"""
Description: This module is the manually starting point our Tutti Salesman application.
Author: Josef Sertl
Copyright: SERTL Analytics, https://sertl-analytics.com
Date: 2019-04-02
"""
from salesman_tutti.tutti import Tutti
from time import sleep
from salesman_system_configuration import SystemConfiguration
from salesman_search import SalesmanSearchApi
from sertl_analytics.constants.salesman_constants import REGION
from salesman_tutti.tutti_constants import PRSUBCAT, PRCAT
from salesman_nlp.salesman_spacy import SalesmanSpacy
from spacy import displacy
my_spacy = SalesmanSpacy()
doc = my_spacy.nlp('Das ist ein Satz')
doc = my_spacy.nlp(
'Leder schwarz. Den Widerstand der Rückenlehne kann man noch einstellen. '
'Untergestell hat leichte Kratzspuren. Leder ist aber wie neu.')
displacy.serve(doc, style='dep')
"""
Using the 'dep' visualizer
Serving on http://0.0.0.0:5000 ...
NOW: open http://localhost:5000/
"""
class Test4SalesmanSpacy:
def __init__(self):
self._salesman_spacy = SalesmanSpacy(load_sm=True)
def run_test(self, print_token=False):
test_case_dict = self.__get_test_case_dict__()
tc_handler = MyTestCaseHandler('Testing "{}":'.format(
self.__class__.__name__))
api = SalesmanSearchApi('')
for key, test_case_list in test_case_dict.items():
for tc in test_case_list:
doc = self._salesman_spacy.nlp(tc[0])
expected = tc[1]
self._salesman_spacy.print_entities_for_doc(doc)
if print_token:
self._salesman_spacy.print_tokens_for_doc(doc)
result = self._salesman_spacy.get_entity_list(doc)
tc_handler.add_test_case(MyTestCase(key, tc, expected, result))
tc_handler.print_result()
def __get_test_case_dict__(self):
return {
'REPLACEMENT': [
[
'Das ist ein Text mit GoreTex und goretex und Goretex und nochmals Gore-Tex',
'Goretex (MATERIAL)'
],
],
'COMPANY': [
[
'BMW und Apple sind beides tolle Firmen',
'Apple (COMPANY), BMW (COMPANY)'
],
],
'PRODUCT': [
[
'Die Serie 7 ist gut. Das beste an der Serie 7 ist der Preis',
'Serie 7 (PRODUCT)'
],
[
'Vitra Aluminium Chair EA107 oder EA 108 oder MacBook',
'Aluminium (MATERIAL), EA 108 (PRODUCT), EA107 (PRODUCT), MacBook (PRODUCT), Vitra (COMPANY)'
],
[
'Vitra Bürostuhl EA108 günstig',
'Bürostuhl (OBJECT), EA108 (PRODUCT), Vitra (COMPANY)'
],
],
'UNKNOWN_ORG': [
[
'Audi und Volvo bauen die sichersten Autos.',
'Audi (COMPANY), Volvo (COMPANY)'
],
[
'Im Dow Jones sind u.a. Travelers Companies und Verizon enthalten',
'Dow Jones (LOC), Travelers Companies (COMPANY), Verizon (COMPANY)'
],
],
'TEST': [
[
'Winterschuhe Kinder, Lowa, Rufus III GTX, Gr. 37, Goretex Sommerhut, Strohhut',
'Dow Jones (LOC), Travelers Companies (COMPANY), Verizon (COMPANY)'
],
]
}