def check_and_install_transliteration(force=False):
# 'en' is not a supported transliteration language
transliteration_languages = [l.code for l in languages if l.code != "en"]
# check installed
install_needed = []
if not force:
t_packages_path = Path(polyglot.polyglot_path) / "transliteration2"
for lang in transliteration_languages:
if not (t_packages_path / lang /
f"transliteration.{lang}.tar.bz2").exists():
install_needed.append(lang)
else:
install_needed = transliteration_languages
# install the needed transliteration packages
if install_needed:
print(
f"Installing transliteration models for the following languages: {', '.join(install_needed)}"
)
from polyglot.downloader import Downloader
downloader = Downloader()
for lang in install_needed:
downloader.download(f"transliteration2.{lang}")
def ner():
if not request.json or not 'text' in request.json:
abort(400)
input_text = request.json['text'].strip()
if input_text == '':
return jsonify({}), 200
text = Text(input_text)
if 'lang' in request.json:
downloader = Downloader(download_dir=polyglot.data_path+'/polyglot_data')
supported_languages = [x.language for x in downloader.get_collection(task="ner2").packages]
lang_ = request.json['lang']
if lang_ not in supported_languages:
abort(400, {'message': 'language {} is not supported'.format(lang_)})
text.language = lang_
entities = text.entities
result = {}
for entity in entities:
if entity.tag not in result:
result[entity.tag] = set()
result[entity.tag].add(' '.join(entity))
for tagType in result:
result[tagType] = list(result[tagType])
return jsonify(result), 200
def status():
data = dict(default_data)
data['missing_libraries'] = []
try:
import textblob
except ImportError:
data['missing_libraries'].append('textblob')
try:
import spacy
except ImportError:
data['missing_libraries'].append('spacy')
try:
import gensim
except ImportError:
data['missing_libraries'].append('gensim')
try:
import newspaper
except ImportError:
data['missing_libraries'].append('newspaper')
try:
import langid
except ImportError:
data['missing_libraries'].append('langid')
try:
import readability
except ImportError:
data['missing_libraries'].append('readability')
try:
import bs4
except ImportError:
data['missing_libraries'].append('bs4')
try:
import afinn
except ImportError:
data['missing_libraries'].append('afinn')
try:
import polyglot
except ImportError:
data['missing_libraries'].append('polyglot')
else:
from polyglot.downloader import Downloader
dwnld = Downloader()
data['polyglot_lang_models'] = {}
for info in sorted(dwnld.collections(), key=str):
status = dwnld.status(info)
if info.id.startswith('LANG:') and status != 'not installed':
data['polyglot_lang_models'][info.id] = status
return jsonify(data)
def download(args):
""" Download polyglot packages and models."""
downloader = Downloader(server_index_url = args.server_index_url)
if args.packages:
for pkg_id in args.packages:
rv = downloader.download(info_or_id=unicode(pkg_id), download_dir=args.dir,
quiet=args.quiet, force=args.force,
halt_on_error=args.halt_on_error)
if rv == False and args.halt_on_error:
break
else:
downloader.download(download_dir=args.dir, quiet=args.quiet, force=args.force,
halt_on_error=args.halt_on_error)
def _download_polyglot_data():
downloader = Downloader()
# Download PT and ES embeddings for mention detection
downloader.download('embeddings2.pt')
downloader.download('embeddings2.es')
# Download NER models
downloader.download('ner2.pt')
downloader.download('ner2.es')
def polyglot_default_install():
"""checking and caching polyglot"""
try:
os.path.sep = '/'
polyglot_path = POLIGLOT['path_polyglot_data']
polyglot_path = tools.get_abs_path(polyglot_path)
if not os.path.exists(polyglot_path):
os.makedirs(polyglot_path)
load.polyglot_path = polyglot_path
downloader = Downloader(download_dir=polyglot_path)
for language in SERVER['language']:
sentiment = False
entities = False
morph = False
pos = False
full_name = Language.from_code(language).name
if language in get_supported_languages('embeddings2'):
#if not downloader.is_installed(unicode('embeddings2.' + language)):
if not downloader.is_installed('embeddings2.' + language):
raise EnvironmentError(
'The {0} module for {1}({2}) was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'embeddings2', full_name, language))
if language in get_supported_languages('ner2'):
entities = True
#if not downloader.is_installed(unicode('ner2.' + language)):
if not downloader.is_installed('ner2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'ner2', full_name))
if language in get_supported_languages('sentiment2'):
sentiment = True
#if not downloader.is_installed(unicode('sentiment2.' + language)):
if not downloader.is_installed('sentiment2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'sentiment2', full_name))
if language in get_supported_languages('morph2'):
morph = True
# if not downloader.is_installed(unicode('morph2.' + language)):
if not downloader.is_installed('morph2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'morph2', full_name))
if language in get_supported_languages('pos2'):
pos = True
# if not downloader.is_installed(unicode('pos2.' + language)):
if not downloader.is_installed('pos2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'pos2', full_name))
# FOR POLYGLOT DOWNLOAD ON START SERVER
if sentiment or entities or morph:
text_polyglot = Text('Testing and cashing',
hint_language_code=language)
if sentiment:
_ = text_polyglot.words[0].polarity
_ = text_polyglot.sentences[0].polarity
if entities:
_ = text_polyglot.entities
if morph:
_ = text_polyglot.morphemes
except:
ex_type, ex, tb = sys.exc_info()
# tools.message_box(str(ex) + 'TracebackError'+''.join(traceback.format_exc()),
# str(ex_type), 0)
raise EnvironmentError(
str(ex) + 'TracebackError' + ''.join(traceback.format_exc()))
