def get_label(alt_url='https://api.rosette.com/rest/v1/', sentence = "Default Sentence"):
api = API(user_key=API_KEYS.ROSETTE_API_KEY, service_url=alt_url)
params = DocumentParameters()
label_set = set()
final_response_list = []
params["content"] = sentence
params["language"] = "eng"
try:
response_dict = json.loads(json.dumps(api.categories(params)))["categories"]
response_list = []
for item in response_dict:
response_list.append([str(item["label"]).replace("_", " "), float(item["confidence"])])
response_list_sorted = sorted(response_list, key=getKey, reverse=True)
for item in response_list_sorted:
label = item[0]
if not label in label_set:
final_response_list.append(item)
label_set.add(label)
return final_response_list[:5]
except RosetteException as exception:
print(exception)
def run(key, altUrl='https://api.rosette.com/rest/v1/'):
url = "Sony Pictures is planning to shoot a good portion of the new "Ghostbusters" in Boston as well."
# Create an API instance
api = API(user_key=key, service_url=altUrl)
params = DocumentParameters()
# Use a URL to input data instead of a string
params["contentUri"] = url
return api.categories(params)
def run(alt_url='https://api.rosette.com/rest/v1/', sentence = "Default Sentence"):
api = API(user_key=API_KEYS.ROSETTE_API_KEY, service_url=alt_url)
params = DocumentParameters()
params["content"] = sentence
try:
return api.categories(params)
except RosetteException as exception:
print(exception)
def run(key, altUrl='https://api.rosette.com/rest/v1/'):
categories_url_data = "http://www.onlocationvacations.com/2015/03/05/the-new-ghostbusters-movie-begins-filming-in-boston-in-june/"
url = categories_url_data
# Create an API instance
api = API(user_key=key, service_url=altUrl)
params = DocumentParameters()
# Use a URL to input data instead of a string
params["contentUri"] = url
return api.categories(params)
def run(key, alt_url='https://api.rosette.com/rest/v1/'):
""" Run the example """
categories_url_data = "https://onlocationvacations.com/2018/02/06/downton-abbey-exhibition-extended-april-2-nyc/"
url = categories_url_data
# Create an API instance
api = API(user_key=key, service_url=alt_url)
# Set selected API options
# For more information on the functionality of these
# and other available options, see Rosette Features & Functions
# https://developer.rosette.com/features-and-functions#categorization
# api.set_option('singleLabel', 'true')
# api.set_option('scoreThreshold',- 0.20)
params = DocumentParameters()
# Use a URL to input data instead of a string
params["contentUri"] = url
try:
return api.categories(params)
except RosetteException as exception:
print(exception)
"""
Example code to call Rosette API to get the category of a document (at a given URL).
"""
import argparse
import pprint
from rosette.api import API, DocumentParameters
parser = argparse.ArgumentParser(description="Get the category of a piece of a document at a URL")
parser.add_argument("--key", required=True, help="Rosette API key")
parser.add_argument("--service_url", nargs="?", help="Optional user service URL")
parser.add_argument("--url", nargs="?", default="http://www.basistech.com/about/", help="Optional URL for data")
args = parser.parse_args()
# Create an API instance
if args.service_url:
api = API(service_url=args.service_url, user_key=args.key)
else:
api = API(user_key=args.key)
params = DocumentParameters()
# Use a URL to input data instead of a string
params["contentUri"] = args.url
result = api.categories(params)
pprint.pprint(result)
class MonitorBasisAPI():
API_KEY = "cfdb0d7990f5c1d2546c61a94cf0a445"
DO_NOT_CALL = False
def __init__(self):
self.rosette_handle = API(user_key=self.API_KEY)
# altUrl='https://api.rosette.com/rest/v1/'
# OPTION: service_url=altUrl
result = self.rosette_handle.ping()
print("/ping: ", result)
def temp_demo(self):
pass
# NOT TESTED
def get_basis_sentiment(self, txt_str):
params = DocumentParameters()
params["language"] = "eng"
# Use an HTML file to load data instead of a string
params.load_document_string(txt_str)
try:
result = self.rosette_handle.sentiment(params)
except RosetteException as e:
print(e)
finally:
print " DONE "
# NOT TESTED
def get_basis_categories(self, txt_str):
# if self.DO_NOT_CALL == True:
# return "POS"
params = DocumentParameters()
params["language"] = "eng"
# Use an HTML file to load data instead of a string
params.load_document_string(txt_str)
typed_relations_resp = self.rosette_handle.categories(text=txt_str)
print " >>> CALLED IBM - typed_relations_resp: " + str(
typed_relations_resp)
return json.dumps(typed_relations_resp)
def get_basis_service_by_tag(self, txt_str, basis_str):
'''
To Be Determined:
- syntax_dependencies
- morphology
- name_translation
- name_similarity
'''
print "~~~~~~~~~~~~~~~~~~~~~~~~~~~ going to get_basis_service_by_tag: " + str(
basis_str)
# if ibm_str == "authors":
# return self.get_basis_service(txt_str, self.rosette_handle.authors)
# elif ibm_str == "concepts":
# return self.get_basis_service(txt_str, self.rosette_handle.concepts)
# elif ibm_str == "dates":
# return self.get_basis_service(txt_str, self.rosette_handle.dates)
# elif ibm_str == "emotion":
# return self.get_basis_service(txt_str, self.rosette_handle.emotion)
# elif ibm_str == "feeds":
# return self.get_basis_service(txt_str, self.rosette_handle.feeds)
# elif ibm_str == "keywords":
# return self.get_basis_service(txt_str, self.rosette_handle.keywords)
# elif ibm_str == "microformats":
# return self.get_basis_service(txt_str, self.rosette_handle.microformats)
# elif ibm_str == "publication_date":
# return self.get_basis_service(txt_str, self.rosette_handle.publication_date)
#
# CURRENT LIST: "entities", "language", "relationships", "sentences", "sentiment", "syntax_dependencies", "tokens"
if basis_str == "entities":
return self.get_basis_service(
txt_str, self.rosette_handle.entities
) # also could use params["genre"] = "social-media"
elif basis_str == "language":
return self.get_basis_service(txt_str,
self.rosette_handle.language) #!
elif basis_str == "relationships":
return self.get_basis_service(
txt_str, self.rosette_handle.relationships) #!
elif basis_str == "sentences":
return self.get_basis_service(txt_str,
self.rosette_handle.sentences) #!
elif basis_str == "sentiment":
return self.get_basis_service(txt_str,
self.rosette_handle.sentiment)
elif basis_str == "syntax_dependencies":
return self.get_basis_service(
txt_str, self.rosette_handle.syntax_dependencies)
elif basis_str == "tokens":
return self.get_basis_service(txt_str,
self.rosette_handle.tokens) #!
# elif ibm_str == "title":
# return self.get_basis_service(txt_str, self.rosette_handle.title)
# elif ibm_str == "typed_relations":
# return self.get_basis_service(txt_str, self.rosette_handle.typed_relations)
def get_basis_service(self, txt_str, basis_func):
if self.DO_NOT_CALL == True:
return "POS"
callStart = datetime.datetime.now()
params = DocumentParameters()
params["language"] = "eng"
# Use an HTML file to load data instead of a string
params.load_document_string(txt_str)
# TODO: determine when to use
# params["content"] vs load_document_string
resp = {"return_data": basis_func(params)}
# print " >>> CALLED IBM - resp: " + str(resp)
resp_time = (datetime.datetime.now() - callStart).total_seconds()
resp['resp_time'] = resp_time
return resp
# mba = MonitorBasisAPI()
