15: "Social Sciences , Religion and Literature",
16: "Europe",
17: "Military and War",
18: "Africa, India and Middle East",
19: "Asia",
20: "Great Britain",
21: "Technology and Science"}
DATADIR = 'models/'
id2word = corpora.Dictionary.load_from_text(
os.path.join(DATADIR, 'wiki_wordids.txt.bz2'))
lda_model = LdaModel.load(os.path.join(DATADIR, 'lda_model'))
geoip_reader = geoip2.database.Reader(
os.path.join(DATADIR, 'GeoLite2-Country.mmdb'))
wikipedia.set_user_agent(
'EditsGeoVisualization/1.0 (http://yasermartinez.com; [email protected])')
logging.info("Loaded models")
#--------------------------------------------------------#
# Helpers #
#--------------------------------------------------------#
def wiki_bow(title):
"""This function downloads text from Wikipedia
Parameters
----------
title: string
The title of the article
for name in ['geosearch', 'languages', 'page', 'search', 'suggest', 'summary']:
setattr(
wikipedia, name,
try_again_dec(
wikipedia.exceptions.HTTPTimeoutError,
wikipedia.exceptions.RedirectError,
requests.exceptions.RequestException,
retry=3
)(getattr(wikipedia, name))
)
# wikipedia configuration
wikipedia.set_lang('en')
wikipedia.set_rate_limiting(True)
wikipedia.set_user_agent(
'Newsparser NE comparison (http://newsparser704.pythonanywhere.com/)'
)
class WikiData:
pages = 'pages'
ne_mapping = 'ne_mapping'
nlp = nlp
def __init__(self):
self.mongo_client = MongoClient()
self.db = self.mongo_client.wiki
from .utils import lazyinit
from .utils.decorators import try_again_dec
# Use try_again_dec with the main methods of wikipedia
for name in ['geosearch', 'languages', 'page', 'search', 'suggest', 'summary']:
setattr(
wikipedia, name,
try_again_dec(wikipedia.exceptions.HTTPTimeoutError,
wikipedia.exceptions.RedirectError,
requests.exceptions.RequestException,
retry=3)(getattr(wikipedia, name)))
# wikipedia configuration
wikipedia.set_lang('en')
wikipedia.set_rate_limiting(True)
wikipedia.set_user_agent(
'Newsparser NE comparison (http://newsparser704.pythonanywhere.com/)')
class WikiData:
pages = 'pages'
ne_mapping = 'ne_mapping'
nlp = nlp
def __init__(self):
self.mongo_client = MongoClient()
self.db = self.mongo_client.wiki
def __del__(self):
self.mongo_client.close()