def create_all_relationship(self):
"""
1.ruling -[have]->ruling_text
:return:
"""
rts = self.base.query(
sql={'metaModel': '裁判文书'},
no_cursor_timeout=True)
i, k = 0, 0
# eg = EtpGraph()
etp_count = rts.count()
relationships = []
# prs = Person()
ruling = Ruling()
for r in rts:
k += 1
rt = RulingText.create_from_original_text(
r['content'], **{'链接': r['url']}
)
rl_n = self.NodeMatcher.match(ruling.label).where(
'_.CASE_NUM="{}"'.format( # OR _.URL="{}"
rt.BaseAttributes['CASE_NUM'],
# rt.BaseAttributes['URL']
)
).first()
if rl_n is None:
continue
relationships.append(
Have(rl_n, rt.get_neo_node(primarykey=rt.primarykey)
).get_relationship()
)
if len(relationships) > 1000:
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, len(relationships)
)))
relationships.clear()
if len(relationships):
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, len(relationships)
)))
relationships.clear()
pass
# rtg = JusRulingTextGraph()
# rtg.create_all_relationship()
def get_all_nodes_and_relationships_from_enterprise(self, etp):
etp_n = Enterprise(URL=etp['url'], NAME=etp['name'])
etp_n = self.get_neo_node(etp_n)
if etp_n is None:
return [], []
nodes, relationships = [], []
nodes.append(etp_n)
if '新闻舆情' in etp['content'].keys():
data = self.get_format_dict(etp['content']['新闻舆情'])
ns = News.create_from_dict(data)
for n in ns:
n_ = n.pop('news')
n_n = self.get_neo_node(n_)
if n_n is not None:
nodes.append(n_n)
relationships.append(
Have(etp_n, n_n, **n)
)
pass
return nodes, relationships
def create_all_relationship(self):
"""
1.enterprise -[have or x]->x
:return:
"""
ops = self.base.query(
sql={'metaModel': '公司新闻'},
# limit=10,
skip=2020,
no_cursor_timeout=True)
i, k = 0, 0
eg = EtpGraph()
etp_count = ops.count()
relationships = []
# etp = Enterprise()
s_t = time.time()
for o in ops:
k += 1
# if k < 43500:
# continue
# TODO(leung): 这里要注意，基本信息以外的模块中的url确定不了公司
etp_n = self.match_node(
*legal,
cypher='_.NAME = "{}"'.format(o['name'])
)
if etp_n is None:
# 如果这个公司还没在数据库里面，那么应该创建这个公司
_ = self.base.query_one(
sql={'metaModel': '基本信息', 'name': o['name']}
)
if _ is not None:
etp = Enterprise(_)
etp_n = self.get_neo_node(etp)
# 虽然在创建司法关系的时候会创建未在库中的企业，但不会创建
# 这个企业的基本关系，因此需要添加其基本关系
relationships += eg.create_relationship_from_enterprise_baseinfo(_)
pass
else:
# 没有这个公司的信息，那就创建一个信息不全的公司
etp = Related(**{'名称': o['name'], '链接': o['url']})
# etp['NAME'] = o['name']
# etp['URL'] = o['url']
etp_n = self.get_neo_node(etp)
if etp_n is None:
continue
pass
if '新闻舆情' in o['content'].keys():
data = self.get_format_dict(o['content']['新闻舆情'])
ns = News.create_from_dict(data)
for n in ns:
n_ = n.pop('news')
n_n = self.get_neo_node(n_)
if n_n is not None:
relationships.append(
Have(etp_n, n_n, **n).get_relationship()
)
pass
if len(relationships) > 1000:
i += 1
sp = int(time.time() - s_t)
s_t = time.time()
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise and spend {} '
'seconds,and merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, sp, len(relationships)
)))
relationships.clear()
# return
if len(relationships):
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, len(relationships)
)))
relationships.clear()
pass
def get_all_nodes_and_relationships_from_enterprise(self, etp):
etp_n = Enterprise(URL=etp['url'], NAME=etp['name'])
etp_n = self.get_neo_node(etp_n)
if etp_n is None:
return [], []
nodes, relationships = [], []
nodes.append(etp_n)
if '产权交易' in etp['content'].keys():
# data = self.get_format_dict(etp['content']['产权交易'])
# for d in data:
# bd = d.pop('标的')
# bd_n =
pass
if '行政许可' in etp['content'].keys():
data = etp['content']['行政许可']
if '工商局' in data.keys():
d1 = self.get_format_dict(data['工商局'])
ls = License.create_from_dict(d1, '工商局')
for l in ls:
l_ = l.pop('license')
l_n = self.get_neo_node(l_)
if l_n is None:
continue
nodes.append(l_n)
relationships.append(Have(etp_n, l_n, **l))
pass
if '信用中国' in data.keys():
d2 = self.get_format_dict(data['信用中国'])
ls = License.create_from_dict(d2, '信用中国')
for l in ls:
l_ = l.pop('license')
l_n = self.get_neo_node(l_)
if l_n is None:
continue
nodes.append(l_n)
relationships.append(Have(etp_n, l_n, **l))
pass
pass
if '招投标信息' in etp['content'].keys():
# 公示的招投标信息一般都是结果，一般情况下是找不到
# 共同投标的单位，除非是共同中标
data = self.get_format_dict(etp['content']['招投标信息'])
bs = Bidding.create_from_dict(data)
for b in bs:
_ = b.pop('bidding')
b_n = self.get_neo_node(_)
if b_n is None:
continue
# TODO(leung):项目分类用作了招投标结果
nodes.append(b_n)
relationships.append(
TakePartIn(etp_n, b_n, **dict(b, **{'RESULT':
b_n['TYPE']})))
pass
if '抽查检查' in etp['content'].keys():
data = self.get_format_dict(etp['content']['抽查检查'])
cs = Check.create_from_dict(data)
for c in cs:
_ = c.pop('check')
n = self.get_neo_node(_)
if n is None:
continue
nodes.append(n)
relationships.append(
Have(etp_n, n, **dict(c, **{'RESULT': n['RESULT']})))
pass
if '双随机抽查' in etp['content'].keys():
data = self.get_format_dict(etp['content']['双随机抽查'])
rcs = RandomCheck.create_from_dict(data)
# rcs_n = self.get_neo_node(rcs)
for rc in rcs:
# TODO(leung):随机抽查没有结果
_ = rc.pop('check')
n = self.get_neo_node(_)
if n is None:
continue
nodes.append(n)
relationships.append(Have(etp_n, n, **rc))
pass
if '税务信用' in etp['content'].keys():
data = self.get_format_dict(etp['content']['税务信用'])
ts = TaxCredit.create_from_dict(data)
# ts_n = self.get_neo_node(ts)
for t in ts:
_ = t.pop('TaxCredit')
n = self.get_neo_node(_)
if n is None:
continue
# TODO(leung):纳税信用等级作为税务信用评级结果
nodes.append(n)
relationships.append(
