def make_concept_uri(text, lang, disambiguation=None):
text = ftfy.ftfy(text).strip()
if disambiguation is None:
text, disambiguation = handle_disambig(text)
if disambiguation is not None:
if isinstance(disambiguation, str):
disambiguation = disambiguation.decode("utf-8")
disambiguation = ftfy.ftfy(disambiguation)
if lang == "en":
normalized = english.normalize(text)
elif lang == "ja" and disambiguation is not None:
match = re.search(r"\((.*?)\)", disambiguation)
if match:
parenthesized = match.group(1)
pos, rest = disambiguation.split("/", 1)
if parenthesized in JAPANESE_PARTS_OF_SPEECH:
pos = JAPANESE_PARTS_OF_SPEECH[parenthesized]
else:
pos = "n"
disambiguation = pos + "/" + re.sub(r"\s*\((.*?)\)\s*", "", rest)
normalized = preprocess_text(text).lower()
else:
normalized = preprocess_text(text).lower()
if disambiguation is not None:
disambiguation = disambiguation.strip().replace(" ", "_").lower()
if disambiguation:
return "/c/%s/%s/%s" % (lang, normalized.replace(" ", "_"), disambiguation)
else:
return "/c/%s/%s" % (lang, normalized.replace(" ", "_"))
def make_concept_uri(text, lang, disambiguation=None):
text = ftfy.ftfy(text).strip()
if disambiguation is None:
text, disambiguation = handle_disambig(text)
if disambiguation is not None:
if isinstance(disambiguation, str):
disambiguation = disambiguation.decode('utf-8')
disambiguation = ftfy.ftfy(disambiguation)
if lang == 'en':
normalized = normalize(text)
elif lang == 'ja' and disambiguation is not None:
match = re.search(r'\((.*?)\)', disambiguation)
if match:
parenthesized = match.group(1)
pos, rest = disambiguation.split('/', 1)
if parenthesized in JAPANESE_PARTS_OF_SPEECH:
pos = JAPANESE_PARTS_OF_SPEECH[parenthesized]
else:
pos = 'n'
disambiguation = pos + '/' + re.sub(r'\s*\((.*?)\)\s*', '', rest)
normalized = text.lower()
else:
normalized = text.lower()
if disambiguation is not None:
disambiguation = disambiguation.strip().replace(' ', '_').lower()
if disambiguation:
return '/c/%s/%s/%s' % (lang, normalized.replace(' ', '_'), disambiguation)
else:
return '/c/%s/%s' % (lang, normalized.replace(' ', '_'))
def make_concept_uri(text, lang, disambiguation=None):
text = ftfy.ftfy(text)
if disambiguation is None:
text, disambiguation = handle_disambig(text)
if disambiguation is not None:
if isinstance(disambiguation, str):
disambiguation = disambiguation.decode('utf-8')
disambiguation = ftfy.ftfy(disambiguation)
if lang == 'en':
normalized = english.normalize(text)
elif lang == 'ja' and disambiguation is not None:
match = re.search(r'\((.*?)\)', disambiguation)
if match:
parenthesized = match.group(1)
pos, rest = disambiguation.split('/', 1)
if parenthesized in JAPANESE_PARTS_OF_SPEECH:
pos = JAPANESE_PARTS_OF_SPEECH[parenthesized]
else:
pos = 'n'
disambiguation = pos + '/' + re.sub(r'\s*\((.*?)\)\s*', '', rest)
normalized = preprocess_text(text).lower()
else:
normalized = preprocess_text(text).lower()
if disambiguation is not None:
disambiguation = disambiguation.replace(' ', '_')
if disambiguation:
return '/c/%s/%s/%s' % (lang, normalized.replace(' ', '_'), disambiguation)
else:
return '/c/%s/%s' % (lang, normalized.replace(' ', '_'))
def load_data(messages_directory):
for threadname in os.listdir(messages_directory):
if threadname.startswith('.'):
continue
if threadname == 'stickers_used':
continue
# the messages are stored in the "message.json" file in the threadname directory
# we want to load those json dictionaries, put them in the threads list, and do some data conversion
filename = os.path.join(messages_directory, threadname, "message.json")
