def handle_callback_for_search_markdown(*params):
self._callback_for_search_input_markdown.set_values(*params)
actual_search_value = self._callback_for_search_input_markdown.get_actual_search_value(
)
if actual_search_value == '':
return ''
prefix_doc_dict = {
'Original': self._salesman_spacy.nlp_sm(actual_search_value),
'Modified': self._salesman_spacy.nlp(actual_search_value)
}
value_list_dict = {
prefix:
SalesmanSpacy.get_sorted_entity_label_values_for_doc(doc)
for prefix, doc in prefix_doc_dict.items()
}
value_joined_list = [
'**{}**: {}'.format(prefix, values)
for prefix, values in value_list_dict.items()
]
return ' \n'.join(value_joined_list)
def __init__(self):
self.sys_config = SystemConfiguration()
self._url_search_q = 'https://www.tutti.ch/de/li/ganze-schweiz?q='
self._spacy = SalesmanSpacy(load_sm=True)
"""
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/
"""
def __init__(self):
self._salesman_spacy = SalesmanSpacy(load_sm=True)
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)'
],
]
}
TC_MATCHER_IS_LIKE_NEW = 'TC_MATCHER_IS_LIKE_NEW'
TC_MATCHER_SIZE = 'TC_MATCHER_SIZE'
TC_MATCHER_NUMBER = 'TC_MATCHER_NUMBER'
TC_MATCHER_ORIGINAL_PRICE = 'TC_MATCHER_ORIGINAL_PRICE'
TC_MATCHER_IS_USED = 'TC_MATCHER_IS_USED'
TC_MATCHER_IS_TOTAL_PRICE = 'TC_MATCHER_IS_TOTAL_PRICE'
TC_MATCHER_SINGLE_PRICE = 'TC_MATCHER_SINGLE_PRICE'
TC_MATCHER_COVER_AVAILABLE = 'TC_MATCHER_COVER_AVAILABLE'
TC_MATCHER_AGE = 'TC_MATCHER_AGE'
TC_MATCHER_USAGE = 'TC_MATCHER_USAGE'
TC_MATCHER_WARRANTY = 'TC_MATCHER_WARRANTY'
tc = TC.TC_MATCHER_ORIGINAL_PRICE
spacy = SalesmanSpacy()
if tc == TC.TC_MATCHER_IS_NEW:
matcher = TuttiMatcher4IsNew(spacy.nlp)
matcher.run_test(spacy, False)
elif tc == TC.TC_MATCHER_IS_LIKE_NEW:
matcher = TuttiMatcher4IsLikeNew(spacy.nlp)
matcher.run_test(spacy, False)
elif tc == TC.TC_MATCHER_IS_USED:
matcher = TuttiMatcher4IsUsed(spacy.nlp)
matcher.run_test(spacy, False)
elif tc == TC.TC_MATCHER_SIZE:
matcher = TuttiMatcher4Size(spacy.nlp)
matcher.run_test(spacy, False)
elif tc == TC.TC_MATCHER_ORIGINAL_PRICE:
matcher = TuttiMatcher4PriceOriginal(spacy.nlp)
def __init__(self):
self._salesman_spacy = SalesmanSpacy(load_sm=True)
self._sys_config = SystemConfiguration()
self._factory = SalesmanSaleFactory(self._sys_config,
self._salesman_spacy)
def __init__(self):
self.sys_config = SystemConfiguration()
self._spacy = SalesmanSpacy(entity_handler=self.sys_config.entity_handler, load_sm=self.sys_config.load_sm)
self._sale_factory = SalesmanSaleFactory(self.sys_config, self._spacy)