Have(etp_n, n, **dict(RESULT=n['GRADE'], **t)))
pass
if '进出口信用' in etp['content'].keys():
data = self.get_format_dict(etp['content']['进出口信用'])
ies = IAE.create_from_dict(data)
# ies_n = self.get_neo_node(ies)
for ie in ies:
_ = ie.pop('iae')
n = self.get_neo_node(_)
if n is None:
continue
nodes.append(n)
relationships.append(Have(etp_n, n, **ie))
pass
if '招聘' in etp['content'].keys():
data = self.get_format_dict(etp['content']['招聘'])
rs = Position.create_from_dict(data)
for r in rs:
_ = r.pop('position')
n = self.get_neo_node(_)
if n is None:
continue
nodes.append(n)
relationships.append(Recruit(etp_n, n, **r))
pass
if '客户' in etp['content'].keys():
data = self.get_format_dict(etp['content']['客户'])
cs = Client.create_from_dict(data)
for c in cs:
cli = c.pop('client')
cli_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
cli['URL'], cli['NAME']))
if cli_n is None:
if cli.isEnterprise():
cli = Enterprise(**cli.to_dict(with_label=False))
cli_n = self.get_neo_node(cli)
if cli_n is None:
continue
nodes.append(cli_n)
relationships.append(SellTo(etp_n, cli_n, **c))
pass
if '供应商' in etp['content'].keys():
data = self.get_format_dict(etp['content']['供应商'])
ss = Supplier.create_from_dict(data)
for s in ss:
sup = s.pop('supplier')
sup_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
sup['URL'], sup['NAME']))
if sup_n is None:
if sup.isEnterprise():
sup = Enterprise(**sup.to_dict(with_label=False))
sup_n = self.get_neo_node(sup)
if sup_n is None:
continue
nodes.append(sup_n)
relationships.append(BuyFrom(etp_n, sup_n, **s))
pass
if '信用评级' in etp['content'].keys():
data = self.get_format_dict(etp['content']['信用评级'])
for d in data:
_ = d.pop('评级公司')
_['链接'] = Enterprise.parser_url(_['链接'])
n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
_['链接'], _['名称']))
if n is None:
n = Enterprise(**_)
n = self.get_neo_node(n)
if n is None:
continue
__ = d.pop('内容')
d['评级内容'] = __['内容']
d['评级链接'] = __['链接']
nodes.append(n)
relationships.append(Appraise(n, etp_n, **d))
pass
if '土地转让' in etp['content'].keys():
data = self.get_format_dict(etp['content']['土地转让'])
for d in data:
e1 = d.pop('原土地使用权人')
e2 = d.pop('现有土地使用权人')
p = Plot(**d)
p_n = self.get_neo_node(p)
if p_n is None:
continue
e1['链接'] = Enterprise.parser_url(e1['链接'])
if e1['名称'] == etp['name'] or e1['链接'] == etp['url']:
n1 = etp_n
else:
# 有可能是人
n1 = self.match_node(*legal,
cypher='_.URL = "{}"'.format(
e1['链接']))
if n1 is None:
n1 = Enterprise(**e1)
if not n1.isEnterprise():
n1 = Person(**e1)
if not n1.isPerson():
n1 = Related(**e1)
n1 = self.get_neo_node(n1)
if n1 is not None:
nodes.append(n1)
nodes.append(p_n)
relationships.append(Sell(n1, p_n))
e2['链接'] = Enterprise.parser_url(e2['链接'])
if e2['名称'] == etp['name'] or e2['链接'] == etp['url']:
n2 = etp_n
else:
n2 = self.match_node(*legal,
cypher='_.URL = "{}"'.format(
e2['链接']))
if n2 is None:
n2 = Enterprise(**e2)
if not n2.isEnterprise():
n2 = Person(**e2)
if not n2.isPerson():
n2 = Related(**e2)
n2 = self.get_neo_node(n2)
if n2 is not None:
nodes.append(n2)
nodes.append(p_n)
relationships.append(Buy(n2, p_n))
pass
return nodes, relationships
def create_all_relationship(self):
"""
1.enterprise -[have or x]->x
:return:
"""
ops = self.base.query(
sql={
'metaModel': '经营状况',
# 'name': '重庆轩烽建材有限公司'
},
limit=1000,
# skip=2000,
no_cursor_timeout=True)
i, k = 0, 0
eg = EtpGraph()
etp_count = ops.count()
relationships = []
# etp = Enterprise()
for o in ops:
k += 1
# TODO(leung): 这里要注意，基本信息以外的模块中的url确定不了公司
etp_n = self.match_node(*legal,
cypher='_.NAME = "{}"'.format(o['name']))
if etp_n is None:
# 如果这个公司还没在数据库里面，那么应该创建这个公司
_ = self.base.query_one(sql={
'metaModel': '基本信息',
'name': o['name']
})
if _ is not None:
etp = Enterprise(_)
etp_n = self.get_neo_node(etp)
# 虽然在创建司法关系的时候会创建未在库中的企业，但不会创建
# 这个企业的基本关系，因此需要添加其基本关系
relationships += eg.create_relationship_from_enterprise_baseinfo(
_)
pass
else:
# 没有这个公司的信息，那就创建一个信息不全的公司
# 如果在neo4j里面存着只有name,url的公司，意味着
# 这家公司没有“基本信息”
etp = Related()
etp['NAME'] = o['name']
etp['URL'] = o['url']
etp_n = self.get_neo_node(etp)
pass
if '产权交易' in etp['content'].keys():
# data = self.get_format_dict(etp['content']['产权交易'])
# for d in data:
# bd = d.pop('标的')
# bd_n =
pass
if '行政许可' in etp['content'].keys():
data = etp['content']['行政许可']
if '工商局' in data.keys():
d1 = self.get_format_dict(data['工商局'])
ls = License.create_from_dict(d1, '工商局')
for l in ls:
l_ = l.pop('license')
l_n = self.get_neo_node(l_)
if l_n is None:
continue
relationships.append(
Have(etp_n, l_n, **l).get_relationship())
pass
if '信用中国' in data.keys():
d2 = self.get_format_dict(data['信用中国'])
ls = License.create_from_dict(d2, '信用中国')
for l in ls:
l_ = l.pop('license')
l_n = self.get_neo_node(l_)
if l_n is None:
continue
relationships.append(
Have(etp_n, l_n, **l).get_relationship())
pass
pass
if '招投标信息' in etp['content'].keys():
# 公示的招投标信息一般都是结果，一般情况下是找不到
# 共同投标的单位，除非是共同中标
data = self.get_format_dict(etp['content']['招投标信息'])
bs = Bidding.create_from_dict(data)
for b in bs:
_ = b.pop('bidding')
b_n = self.get_neo_node(_)
if b_n is None:
continue
# TODO(leung):项目分类用作了招投标结果
relationships.append(
TakePartIn(etp_n, b_n,
**dict(b,
**{'RESULT':
b_n['TYPE']})).get_relationship())
pass
if '抽查检查' in etp['content'].keys():
data = self.get_format_dict(etp['content']['抽查检查'])
cs = Check.create_from_dict(data)
for c in cs:
_ = c.pop('check')
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
Have(etp_n, n,
**dict(c, **{'RESULT':
n['RESULT']})).get_relationship())
pass
if '双随机抽查' in etp['content'].keys():
