def __generate_filter(self, kv):
pending = []
if kv.get('round'):
pending.append(templates.get_round(kv.get('round')))
if kv.get('location'):
pending.append(templates.get_terms('location', kv.get('location')))
if kv.get('domestic', None) is not None:
pending.append(self.__generate_domestic_filter(kv.get('domestic')))
if kv.get('public', None) is not None:
pending.append(self.__generate_public_filter(kv.get('public')))
if kv.get('tag'):
if kv.get('operator', 'or') == 'and':
for t in kv.get('tag', []):
pending.append(
templates.get_term('tags',
t.lower().strip()))
else:
pending.append(
templates.get_terms('tags',
[t.lower() for t in kv.get('tag')]))
if kv.get('investor'):
for iv in kv.get('investor'):
pending.append(templates.get_fuzzy('investor', iv))
if kv.get('investorId'):
pending.append(
templates.get_terms('investorId', kv.get('investorId')))
if kv.get('previous_investor'):
for iv in kv.get('previous_investor'):
pending.append(templates.get_fuzzy('previous_investor', iv))
if kv.get('date'):
pending.append(templates.get_terms('funding_year', kv.get('date')))
if kv.get('funding_date'):
today = datetime.today().date()
for fd in kv.get('funding_date'):
if fd == 'latest7':
start, end = today - timedelta(days=7), today + timedelta(
days=1)
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd == 'latest30':
start, end = today - timedelta(days=30), today + timedelta(
days=1)
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd == 'latest90':
start, end = today - timedelta(days=90), today + timedelta(
days=1)
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd.isalnum():
pending.append(templates.get_term('funding_year', int(fd)))
if kv.get('source'):
pending.append(templates.get_terms('source', kv.get('source')))
return pending
def generate_query(self):
self.__intent_classify()
query = {}
if self.intent == 'tag':
for tag in self.query.get('and', []):
query.setdefault('bool', {}).setdefault('must', []).append(
templates.get_terms('tags', [tag.lower()]))
# query.setdefault('bool', {}).setdefault('must', []).append(templates.get_tag_template(tag))
for tag in self.query.get('or', []):
query.setdefault('bool', {}).setdefault('should', []).extend(
templates.get_fast_tag_template(tag.lower()))
if self.query.get('or'):
query['bool']['minimum_number_should_match'] = 1
for tag in self.query.get('not', []):
query.setdefault('bool', {}).setdefault('must_not', []).append(
templates.get_tag_template(tag.lower()))
if self.intent == 'general':
name = self.query
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_name_template(name))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_fuzzy('name', name, 5))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_fuzzy('alias', name))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_term('members', name))
query.setdefault('bool', {})['minimum_number_should_match'] = 1
if self.filters:
query.setdefault('bool', {}).setdefault('must', []).extend(
self.__generate_filter(self.filters))
if self.nested:
query.setdefault('bool', {}).setdefault('must', []).extend(
self.__generate_filter(self.nested))
return query
def generate_query(self, logger=None):
self.__intent_classify(logger)
# logger.info('End to classify intent')
query = {}
if isinstance(self.input, str) or isinstance(self.input, unicode):
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_term('name', self.input, 10))
if self.intent == 'tag':
for tag in self.query.get('and', []):
query.setdefault('bool', {}).setdefault('must', []).append(
templates.get_terms('tags', [tag.lower()]))
# query.setdefault('bool', {}).setdefault('must', []).append(templates.get_tag_template(tag))
for tag in self.query.get('or', []):
query.setdefault('bool', {}).setdefault('should', []).extend(
templates.get_fast_tag_template(tag.lower()))
if self.query.get('or'):
query['bool']['minimum_number_should_match'] = 1
for tag in self.query.get('not', []):
query.setdefault('bool', {}).setdefault('must_not', []).append(
templates.get_tag_template(tag.lower()))
if self.intent == 'general':
name = self.query
# parsed_names = self.name_parser.segment(name)
# query.setdefault('bool', {}).setdefault('should', []).append(templates.get_term('alias', name, 100))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_name_template(name))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_fuzzy('name', name, 5))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_fuzzy('alias', name))
query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_term('members', name))
# fuzzy name
# if parsed_names:
# for parsed_name in parsed_names.split():
# query.setdefault('bool', {}).setdefault('should', []).append(
# templates.get_term('alias', parsed_name))
query.setdefault('bool', {})['minimum_number_should_match'] = 1
if self.filters:
if logger:
logger.info('Filter', self.filters)
query.setdefault('bool', {}).setdefault('must', []).extend(
self.__generate_filter(self.filters))
return query
def generate_query(self, field, extend=False):
query = dict()
if self.input:
if self.input.strip() in self.news_tags:
query = templates.get_term(
'features', self.news_tags.get(self.input.strip()))
elif extend:
if self.len <= self.tag_max_len:
key = filter(
lambda x: (x not in self.stopwords) and len(x) > 1,
self.seg.cut4search(self.input))
query.setdefault('bool', {}).setdefault('must', []).append(
templates.get_string_template(field, ' '.join(key),
'100%'))
else:
if isinstance(self.input, str) or isinstance(
self.input, unicode):
key = filter(
lambda x: (x not in self.stopwords) and len(x) > 1,
