(unk_token=UNK_TOKEN, pad_token=PAD_TOKEN, init_token=None, eos_token=EOS_TOKEN, lower=args.lower)
TGT = data.Field(unk_token=UNK_TOKEN, pad_token=PAD_TOKEN, init_token=SOS_TOKEN, eos_token=EOS_TOKEN,
lower=args.lower)
MAX_LEN = args.max_lenght
if args.iwslt:
datatype = IWSLT14
else:
datatype = datasets.Multi30k
train_data, val_data, test_data = datatype.splits(
exts=('.de', '.en'), fields=(SRC, TGT),
filter_pred=lambda x: len(vars(x)['src']) <= MAX_LEN and
len(vars(x)['trg']) <= MAX_LEN
)
MIN_FREQ = args.min_freq
SRC.build_vocab(train_data.src, min_freq=MIN_FREQ)
TGT.build_vocab(train_data.trg, min_freq=MIN_FREQ)
return train_data, val_data, test_data, SRC, TGT
def build_vocabs(filepath, min_count):
"""Build the word and char counter vocabularies"""
toktok = ToktokTokenizer()
word_vocab = Counter()
char_vocab = Counter()
with open(filepath, 'r', encoding='utf8') as f:
try:
line = f.read()
if 'numbers_' in filepath:
tmp = toktok.tokenize(line.lower())
for i in range(min_count):
word_vocab.update(tmp)
else:
word_vocab.update(word_tokenize(line.lower()))
char_vocab.update(line)
except Exception as error:
print('Error with file: {}'.format(filepath))
print(error)
return word_vocab, char_vocab
def tokenize_nltk(self, text):
self.tokenizer_counter += 1
logger.debug("item:" + str(self.tokenizer_counter) +
"/" + str(self.tokenizer_len))
toktok = ToktokTokenizer()
# tokens =[toktok.tokenize(sent) for sent in sent_tokenize(text)]
tokens = nltk.word_tokenize(text)
# logger.debug("Number of tokens:" + str(len(tokens)))
stems = [self.stemmer.stem(t) for t in tokens]
return stems
def word_frequencies(contents):
toktok = ToktokTokenizer()
string_corpus = brown.raw()
# Frequencies for each file
list = []
for file in contents.keys():
print("Tokenising", file)
tokenised = [
toktok.tokenize(sent) for sent in sent_tokenize(string_corpus)
]
fdist = Counter(chain(*tokenised))
list.append(fdist)
# Combine keys into one set, eliminating duplicates
print("Making frequency distribution of all words that we care about.")
keys = []
for sublist in list:
keys += sublist
keys = set(keys)
# Build combined frequency dict
# Tuple of identifiers for connectives and other common words
unwanted = ('at', 'to', 'in', 'ma', 'bez', 'ppss', 'pp$', 'dt', 'bedz',
'hv', 'cc', 'cs', 'hvd', 'wdt', '*', 'bed', 'ber', 'be', 'np$',
'ppo', 'pps', 'abn', 'cd', 'md', 'ben', 'ben', 'wps', 'vbd',
'jj', 'rb', 'do', 'ql', 'dts', 'rp', 'in-tl', 'ex', 'i', 'dti',
'dod', 'wrb', 'hvz', 'nn$')
# This is far from the best way to do this, but I couldn't find the documentation for these identifiers
frequencies = {}
for key in keys:
total = 0
if (key[0] not in string.punctuation) and (
key.split('/')[-1]
not in unwanted): # Gets rid of unwanted tokens
for sublist in list:
if key in sublist.keys():
total += sublist[key]
frequencies[key.split('/')[0].lower()] = total
print("Total words (that we care about): " + str(len(frequencies.keys())))
return frequencies
def __init__(self, doc_map):
self.posting_list = {}
self.mine = ['br','\'','http','url','web','www','blp','ref','external','links']
self.stop_words = set(stopwords.words('english')).union(self.mine)
# self.ps = PorterStemmer().stem
self.ps = SnowballStemmer("english").stem
self.tokenizer = RegexpTokenizer(r'[a-zA-Z]+|[0-9]{,4}')
self.d = doc_map
self.sent = nltk.data.load('tokenizers/punkt/english.pickle').tokenize
self.toktok = ToktokTokenizer()
def __init__(self, doc_map):
self.posting_list = {}
self.mine = [
'br', '\'', 'http', 'url', 'web', 'www', 'blp', 'ref', 'external',
'links'
]
self.stop_words = set(stopwords.words('english')).union(self.mine)
self.ps = PorterStemmer().stem
self.tokenizer = RegexpTokenizer(r'[a-zA-Z0-9]+')
self.d = doc_map
self.t = 0
self.toktok = ToktokTokenizer()
def __init__(self, stopwords: Set[str]) -> None:
self.stopwords = stopwords
self.ps = WordNetLemmatizer()
self.stemmer = SnowballStemmer("english")
self.tokenizer = ToktokTokenizer()
self.puncuation = set(string.punctuation)
# self.words = set(nltk.corpus.words.words())
self.pipeline = [
self.remove_punctuation,
self.tokenize,
self.lowering,
self.remove_words,
self.remove_stopwords,
self.remove_digits_and_punctuation,
self.remove_dangling_puncuation,
self.remove_single,
self.stemm,
self.remove_starting_with_file,
]
self.words_to_remove = set(
"edit wookieepedia format registerr wrapup wiki sandbox click edit page link code preview button format"
.split(" "))