data = self.get_format_dict(etp['content']['双随机抽查'])
rcs = RandomCheck.create_from_dict(data)
# rcs_n = self.get_neo_node(rcs)
for rc in rcs:
# TODO(leung):随机抽查没有结果
_ = rc.pop('check')
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
Have(etp_n, n, **rc).get_relationship())
pass
if '税务信用' in etp['content'].keys():
data = self.get_format_dict(etp['content']['税务信用'])
ts = TaxCredit.create_from_dict(data)
# ts_n = self.get_neo_node(ts)
for t in ts:
_ = t.pop('TaxCredit')
n = self.get_neo_node(_)
if n is None:
continue
# TODO(leung):纳税信用等级作为税务信用评级结果
relationships.append(
Have(etp_n, n, **dict(RESULT=n['GRADE'],
**t)).get_relationship())
pass
if '进出口信用' in etp['content'].keys():
data = self.get_format_dict(etp['content']['进出口信用'])
ies = IAE.create_from_dict(data)
# ies_n = self.get_neo_node(ies)
for ie in ies:
_ = ie.pop('iae')
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
Have(etp_n, n, **ie).get_relationship())
pass
if '招聘' in etp['content'].keys():
data = self.get_format_dict(etp['content']['招聘'])
rs = Position.create_from_dict(data)
for r in rs:
_ = r.pop('position')
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
Recruit(etp_n, n, **r).get_relationship())
pass
if '客户' in etp['content'].keys():
data = self.get_format_dict(etp['content']['客户'])
cs = Client.create_from_dict(data)
for c in cs:
_ = c.pop('client')
n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
_['URL'], _['NAME']))
if n is None:
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
SellTo(etp_n, n, **c).get_relationship())
pass
if '供应商' in etp['content'].keys():
data = self.get_format_dict(etp['content']['供应商'])
ss = Supplier.create_from_dict(data)
for s in ss:
_ = s.pop('supplier')
n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
_['URL'], _['NAME']))
if n is None:
n = self.get_neo_node(_)
if n is None:
continue
relationships.append(
BuyFrom(etp_n, n, **s).get_relationship())
pass
if '信用评级' in etp['content'].keys():
data = self.get_format_dict(etp['content']['信用评级'])
for d in data:
_ = d.pop('评级公司')
n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
_['链接'], _['名称']))
if n is None:
n = Related()
n['NAME'] = _['名称']
n['URL'] = _['链接']
n = self.get_neo_node(n)
if n is None:
continue
__ = d.pop('内容')
d['评级内容'] = __['内容']
d['评级链接'] = __['链接']
relationships.append(
Appraise(n, etp_n, **d).get_relationship())
pass
if '土地转让' in etp['content'].keys():
data = self.get_format_dict(etp['content']['土地转让'])
for d in data:
e1 = d.pop('原土地使用权人')
e2 = d.pop('现有土地使用权人')
p = Plot(**d)
p_n = self.get_neo_node(p)
if p_n is None:
continue
if e1['名称'] == o['name'] or e1['链接'] == o['url']:
n1 = etp_n
else:
# 有可能是人
n1 = self.match_node(*legal,
cypher='_.URL = "{}"'.format(
e1['链接']))
if n1 is None:
n1 = Related(**e1)
n1 = self.get_neo_node(n1)
if n1 is not None:
relationships.append(Sell(n1, p_n).get_relationship())
if e2['名称'] == o['name'] or e2['链接'] == o['url']:
n2 = etp_n
else:
n2 = self.match_node(*legal,
cypher='_.URL = "{}"'.format(
e2['链接']))
if n2 is None:
n2 = Related(**e2)
n2 = self.get_neo_node(n2)
if n2 is not None:
relationships.append(Buy(n2, p_n).get_relationship())
pass
if len(relationships) > 1000:
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(
SuccessMessage(
'{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count,
len(relationships))))
relationships.clear()
# return
if len(relationships):
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(
SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count,
len(relationships))))
relationships.clear()
pass
def get_all_nodes_and_relationships_from_enterprise(self, etp):
"""
创建从公司基本信息可以看出的关系：
1.person-[lr]->enterprise
2.person-[be_in_office]->enterprise
3.enterprise-[located]->address
4.person|enterprise-[holding]->enterprise
5.enterprise-[have]->telephone
6.enterprise-[have]->email
:param :
:return:
"""
# 如果关系上的节点不存在，数据库同样会补充创建节点，这一点很重要
nodes, rps = [], []
etp_n = self.get_neo_node(etp)
if etp_n is None:
self.logger.debug('{} filed initialize enterprise '
'Neo node'.format(etp['NAME']))
return nodes, rps
nodes.append(etp_n)
try:
lr = etp.get_legal_representative()
# 法定代表人有可能会是以下这些对象
lr_n = self.match_node(
*['Person'] + legal,
cypher='_.URL = "{}"'.format(lr['URL'])
)
if lr_n is None:
lr_n = self.get_neo_node(lr)
if lr_n is None:
self.logger.debug('{} filed initialize legal representative '
'Neo node'.format(etp['NAME']))
else:
nodes.append(lr_n)
rps.append(LegalRep(lr_n, etp_n))
except Exception as e:
ExceptionInfo(e)
self.logger.error('{} deal legal representative raise '
'({})'.format(etp['NAME'], e),
exc_info=True)
try:
ms = etp.get_manager()
if len(ms):
for m in ms:
# 主要人员 下面必然是人
m_n = m.pop('person')
m_n = self.get_neo_node(m_n)
if m_n is None:
self.logger.debug('{} filed initialize major manager '
'Neo node'.format(etp['NAME']))
else:
nodes.append(m_n)
rps.append(BeInOffice(m_n, etp_n, **m))
except Exception as e:
self.logger.error('{} deal major managers raise '
'({})'.format(etp['NAME'], e),
exc_info=True)
try:
dz = etp.get_address()
dz_n = self.get_neo_node(dz)
if dz_n is None:
self.logger.debug('{} filed initialize address '
'Neo node'.format(etp['NAME']))
else:
nodes.append(dz_n)
rps.append(Located(etp_n, dz_n))
except Exception as e:
self.logger.error('{} deal address raise '
'({})'.format(etp['NAME'], e),
exc_info=True)
try:
sh = etp.get_share_holder()
if len(sh):
for s in sh:
s_ = s.pop('share_holder')
# 股东有可能会是以下这些对象
sh_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(s_['URL'])
)
if sh_n is None:
