class NerService(tornado.web.RequestHandler):
def initialize(self, concepts_inlinks, stopwords, entities, inlinks_threshold=400, MAX_WORDS=400, MAX_CHARS=2000):
"""
"""
self.MAX_WORDS = MAX_WORDS
self.MAX_CHARS = MAX_CHARS
self.ner = Ner(concepts_inlinks, entities, stopwords, inlinks_threshold=inlinks_threshold, max_words=MAX_WORDS)
def get(self):
# Get parameters
inlinks_threshold = int(self.get_argument("inlinks_threshold", default=400))
self.ner.inlinks_threshold=inlinks_threshold
text = self.get_argument("text")
debug = self.get_argument("debug", default=False)
# Check warnings if exists
warning = []
if len(text) > self.MAX_CHARS:
warning.append('Only the first %d chars will be processed. This request is over this limit.' % self.MAX_CHARS)
if len(text.split(' ')) > self.MAX_WORDS:
warning.append('Only the first %d words will be processed. This request is over this limit.' % self.MAX_WORDS)
#result = json.loads(self.ner.fetch_entities(text))
result = self.ner.fetch_entities(text)
# Erase text at response
del(result['text'])
# if exists warning, append the flags to the output
if len(warning) > 0:
result['warnings'] = warning
if debug:
self.write(result)
else:
self.write({"concepts": list(result["results"].keys())})
def post(self):
results = list()
for line in str(self.request.body, 'utf8').split('\n'):
if line:
fields = line.split('\t')
text = fields[0]
#concepts = self.__format_post_result(self.ner.fetch_entities(text))
concepts = self.ner.fetch_entities(text)
print(concepts['results'])
#result_fields = [concepts] + fields
#result = '\t'.join(result_fields)
concept_names = list(concepts['results'].keys())
results.append({"text":text, "concepts":concept_names})
#for temp in results:
# self.write('%s\n' % (temp))
print(results)
self.write({"response":results})
def __format_post_result(self, response):
response_dict = json.loads(response)
concepts = response_dict['results'].keys()
if concepts:
return ";;".join(concepts)
else:
return ""
class NerService(tornado.web.RequestHandler):
def initialize(self,
concepts_inlinks,
stopwords,
inlinks_threshold=400,
MAX_WORDS=400,
MAX_CHARS=2000):
"""
"""
self.MAX_WORDS = MAX_WORDS
self.MAX_CHARS = MAX_CHARS
self.inlinks_threshold = inlinks_threshold
self.ner = Ner(concepts_inlinks,
stopwords,
inlinks_threshold=inlinks_threshold,
max_words=MAX_WORDS)
def get(self):
# Get parameters
inlinks_threshold = int(
self.get_argument("inlinks_threshold",
default=self.inlinks_threshold))
self.ner.inlinks_threshold = inlinks_threshold
text = self.get_argument("text")
debug = self.get_argument("debug", default=False)
# Check warnings if exists
warning = []
if len(text) > self.MAX_CHARS:
warning.append(
'Only the first %d chars will be processed. This request is over this limit.'
% self.MAX_CHARS)
if len(text.split(' ')) > self.MAX_WORDS:
warning.append(
'Only the first %d words will be processed. This request is over this limit.'
% self.MAX_WORDS)
result = self.ner.fetch_entities(text)
# Erase text at response
del (result['text'])
# if exists warning, append the flags to the output
if len(warning) > 0:
result['warnings'] = warning
if debug:
self.write(result)
else:
self.write({"concepts": list(result["results"].keys())})
def post(self):
results = list()
for line in str(self.request.body, 'utf8').split('\n'):
if line:
fields = line.split('\t')
text = fields[0]
concepts = self.ner.fetch_entities(text)
concept_names = list(concepts['results'].keys())
results.append({"text": text, "concepts": concept_names})
self.write({"response": results})
def __format_post_result(self, response):
response_dict = json.loads(response)
concepts = response_dict['results'].keys()
if concepts:
return ";;".join(concepts)
else:
return ""