threaddict = {}
with open(filename) as f:
threaddict = json.load(f)
# alter every message a bit
for message in threaddict['messages']:
# now convert the timestamp into a real datetime object
message['date'] = datetime.fromtimestamp(message['timestamp'])
# for consistency, change sender_name to sender
message['sender'] = message['sender_name']
if 'content' not in message:
message['content'] = ''
message['content'] = ftfy.ftfy(message['content'])
message['sender'] = ftfy.ftfy(message['sender'])
# for consistency, copy participants to members
threaddict['members'] = []
if 'participants' in threaddict:
for participant in threaddict['participants']:
threaddict['members'].append(ftfy.ftfy(participant))
threaddict['title'] = ftfy.ftfy(threaddict['title'])
threaddict['members'].append(setup.user)
# add index
global threads
threaddict['index'] = len(threads)
threads.append(threaddict)
def safe_path(origtitle):
title = safe_path_component(ftfy(origtitle))
if len(title) == 0:
title = origtitle = u'_'
if title.startswith(u'-') or title.startswith(u'.'):
title = u'_' + title
try:
charname = safe_path_component(unicodedata.name(origtitle[0]))
except ValueError:
charname = u'UNKNOWN'
category = charname.split('_')[0]
# some ridiculous stuff to give every article a unique name that can be
# stored on multiple file systems and tab-completed
if len(origtitle) == 1:
pieces = [u'single_character', category, charname + '.json']
else:
try:
charname2 = safe_path_component(unicodedata.name(origtitle[1]))
except ValueError:
charname2 = u'UNKNOWN'
text_to_encode = unicodedata.normalize("NFKD",
safe_path_component(title[:64]))
finalpart = text_to_encode.encode('punycode').rstrip('-')
pieces = [charname, charname2, finalpart + '.json']
path = u'/'.join(pieces)
return path
def leeds_corpus_frequencies(corpusfile, stemmer):
if stemmer is None:
stemmer = lambda x: x
infile = codecs.open(corpusfile, encoding='utf-8')
freqs = defaultdict(int)
tokenfreqs = defaultdict(int)
for line in infile:
line = ftfy(line.strip())
if line:
rank = line.split(' ')[0]
if NUMBER_RE.match(rank) and line.count(' ') == 2:
rank, freq, token = line.split(' ')
stemmed = stemmer(token)
print "%s -> %s" % (token, stemmed)
freq = float(freq)
freq_int = int(freq*100)
for word in stemmed.split(' '):
if ',' not in word:
freqs[word] += freq_int
if ',' not in token:
tokenfreqs[token.lower()] += freq_int
for key in tokenfreqs:
if tokenfreqs[key] > freqs[key]:
freqs[key] = tokenfreqs[key]
return freqs
def safe_path(origtitle):
title = safe_path_component(ftfy(origtitle))
if len(title) == 0:
title = origtitle = u'_'
if title.startswith(u'-') or title.startswith(u'.'):
title = u'_' + title
try:
charname = safe_path_component(unicodedata.name(origtitle[0]))
except ValueError:
charname = u'UNKNOWN'
category = charname.split('_')[0]
# some ridiculous stuff to give every article a unique name that can be
# stored on multiple file systems and tab-completed
if len(origtitle) == 1:
pieces = [u'single_character', category, charname + '.json']
else:
try:
charname2 = safe_path_component(unicodedata.name(origtitle[1]))
except ValueError:
charname2 = u'UNKNOWN'
text_to_encode = unicodedata.normalize("NFKD", safe_path_component(title[:64]))
finalpart = text_to_encode.encode('punycode').rstrip('-')
pieces = [charname, charname2, finalpart + '.json']
path = u'/'.join(pieces)
return path
def load(self, path_to_file):
"""Loads .txt file from `path_to_file`.