sh_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
s_['URL'], s_['NAME'])
)
if sh_n is None: # 在以有的对象里面没找到这个股东
# 创建这个意外的股东
sh_n = self.get_neo_node(s_)
if sh_n is None:
self.logger.debug('{} filed initialize unexpected share '
'holder Neo node'.format(etp['NAME']))
if sh_n is not None:
nodes.append(sh_n)
rps.append(Share(etp_n, sh_n, **s))
except Exception as e:
self.logger.error('{} deal share holder raise '
'({})'.format(etp['NAME'], e),
exc_info=True)
try:
tel = etp.get_telephone_number()
if tel is None:
# self.to_logs('there is not valid telephone for'
# ' this enterprise.', 'ERROR', eb['name'])
pass
else:
tel_n = self.get_neo_node(tel)
if tel_n is None:
self.logger.debug('{} filed initialize telephone '
'Neo node'.format(etp['NAME']))
else:
nodes.append(tel_n)
rps.append(Have(etp_n, tel_n))
pass
except Exception as e:
self.logger.error('{} deal telephone number raise '
'({})'.format(etp['NAME'], e),
exc_info=True)
try:
eml = etp.get_email()
if eml is None:
# self.to_logs('there is not valid email for'
# ' this enterprise.', 'ERROR', eb['name'])
pass
else:
eml_n = self.get_neo_node(eml)
if eml_n is None:
self.logger.debug('{} filed initialize email '
'Neo node'.format(etp['NAME']))
else:
nodes.append(eml_n)
rps.append(Have(etp_n, eml_n))
pass
except Exception as e:
self.logger.debug('{} deal email raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
try:
ivs = etp.get_invest_outer()
if len(ivs):
for iv in ivs:
iv_ = iv.pop('invested')
# 被投资企业可能是下面这些对象
iv_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
iv_['URL'], iv_['NAME'])
)
if iv_n is None:
iv_n = self.get_neo_node(iv_)
if iv_n is None:
self.logger.debug('{} filed initialize unexpected invested '
'Neo node'.format(etp['NAME']))
continue
nodes.append(iv_n)
rps.append(Investing(etp_n, iv_n, **iv))
except Exception as e:
self.logger.error('{} deal invest raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
try:
brs = etp.get_branch()
if len(brs):
for b in brs:
b_ = b.pop('branch')
# 分支机构可能是下面这些对象
b_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
b_['URL'], b_['NAME'])
)
if b_n is None:
b_n = self.get_neo_node(b_)
if b_n is None:
self.logger.debug('{} filed initialize unexpected branch '
'Neo node'.format(etp['NAME']))
continue
p_ = b['principal']
p_n = self.get_neo_node(p_)
if p_n is not None:
nodes.append(p_n)
rps.append(Principal(p_n, b_n))
b.pop('principal')
nodes.append(b_n)
rps.append(BranchAgency(
etp_n, b_n, **b
))
except Exception as e:
self.logger.error('{} deal branch raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
try:
hcs = etp.get_head_company()
if len(hcs):
for h in hcs:
h_ = h.pop('head')
# 总公司可能是下面这些对象
h_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
h_['URL'], h_['NAME'])
)
if h_n is None:
h_n = self.get_neo_node(h_)
if h_n is None:
self.logger.debug('filed initialize unexpected head '
'company Neo node'.format(etp['NAME']))
continue
p_ = h['principal']
p_n = self.get_neo_node(p_)
if p_n is not None:
nodes.append(p_n)
rps.append(Principal(p_n, h_n))
h.pop('principal')
nodes.append(h_n)
rps.append(SuperiorAgency(
etp_n, h_n, **h
))
except Exception as e:
self.logger.error('{} deal head company raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
try:
cps = etp.get_construction_project()
if len(cps):
for c in cps:
c_ = c.pop('project')
c_n = self.get_neo_node(c_)
if c_n is None:
self.logger.debug('filed initialize unexpected construction '
'project Neo node'.format(etp['NAME']))
continue
jsdw = c.pop('jsdw')
# 查询这个建设单位是否已经存在
j_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
jsdw['URL'], jsdw['NAME'])
)
if j_n is None:
j_n = self.get_neo_node(jsdw)
if j_n is None:
self.logger.debug('filed initialize unexpected construction '
'agency Neo node'.format(etp['NAME']))
continue
# TODO(lj):需要考虑是否将承建、建设单独列为一种关系
nodes.append(c_n)
rps.append(Have(
etp_n, c_n, **dict(角色='承建单位', **c)
))
nodes.append(j_n)
rps.append(Have(
j_n, c_n, **dict(角色='建设单位', **c)
))
except Exception as e:
self.logger.error('{} deal construction project raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
try:
ccs = etp.get_construction_certificate()
if len(ccs):
for c in ccs:
c_ = c.pop('ctf')
c_n = self.get_neo_node(c_)
if c_n is None:
self.logger.debug('filed initialize unexpected construction '
'certificate Neo node'.format(etp['NAME']))
continue
nodes.append(c_n)
rps.append(Have(etp_n, c_n, **c))
except Exception as e:
ExceptionInfo(e)
self.logger.error('deal construction certificate raise ({})'
''.format(etp['NAME'], e),
exc_info=True)
return nodes, rps
def create_all_relationship(self):
"""
1.enterprise -[have]->punishment
:return:
"""
ors = self.base.query(
sql={
'metaModel': '经营风险',
# 'name': '重庆铭悦机械设备有限公司'
},
limit=1000,
# skip=2000,
no_cursor_timeout=True)
i, k = 0, 0
eg = EtpGraph()
etp_count = ors.count()
relationships = []
# prs = Person()
etp = Enterprise()
for j in ors:
# 每个公司经营风险下列式的东西，肯定就是这家公司的
k += 1
# if k < 43500:
# continue
# TODO(leung): 这里要注意，基本信息以外的模块中的url确定不了公司
etp_n = self.match_node(
*legal,
cypher='_.NAME = "{}"'.format(j['name'])
)
if etp_n is None:
# 如果这个公司还没在数据库里面，那么应该创建这个公司
_ = self.base.query_one(
sql={'metaModel': '基本信息', 'name': j['name']}
)
if _ is not None:
etp = Enterprise(_)
etp_n = self.get_neo_node(etp)
# 虽然在创建司法关系的时候会创建未在库中的企业，但不会创建
# 这个企业的基本关系，因此需要添加其基本关系
relationships += eg.create_relationship_from_enterprise_baseinfo(_)
pass
else:
# 没有这个公司的信息，那么就简单的把这个公司理解成一个涉案者
# 这里就相当于把一个公司当做了一个风险提示的涉及者
# etp = Related(**{'名称': j['name'], '链接': j['url']})
etp = Related()