self.seg.cut4search(self.input))
if self.len <= self.tag_max_len:
query.setdefault('bool', {}).setdefault(
'must',
[]).append(templates.get_term(field, self.input))
elif self.len <= 20:
query.setdefault('bool',
{}).setdefault('must', []).append(
templates.get_string_template(
field, ' '.join(key), '100%'))
else:
query.setdefault('bool',
{}).setdefault('must', []).append(
templates.get_string_template(
field, ' '.join(key), '95%'))
if self.filters:
for key in self.filters.keys():
if self.filters.get(key):
query.setdefault('bool', {}).setdefault('must', []).append(
templates.get_terms(key, self.filters.get(key)))
if not extend and (not query):
return {"match_all": {}}
return query
def extend_query(self, es_query):
query = self.__intent_classify()
print 'intent', self.intent
print 'query', query
if self.intent == 'yellow':
es_query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_term('tags', query, boost=5))
return es_query
elif self.intent == 'tag':
# es_query.setdefault('bool', {}).setdefault('should', []).append(templates.get_term('name', ''.join(query),
# boost=50))
es_query.setdefault('bool', {}).setdefault('should', []).append(
templates.get_keyword_template(*query))
# es_query["bool"]["should"].append(templates.get_fuzzy('name', ''.join(query), boost=10))
es_query["bool"]["minimum_number_should_match"] = 1
# for tag in query:
# es_query.setdefault('bool', {}).setdefault('should', []).append(
# templates.get_term('description', tag, 0.5))
# print es_query
return es_query
elif self.intent == 'general':
es_query["bool"]["should"].append(
templates.get_fuzzy('name', query, boost=50))
es_query["bool"]["should"].append(
templates.get_fuzzy('alias', query))
es_query["bool"]["should"].append(
templates.get_term('description', query))
es_query["bool"]["should"].append(
templates.get_term('tags', query, boost=5))
es_query["bool"]["minimum_number_should_match"] = 1
# print es_query
return es_query
elif self.intent == 'empty':
return es_query
def generate_query(self):
query = {}
self.__intent_classify()
if self.intent == 'tag':
query_piece = templates.get_nested_template(
'investor_tag', 'investor_tag.tag', self.query)
query.setdefault('bool', {}).setdefault('must',
[]).append(query_piece)
elif self.intent == 'general':
query.setdefault(
'bool', {})['should'] = templates.get_investor_name_completion(
self.query)
query['bool']['minimum_number_should_match'] = 1
if self.online:
query.setdefault('bool', {}).setdefault('must', []).append(
templates.get_term('online', True))
if self.filters:
query.setdefault('bool', {}).setdefault('must', []).extend(
self.__generate_filter(self.filters))
return query
def __generate_filter(self, kv):
pending = [templates.get_term('oid', self.org)]
if kv.get('status'):
pending.append(templates.get_terms('status', kv.get('status')))
if kv.get('location'):
pending.append(templates.get_terms('location', kv.get('location')))
if kv.get('assignee'):
pending.append(templates.get_terms('assignee', kv.get('assignee')))
if kv.get('sponsor'):
sponsor = [s for s in kv.get('sponsor') if s]
pending.append(templates.get_terms('sponsor', sponsor))
if kv.get('portfolioStatus'):
pending.append(
templates.get_terms('portfolioStatus',
kv.get('portfolioStatus')))
if kv.get('stage'):
pending.append(templates.get_terms('stage', kv.get('stage')))
if kv.get('portfolioStage'):
pending.append(
templates.get_terms('portfolioStage',
kv.get('portfolioStage')))
return pending
def __generate_filter(self, kv):
global collection_ranking_threshold
pending = []
if kv.get('round'):
pending.append(templates.get_terms('round', kv.get('round', [])))
if kv.get('date'):
pending.append(
templates.get_terms('established',
self.__extend_date(kv.get('date'))))
if kv.get('location'):
pending.append(templates.get_terms('location', kv.get('location')))
if kv.get('domestic', None) is not None:
pending.append(self.__generate_domestic_filter(kv.get('domestic')))
if kv.get('team'):
pending.append(templates.get_terms('team', kv.get('team')))
if kv.get('threshold'):
pending.append(
templates.get_range('ranking_score', 1,
collection_ranking_threshold))
if kv.get('yellow'):
pending.append(templates.get_terms('yellows', kv.get('yellow')))
if kv.get('status'):
pending.append(templates.get_terms('status', kv.get('status')))
if kv.get('tag'):
if kv.get('operator', 'and') == 'or':
pending.append(
templates.get_terms('tags',
[t.lower() for t in kv.get('tag')]))
else:
for t in kv.get('tag', []):
pending.append(
templates.get_term('tags',
t.lower().strip()))
if kv.get('category'):
pending.append(templates.get_terms('category', kv.get('category')))
if kv.get('industry'):
pending.append(
templates.get_nested_term('nested_tag.id', 'industry',
kv.get('industry')))
if kv.get('topic'):
pending.append(
templates.get_nested_term('nested_tag.id', 'topic',
kv.get('topic')))
if kv.get('funding_date'):
today = datetime.today().date()
for fd in kv.get('funding_date'):
if fd == 'latest7':
start, end = today - timedelta(days=7), today
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd == 'latest30':
start, end = today - timedelta(days=30), today
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd == 'latest90':
start, end = today - timedelta(days=90), today
pending.append(
templates.get_range('last_funding_date', end, start))
elif fd.isalnum():
start = datetime.strptime('%s-01-01' % fd, '%Y-%m-%d')
end = datetime.strptime('%s-12-31' % fd, '%Y-%m-%d')
pending.append(
templates.get_range('last_funding_date', end, start))
return pending