Arguments:
path_to_file (pathlib.Path):
Path to .txt file
Returns:
doc (chomskIE.utils.Document)
Document object corresponding to .txt file in `path_to_file`.
"""
if not self._validate_data_path(path_to_file, is_directory=False):
raise PathError(f'{path_to_file} is not a valid file path.')
try:
text_obj = open(path_to_file, 'r')
text = text_obj.read()
except UnicodeDecodeError:
text_obj = open(path_to_file, 'rb')
text, _ = ftfy.guess_bytes(text_obj.read())
text = ftfy.ftfy(text)
name = str(path_to_file).split('/')[-1]
paragraphs = [p.strip() for p in text.splitlines() if p]
doc = Document(name=name, text=text, paragraphs=paragraphs)
return doc
def load(self, english_model, path_to_file):
"""Loads .txt file from `path_to_file`.
Arguments:
english_model (spacy.lang)
Trained SpaCy language pipeline.)
path_to_file (pathlib.Path):
Path to .txt file
Returns:
doc, spacy_doc (tuple)
``doc`` is a ``chomskIE.utils.Document`` object corresponding
to .txt file in `path`.
``spacy_doc`` is a ``spacy.tokens.Document`` object corresponding
to .txt files in `path` processed by ``english_model``.
"""
if not self._validate_data_path(path_to_file, is_directory=False):
raise PathError(f'{path_to_file} is not a valid file path.')
try:
text_obj = open(path_to_file, 'r')
text = text_obj.read()
except UnicodeDecodeError:
text_obj = open(path_to_file, 'rb')
text, _ = ftfy.guess_bytes(text_obj.read())
text = ftfy.ftfy(text)
name = str(path_to_file).split('/')[-1]
spacy_doc = english_model(text)
doc = Document(name=name, text=None, paragraphs=None)
return doc, spacy_doc
def _read_csv(reader, header, encoding):
"""
Given a constructed CSV reader object, a header row that we've read, and
a detected encoding, yield its rows as dictionaries.
"""
for row in reader:
if len(row) == 0:
continue
row = [ftfy(cell.decode(encoding, 'replace')) for cell in row]
row_list = zip(header, row)
row_dict = dict(row_list)
if len(row_dict['text']) == 0:
continue
row_dict['text'] = unicodedata.normalize(
'NFKC', row_dict['text'].strip()
)
if row_dict.get('title') == '':
del row_dict['title']
if 'date' in row_dict:
if row_dict['date'] == '':
del row_dict['date']
else:
row_dict['date'] = int(row_dict['date'])
if 'query' in row_dict or 'subset' in row_dict:
queries = [cell[1] for cell in row_list
if cell[1] != '' and
(cell[0] == 'query' or cell[0] == 'subset')]
if queries:
row_dict['queries'] = queries
if 'query' in row_dict:
del row_dict['query']
if 'subset' in row_dict:
del row_dict['subset']
yield row_dict
def clean_data(data):
'''
Augment the raw Facebook data for our graphing use cases
'''
# set timezone
data['datetime'] = pd.DatetimeIndex(
pd.to_datetime(data['timestamp_ms'],
unit='ms')).tz_localize('UTC').tz_convert(
config.TIMEZONE)
# column for just date
data['date'] = data["datetime"].apply(lambda d: datetime.datetime(
year=d.year, month=d.month, day=d.day)).map(lambda x: x.date())
# column for term of date
data['term'] = pd.to_datetime(
data['datetime']).apply(lambda d: "{} {}".format(
d.strftime('%Y'), util.to_term(int(d.strftime('%m')))))
# clean up sticker data
data['sticker'] = data['sticker'].apply(lambda s: s['uri']
if not pd.isnull(s) else None)
duplicate_likes = [
"messages/stickers_used/851582_369239386556143_1497813874_n_369239383222810.png",
"messages/stickers_used/851587_369239346556147_162929011_n_369239343222814.png"
]
data['sticker'] = data['sticker'].replace(
duplicate_likes,
"messages/stickers_used/851557_369239266556155_759568595_n_369239263222822.png"
)
# format text properly
data['content'] = data['content'].apply(lambda x: ftfy.ftfy(x)
if type(x) == str else x)
# properly set message type, adding types 'Game', 'Plan Update', 'Chat Update'
warnings.filterwarnings("ignore", 'This pattern has match groups')
data['game'] = data['content'].str.contains(chatstats_constants.GAME_REGEX,
na=False)
data['plan_update'] = data['content'].str.contains(
chatstats_constants.PLAN_UPDATE_REGEX, na=False)
data['chat_update'] = data['content'].str.contains(
chatstats_constants.CHAT_UPDATE_REGEX, na=False)
data['call_update'] = data['content'].str.contains(
chatstats_constants.CALL_UPDATE_REGEX, na=False)
data['type'] = data.apply(lambda x: clean_type(x), axis=1)
# add first name column
data['sender_first_name'] = data['sender_name'].apply(
lambda s: s.split()[0])
return data
def open_csv_somehow(filename):
"""
Given a filename that we're told is a CSV file, detect its encoding,
parse its header, and return a generator yielding its rows as dictionaries.