etp['NAME'] = j['name']
etp['URL'] = j['url']
etp_n = self.get_neo_node(etp)
pass
if '动产抵押' in j['content'].keys():
data = self.get_format_dict(j['content']['动产抵押'])
for d in data:
_ = d.pop('被担保主债权数额')
debt = Debt(**{'债务(金额)': _['金额'],
'债务(单位)': _['单位'],
'履行期限': d.pop('债务人履行债务的期限')
})
debt_n = self.get_neo_node(debt)
dy = d.pop('抵押权人')
zw = d.pop('债务人')
sy = d.pop('所有权或使用权归属')
if dy['名称'] == j['name'] or dy['链接'] == j['url']:
dy_n = etp_n
else:
dy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dy['链接'], dy['名称'])
)
if dy_n is None and len(dy['名称']) > 1:
dy_n = Related(**dy)
dy_n = self.get_neo_node(dy_n)
if dy_n is not None:
relationships.append(Have(
dy_n, debt_n, **dict(角色='抵押权人', **d)
).get_relationship())
if zw['名称'] == j['name'] or zw['链接'] == j['url']:
zw_n = etp_n
else:
zw_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
zw['链接'], zw['名称'])
)
if zw_n is None and len(zw['名称']) > 1:
zw_n = Related(**zw)
zw_n = self.get_neo_node(zw_n)
if zw_n is not None:
relationships.append(Have(
zw_n, debt_n, **dict(角色='债务人', **d)
).get_relationship())
if sy['名称'] == j['name'] or sy['链接'] == j['url']:
sy_n = etp_n
else:
sy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
sy['链接'], sy['名称'])
)
if sy_n is None and len(sy['名称']) > 1:
sy_n = Related(**sy)
sy_n = self.get_neo_node(sy_n)
if sy_n is not None:
relationships.append(Have(
sy_n, debt_n, **dict(角色='所有权或使用权人', **d)
).get_relationship())
pass
if '公示催告' in j['content'].keys():
data = self.get_format_dict(j['content']['公示催告'])
for d in data:
_ = d.pop('票面金额')
bn = Banknote(**{'票据号': d.pop('票据号'),
'票据类型': d.pop('票据类型'),
'票面金额(金额)': _['金额'],
'票面金额(单位)': _['单位']
})
bn_n = self.get_neo_node(bn)
sq = d.pop('申请人')
cp = d.pop('持票人')
if sq['名称'] == j['name'] or sq['链接'] == j['url']:
sq_n = etp_n
else:
sq_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
sq['链接'], sq['名称'])
)
if sq_n is None:
sq_n = Related(**sq)
sq_n = self.get_neo_node(sq_n)
if sq_n is not None:
relationships.append(Have(
sq_n, bn_n, **dict(角色='申请人', **d)
).get_relationship())
if cp['名称'] == j['name'] or cp['链接'] == j['url']:
cp_n = etp_n
else:
cp_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
cp['链接'], cp['名称'])
)
if cp_n is None:
cp_n = Related(**cp)
cp_n = self.get_neo_node(cp_n)
if cp_n is not None:
relationships.append(Have(
cp_n, bn_n, **dict(角色='持票人', **d)
).get_relationship())
relationships.append(Have(
etp_n, bn_n, **dict(角色='出票人', **d)
).get_relationship())
pass
if '行政处罚' in j['content'].keys():
data = j['content']['行政处罚']
d1 = self.get_format_dict(data['工商局'])
ps = Punishment.create_from_dict(d1, '工商局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
relationships.append(
Have(
etp_n, n, **p
).get_relationship()
)
d2 = self.get_format_dict(data['税务局'])
ps = Punishment.create_from_dict(d2, '税务局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
relationships.append(
Have(
etp_n, n, **p
).get_relationship()
)
d3 = self.get_format_dict(data['信用中国'])
ps = Punishment.create_from_dict(d3, '信用中国')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
relationships.append(
Have(
etp_n, n, **p
).get_relationship()
)
d4 = self.get_format_dict(data['其他'])
ps = Punishment.create_from_dict(d4, '其他')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
relationships.append(
Have(
etp_n, n, **p
).get_relationship()
)
pass
if '环保处罚' in j['content'].keys():
data = self.get_format_dict(j['content']['环保处罚'])
ps = Punishment.create_from_dict(data, '环保局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
relationships.append(
Have(
etp_n, n, **p
).get_relationship()
)
if '股权出质' in j['content'].keys():
sh_info = j['content']['股权出质']
sh_info = self.get_format_dict(sh_info)
for sh in sh_info:
sh = dict(sh, **self.get_format_amount(
'出质数额', sh.pop('出质数额')
))
# 确定出质人
cz = sh.pop('出质人')
cz['链接'] = etp.parser_url(cz['链接'])
# 判断出质人是不是当前公司
if j['name'] == cz['名称'] or cz['链接'] == etp_n['URL']:
cz_n = etp_n
else:
# 确定出质人，先在法人主体中找
cz_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
cz['名称'], cz['链接']
)
)
if cz_n is None:
# 在法人中没找到，就通过url在自然人中找
# 这里最好不要通过名称找了，除公司以外出现
# 同名的几率很大
# TODO(leung):在所有实体中去找开销很大，需要注意
cz_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(cz['链接'])
)
if cz_n is None:
# 创建这个股权出质人
if len(cz['名称']) > 1:
cz_n = Involveder(**cz)
cz_n = self.get_neo_node(cz_n)
pass
# 确定质权人
zq = sh.pop('质权人')
zq['链接'] = etp.parser_url(zq['链接'])
# 判断质权人是不是当前公司
if j['name'] == zq['名称'] or zq['链接'] == etp_n['URL']:
zq_n = etp_n
else:
# 确定质权人，先在企业中找
zq_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
zq['名称'], zq['链接']
)
)
if zq_n is None:
# 在企业中没找到，就通过url在所有对象中找
# 这里最好不要通过名称找了，除公司以外出现
# 同名的几率很大
# TODO(leung):在所有实体中去找开销很大，需要注意
zq_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(zq['链接'])
)
if zq_n is None:
# 创建这个股权出质人
if len(zq['名称']) > 1:
zq_n = Involveder(**zq)
zq_n = self.get_neo_node(zq_n)
pass
# 确定出质标的企业
bd = sh.pop('标的企业')
bd['链接'] = etp.parser_url(bd['链接'])
# 判断出质标的是不是当前公司
if j['name'] == bd['名称'] or bd['链接'] == etp_n['URL']:
bd_n = etp_n
else:
# 确定出质标的，先在企业中找
bd_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
bd['名称'], bd['链接']
)
)
if bd_n is None:
# 在企业中没找到，就通过url在所有对象中找
# 这里最好不要通过名称找了，除公司以外出现
# 同名的几率很大
# TODO(leung):在所有实体中去找开销很大，需要注意
bd_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(bd['链接'])
)
if bd_n is None:
# 创建这个出质标的
if len(bd['名称']) > 1:
bd_n = Possession(**bd)
bd_n = self.get_neo_node(bd_n)
pass
# 创建关系
# 1. 抵押
if cz_n is not None and bd_n is not None:
relationships.append(
Guaranty(cz_n, bd_n, **sh).get_relationship()
)
# 2. 质权
if zq_n is not None and bd_n is not None:
relationships.append(
Have(zq_n, bd_n, **sh).get_relationship()
)
if '破产重组' in j['content'].keys():
data = self.get_format_dict(j['content']['破产重组'])
for d in data:
sq = d.pop('申请人')
if sq['名称'] == j['name'] or sq['链接'] == etp_n['URL']:
sq_n = etp_n
else:
sq_n = self.match_node(
*['person'] + legal,
cypher='_.URL = "{}"'.format(sq['链接'])
)
if sq_n is None:
sq_n = Involveder(**sq)
sq_n = self.get_neo_node(sq_n)
bsq = d.pop('被申请人')
if bsq['名称'] == j['name'] or bsq['链接'] == etp_n['URL']:
bsq_n = etp_n
else:
# 被申请破产的一般是法人
bsq_n = self.match_node(
*['person'] + legal,
cypher='_.URL = "{}"'.format(bsq['链接'])
)
if bsq_n is None:
bsq_n = Involveder(**bsq)
bsq_n = self.get_neo_node(bsq_n)
if sq_n is not None and bsq_n is not None:
relationships.append(
Relationship(sq_n, '申请破产', bsq_n, **d)
)
pass
if '土地抵押' in j['content'].keys():
data = self.get_format_dict(j['content']['土地抵押'])
for d in data:
_ = d.pop('抵押面积')
p = Plot(**{'位置': d.pop('位置'),
'面积(数量)': _['数额'],
'面积(单位)': _['单位'],
})
p_n = self.get_neo_node(p)
d = dict(d, **self.get_format_amount(
'抵押金额', d.pop('抵押金额')
))
dy = d.pop('抵押人')
dyq = d.pop('抵押权人')
if dy['名称'] == j['name'] or dy['链接'] == etp_n['URL']:
dy_n = etp_n
else:
dy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dy['链接'], dy['名称'])
)
if dy_n is None:
dy_n = Related(**dy)
dy_n = self.get_neo_node(dy_n)
if dy_n is not None:
relationships.append(
Guaranty(dy_n, p_n, **d).get_relationship()
)
if dyq['名称'] == j['name'] or dyq['链接'] == etp_n['URL']:
dyq_n = etp_n
else:
dyq_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dyq['链接'], dyq['名称'])
)
if dyq_n is None:
dyq_n = Related(**dyq)
dyq_n = self.get_neo_node(dyq_n)
if dyq_n is not None:
relationships.append(
Have(dyq_n, p_n, **d).get_relationship()
)
pass
if len(relationships) > 1000:
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, len(relationships)
)))
relationships.clear()
# return
pass
if len(relationships):
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, len(relationships)
)))
relationships.clear()
pass
def get_all_nodes_and_relationships_from_enterprise(self, etp):
etp_n = Enterprise(URL=etp['url'], NAME=etp['name'])
etp_n = self.get_neo_node(etp_n)
if etp_n is None:
return [], []
nodes, relationships = [], []
nodes.append(etp_n)
if '动产抵押' in etp['content'].keys():
data = self.get_format_dict(etp['content']['动产抵押'])
for d in data:
_ = d.pop('被担保主债权数额')
debt = Debt(**{'债务(金额)': _['金额'],
'债务(单位)': _['单位'],
'履行期限': d.pop('债务人履行债务的期限')
})
debt_n = self.get_neo_node(debt)
nodes.append(debt_n)
dy = d.pop('抵押权人')
zw = d.pop('债务人')
sy = d.pop('所有权或使用权归属')
dy['链接'] = Enterprise.parser_url(dy['链接'])
zw['链接'] = Enterprise.parser_url(zw['链接'])
sy['链接'] = Enterprise.parser_url(sy['链接'])
if dy['名称'] == etp['name'] or dy['链接'] == etp['url']:
dy_n = etp_n
else:
dy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dy['链接'], dy['名称'])
)
if dy_n is None:
# dy_n = Related(**dy)
dy_n = Enterprise(**dy)
if not dy_n.isEnterprise():
dy_n = Person(**dy)
if not dy_n.isPerson():
dy_n = Related(**dy)
dy_n = self.get_neo_node(dy_n)
if dy_n is not None:
nodes.append(dy_n)
relationships.append(Have(
dy_n, debt_n, **dict(角色='抵押权人', **d)
))
if zw['名称'] == etp['name'] or zw['链接'] == etp['url']:
zw_n = etp_n
else:
zw_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
zw['链接'], zw['名称'])
)
if zw_n is None and len(zw['名称']) > 1:
# zw_n = Related(**zw)
zw_n = Enterprise(**zw)
if not zw_n.isEnterprise():
zw_n = Person(**zw)
if not zw_n.isPerson():
zw_n = Related(**zw)
zw_n = self.get_neo_node(zw_n)
if zw_n is not None:
nodes.append(zw_n)
relationships.append(Have(
zw_n, debt_n, **dict(角色='债务人', **d)
))
if sy['名称'] == etp['name'] or sy['链接'] == etp['url']:
sy_n = etp_n
else:
sy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
sy['链接'], sy['名称'])
)
if sy_n is None and len(sy['名称']) > 1:
# sy_n = Related(**sy)
sy_n = Enterprise(**sy)
if not sy_n.isEnterprise():
sy_n = Person(**sy)
if not sy_n.isPerson():
sy_n = Related(**sy)
sy_n = self.get_neo_node(sy_n)
if sy_n is not None:
nodes.append(sy_n)
relationships.append(Have(
sy_n, debt_n, **dict(角色='所有权或使用权人', **d)
))
pass
if '公示催告' in etp['content'].keys():
data = self.get_format_dict(etp['content']['公示催告'])
for d in data:
_ = d.pop('票面金额')
bn = Banknote(**{'票据号': d.pop('票据号'),
'票据类型': d.pop('票据类型'),
'票面金额(金额)': _['金额'],
'票面金额(单位)': _['单位']
})
bn_n = self.get_neo_node(bn)
nodes.append(bn_n)
sq = d.pop('申请人')
cp = d.pop('持票人')
sq['链接'] = Enterprise.parser_url(sq['链接'])
cp['链接'] = Enterprise.parser_url(cp['链接'])
if sq['名称'] == etp['name'] or sq['链接'] == etp['url']:
sq_n = etp_n
else:
sq_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
sq['链接'], sq['名称'])
)
if sq_n is None:
# sq_n = Related(**sq)
sq_n = Enterprise(**sq)
if not sq_n.isEnterprise():
sq_n = Person(**sq)
if not sq_n.isPerson():
sq_n = Related(**sq)
sq_n = self.get_neo_node(sq_n)
if sq_n is not None:
nodes.append(sq_n)
relationships.append(Have(
sq_n, bn_n, **dict(角色='申请人', **d)
))
if cp['名称'] == etp['name'] or cp['链接'] == etp['url']:
cp_n = etp_n
else:
cp_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
cp['链接'], cp['名称'])
)
if cp_n is None:
# cp_n = Related(**cp)
cp_n = Enterprise(**cp)
if not cp_n.isEnterprise():
cp_n = Person(**cp)
if not cp_n.isPerson():
cp_n = Related(**cp)
cp_n = self.get_neo_node(cp_n)
if cp_n is not None:
nodes.append(cp_n)
relationships.append(Have(
cp_n, bn_n, **dict(角色='持票人', **d)
))
relationships.append(Have(