Use the `ftfy` module internally to fix Unicode problems at the level that
chardet can't deal with.
"""
encoding = detect_file_encoding(filename)
csvfile = open(filename, 'rU')
reader = csv.reader(csvfile, dialect='excel')
header = reader.next()
header = [ftfy(cell.decode(encoding).lower()) for cell in header]
return _read_csv(reader, header, encoding)
def open_csv_somehow_py3(filename):
encoding = detect_file_encoding(filename)
csvfile = open(filename, 'rU', encoding=encoding)
line = csvfile.readline()
csvfile.seek(0)
if '\t' in line:
# tab-separated
reader = csv.reader(csvfile, delimiter='\t')
else:
reader = csv.reader(csvfile, dialect='excel', newline='')
header = reader.next()
header = [ftfy(cell.lower().strip()) for cell in header]
return _read_csv(reader, header, encoding)
def read_leeds_corpus(filename):
"""
Load word frequencies from a "Web as Corpus" file, collected and
provided by the University of Leeds.
For more information, see: http://corpus.leeds.ac.uk/list.html
"""
infile = codecs.open(filename, encoding='utf-8')
counts = defaultdict(float)
for line in infile:
line = line.rstrip()
if line:
rank = line.split(u' ')[0]
if NUMBER_RE.match(rank) and line.count(u' ') == 2:
_, freq, token = line.split(u' ')
token = standardize_word(ftfy(token))
freq = float(freq)
counts[token] += freq
return _scale_freqs(counts)
def open_csv_somehow_py2(filename):
"""
Open a CSV file using Python 2's CSV module, working around the deficiency
where it can't handle the null bytes of UTF-16.
"""
encoding = detect_file_encoding(filename)
if encoding.startswith('UTF-16'):
csvfile = transcode_to_utf8(filename, encoding)
encoding = 'UTF-8'
else:
csvfile = open(filename, 'rU')
line = csvfile.readline()
csvfile.seek(0)
if '\t' in line:
# tab-separated
reader = csv.reader(csvfile, delimiter='\t')
else:
reader = csv.reader(csvfile, dialect='excel')
header = reader.next()
header = [ftfy(cell.decode(encoding).lower().strip()) for cell in header]
return _read_csv(reader, header, encoding)
def fix_heading(heading):
return ftfy(heading).strip('[]')
def fix_heading(heading):
return ftfy(heading).strip("[]")
def c_cleverbot(client, message):
yield from client.send_message(
message.channel,
ftfy.ftfy(sh.mention(message) + cw.say(sh.get_args(message, True))))
def fix_heading(heading):
return ftfy(heading).strip('[]')
def fix(text):
return ftfy(text).lower()
def encodeText(text):
return ftfy.ftfy(text)
def format(s): return ftfy.ftfy(s) if type(s) == str else s def format_obj(obj: object):