etp_n, bn_n, **dict(角色='出票人', **d)
))
pass
if '行政处罚' in etp['content'].keys():
data = etp['content']['行政处罚']
d1 = self.get_format_dict(data['工商局'])
ps = Punishment.create_from_dict(d1, '工商局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
nodes.append(n)
relationships.append(
Have(etp_n, n, **p)
)
d2 = self.get_format_dict(data['税务局'])
ps = Punishment.create_from_dict(d2, '税务局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
nodes.append(n)
relationships.append(
Have(etp_n, n, **p)
)
d3 = self.get_format_dict(data['信用中国'])
ps = Punishment.create_from_dict(d3, '信用中国')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
nodes.append(n)
relationships.append(
Have(etp_n, n, **p)
)
d4 = self.get_format_dict(data['其他'])
ps = Punishment.create_from_dict(d4, '其他')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
nodes.append(n)
relationships.append(
Have(etp_n, n, **p)
)
pass
if '环保处罚' in etp['content'].keys():
data = self.get_format_dict(etp['content']['环保处罚'])
ps = Punishment.create_from_dict(data, '环保局')
for p in ps:
_ = p.pop('punishment')
n = self.get_neo_node(_)
if n is not None:
nodes.append(n)
relationships.append(
Have(etp_n, n, **p)
)
if '股权出质' in etp['content'].keys():
sh_info = etp['content']['股权出质']
sh_info = self.get_format_dict(sh_info)
for sh in sh_info:
sh = dict(sh, **self.get_format_amount(
'出质数额', sh.pop('出质数额')
))
# 确定出质人
cz = sh.pop('出质人')
cz['链接'] = Enterprise.parser_url(cz['链接'])
# 判断出质人是不是当前公司
if etp['name'] == cz['名称'] or cz['链接'] == etp_n['URL']:
cz_n = etp_n
else:
# 确定出质人，先在法人主体中找
cz_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
cz['名称'], cz['链接']
)
)
if cz_n is None:
# 在法人中没找到，就通过url在自然人中找
# 这里最好不要通过名称找了，除公司以外出现
# 同名的几率很大
# TODO(leung):在所有实体中去找开销很大，需要注意
cz_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(cz['链接'])
)
if cz_n is None:
# 创建这个股权出质人
if len(cz['名称']) > 1:
# cz_n = Involveder(**cz)
cz_n = Enterprise(**cz)
if not cz_n.isEnterprise():
cz_n = Person(**cz)
if not cz_n.isPerson():
cz_n = Related(**cz)
cz_n = self.get_neo_node(cz_n)
pass
# 确定质权人
zq = sh.pop('质权人')
zq['链接'] = Enterprise.parser_url(zq['链接'])
# 判断质权人是不是当前公司
if etp['name'] == zq['名称'] or zq['链接'] == etp_n['URL']:
zq_n = etp_n
else:
# 确定质权人，先在企业中找
zq_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
zq['名称'], zq['链接']
)
)
if zq_n is None:
# 在企业中没找到，就通过url在所有对象中找
# 这里最好不要通过名称找了，除公司以外出现
# 同名的几率很大
# TODO(leung):在所有实体中去找开销很大，需要注意
zq_n = self.match_node(
'Person',
cypher='_.URL = "{}"'.format(zq['链接'])
)
if zq_n is None:
# 创建这个股权出质人
if len(zq['名称']) > 1:
# zq_n = Involveder(**zq)
zq_n = Enterprise(**zq)
if not zq_n.isEnterprise():
zq_n = Person(**zq)
if not zq_n.isPerson():
zq_n = Related(**zq)
zq_n = self.get_neo_node(zq_n)
pass
# 确定出质标的企业
bd = sh.pop('标的企业')
bd['链接'] = Enterprise.parser_url(bd['链接'])
# 判断出质标的是不是当前公司
if etp['name'] == bd['名称'] or bd['链接'] == etp_n['URL']:
bd_n = etp_n
else:
# 确定出质标的，先在企业中找，不会是人
bd_n = self.match_node(
*legal,
cypher='_.NAME = "{}" OR _.URL = "{}"'.format(
bd['名称'], bd['链接']
)
)
if bd_n is None:
# 创建这个出质标的
if len(bd['名称']) > 1:
bd_n = Enterprise(**bd)
if not bd_n.isEnterprise():
bd_n = Possession(**bd)
bd_n = self.get_neo_node(bd_n)
pass
# 创建关系
if bd_n is None:
continue
nodes.append(bd_n)
# 1. 抵押
if cz_n is not None:
nodes.append(cz_n)
relationships.append(
Guaranty(cz_n, bd_n, **sh)
)
# 2. 质权
if zq_n is not None:
nodes.append(zq_n)
relationships.append(
Have(zq_n, bd_n, **sh)
)
if '破产重组' in etp['content'].keys():
data = self.get_format_dict(etp['content']['破产重组'])
for d in data:
sq = d.pop('申请人')
sq['链接'] = Enterprise.parser_url(sq['链接'])
if sq['名称'] == etp['name'] or sq['链接'] == etp_n['URL']:
sq_n = etp_n
else:
sq_n = self.match_node(
*['person'] + legal,
cypher='_.URL = "{}"'.format(sq['链接'])
)
if sq_n is None:
# sq_n = Involveder(**sq)
sq_n = Enterprise(**sq)
if not sq_n.isEnterprise():
sq_n = Person(**sq)
if not sq_n.isPerson():
sq_n = Related(**sq)
sq_n = self.get_neo_node(sq_n)
bsq = d.pop('被申请人')
bsq['链接'] = Enterprise.parser_url(bsq['链接'])
if bsq['名称'] == etp['name'] or bsq['链接'] == etp_n['URL']:
bsq_n = etp_n
else:
# 被申请破产的一般是法人
bsq_n = self.match_node(
*['person'] + legal,
cypher='_.URL = "{}"'.format(bsq['链接'])
)
if bsq_n is None:
# bsq_n = Involveder(**bsq)
bsq_n = Enterprise(**bsq)
if not bsq_n.isEnterprise():
bsq_n = Person(**bsq)
if not bsq_n.isPerson():
bsq_n = Related(**bsq)
bsq_n = self.get_neo_node(bsq_n)
if sq_n is not None and bsq_n is not None:
nodes += [sq_n, bsq_n]
relationships.append(
ApplyBankrupt(sq_n, bsq_n, **d)
)
pass
if '土地抵押' in etp['content'].keys():
data = self.get_format_dict(etp['content']['土地抵押'])
for d in data:
_ = d.pop('抵押面积')
p = Plot(**{'位置': d.pop('位置'),
'面积(数量)': _['数额'],
'面积(单位)': _['单位'],
})
p_n = self.get_neo_node(p)
nodes.append(p_n)
d = dict(d, **self.get_format_amount(
'抵押金额', d.pop('抵押金额')
))
dy = d.pop('抵押人')
dyq = d.pop('抵押权人')
dy['链接'] = Enterprise.parser_url(dy['链接'])
dyq['链接'] = Enterprise.parser_url(dyq['链接'])
if dy['名称'] == etp['name'] or dy['链接'] == etp_n['URL']:
dy_n = etp_n
else:
dy_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dy['链接'], dy['名称'])
)
if dy_n is None:
# dy_n = Related(**dy)
dy_n = Enterprise(**dy)
if not dy_n.isEnterprise():
dy_n = Person(**dy)
if not dy_n.isPerson():
dy_n = Related(**dy)
dy_n = self.get_neo_node(dy_n)
if dy_n is not None:
nodes.append(dy_n)
relationships.append(
Guaranty(dy_n, p_n, **d)
)
if dyq['名称'] == etp['name'] or dyq['链接'] == etp_n['URL']:
dyq_n = etp_n
else:
dyq_n = self.match_node(
*legal,
cypher='_.URL = "{}" OR _.NAME = "{}"'.format(
dyq['链接'], dyq['名称'])
)
if dyq_n is None:
# dyq_n = Related(**dyq)
dyq_n = Enterprise(**dyq)
if not dyq_n.isEnterprise():
dyq_n = Person(**dyq)
if not dyq_n.isPerson():
dyq_n = Related(**dyq)
dyq_n = self.get_neo_node(dyq_n)
if dyq_n is not None:
nodes.append(dyq_n)
relationships.append(
Have(dyq_n, p_n, **d)
)
pass
return nodes, relationships
def create_all_relationship(self):
"""
1.enterprise -[have]->x
:return:
"""
rts = self.base.query(
sql={'metaModel': '知识产权'},
# limit=100,
skip=79175 + 7909,
no_cursor_timeout=True)
i, k = 0, 0
eg = EtpGraph()
# etp = Enterprise()
etp_count = rts.count()
relationships = []
s_t = time.time()
for r in rts:
k += 1
# TODO(leung): 这里要注意，基本信息以外的模块中的url确定不了公司
etp_n = self.match_node(*legal,
cypher='_.NAME = "{}"'.format(r['name']))
if etp_n is None:
# 如果这个公司还没在数据库里面，那么应该创建这个公司
_ = self.base.query_one(sql={
'metaModel': '基本信息',
'name': r['name']
})
if _ is not None:
etp = Enterprise(_)
etp_n = self.get_neo_node(etp)
# 虽然在创建司法关系的时候会创建未在库中的企业，但不会创建
# 这个企业的基本关系，因此需要添加其基本关系
relationships += eg.create_relationship_from_enterprise_baseinfo(
_)
pass
else:
# 没有这个公司的信息，那就创建一个信息不全的公司
etp = Related(**{'名称': r['name'], '链接': r['url']})
# etp['NAME'] = r['name']
# etp['URL'] = r['url']
etp_n = self.get_neo_node(etp)
pass
pass
if '网站信息' in r['content'].keys():
data = self.get_format_dict(r['content']['网站信息'])
webs = Website.create_from_dict(data)
for web in webs:
w = web.pop('website')
w_n = self.get_neo_node(w)
if w_n is not None:
relationships.append(
Have(etp_n, w_n, **web).get_relationship())
pass
if '证书信息' in r['content'].keys():
data = self.get_format_dict(r['content']['证书信息'])
ctfs = Certificate.create_from_dict(data)
for ctf in ctfs:
c = ctf.pop('certificate')
c_n = self.get_neo_node(c)
if c_n is not None:
relationships.append(
Have(etp_n, c_n, **ctf).get_relationship())
pass
if '专利信息' in r['content'].keys():
data = self.get_format_dict(r['content']['专利信息'])
pats = Patent.create_from_dict(data)
for pat in pats:
p = pat.pop('patent')
p_n = self.get_neo_node(p)
if p_n is not None:
relationships.append(
Have(etp_n, p_n, **pat).get_relationship())
pass
if '商标信息' in r['content'].keys():
data = self.get_format_dict(r['content']['商标信息'])
tms = Trademark.create_from_dict(data)
for tm in tms:
t = tm.pop('trademark')
t_n = self.get_neo_node(t)
if t_n is not None:
relationships.append(
Have(etp_n, t_n, **tm).get_relationship())
pass
if '软件著作权' in r['content'].keys():
data = self.get_format_dict(r['content']['软件著作权'])
scrs = SoftCopyRight.create_from_dict(data)
for scr in scrs:
s = scr.pop('softcopyright')
s_n = self.get_neo_node(s)
if s_n is not None:
relationships.append(
Have(etp_n, s_n, **scr).get_relationship())
pass
if '作品著作权' in r['content'].keys():
data = self.get_format_dict(r['content']['作品著作权'])
wcrs = WorkCopyRight.create_from_dict(data)
for wcr in wcrs:
w = wcr.pop('workcopyright')
w_n = self.get_neo_node(w)
if w_n is not None:
relationships.append(
Have(etp_n, w_n, **wcr).get_relationship())
pass
if '微博' in r['content'].keys():
data = self.get_format_dict(r['content']['微博'])
wbs = Weibo.create_from_dict(data)
for wb in wbs:
w = wb.pop('weibo')
w_n = self.get_neo_node(w)
if w_n is not None:
relationships.append(
Have(etp_n, w_n, **wb).get_relationship())
pass
if '微信公众号' in r['content'].keys():
data = self.get_format_dict(r['content']['微信公众号'])
oas = OfficialAccount.create_from_dict(data)
for oa in oas:
woa = oa.pop('WeChat')
woa_n = self.get_neo_node(woa)
if woa_n is not None:
relationships.append(
Have(etp_n, woa_n, **oa).get_relationship())
pass
if '小程序' in r['content'].keys():
data = self.get_format_dict(r['content']['小程序'])
alts = Applets.create_from_dict(data)
for alt in alts:
a = alt.pop('applets')
a_n = self.get_neo_node(a)
if a_n is not None:
relationships.append(
Have(etp_n, a_n, **alt).get_relationship())
pass
if 'APP' in r['content'].keys():
data = self.get_format_dict(r['content']['APP'])
aps = App.create_from_dict(data)
for ap in aps:
a = ap.pop('app')
a_n = self.get_neo_node(a)
if a_n is not None:
relationships.append(
Have(etp_n, a_n, **ap).get_relationship())
pass
if len(relationships) > 1000:
i += 1
sp = int(time.time() - s_t)
s_t = time.time()
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(
SuccessMessage(
'{}:success merge relationships to database '
'round {} and deal {}/{} enterprise and spend {} '
'seconds,and merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count, sp,
len(relationships))))
relationships.clear()
# return
if len(relationships):
i += 1
self.graph_merge_relationships(relationships)
if not self.index_and_constraint_statue:
self.create_index_and_constraint()
print(
SuccessMessage('{}:success merge relationships to database '
'round {} and deal {}/{} enterprise,and'
' merge {} relationships.'.format(
dt.datetime.now(), i, k, etp_count,
len(relationships))))
relationships.clear()
pass
def get_all_nodes_and_relationships_from_enterprise(self, etp):
etp_n = Enterprise(URL=etp['url'], NAME=etp['name'])
etp_n = self.get_neo_node(etp_n)
if etp_n is None:
return [], []
nodes, relationships = [], []
nodes.append(etp_n)
if '网站信息' in etp['content'].keys():
data = self.get_format_dict(etp['content']['网站信息'])
webs = Website.create_from_dict(data)
for web in webs:
w = web.pop('website')
w_n = self.get_neo_node(w)
if w_n is not None:
nodes.append(w_n)
relationships.append(Have(etp_n, w_n, **web))
pass
if '证书信息' in etp['content'].keys():
data = self.get_format_dict(etp['content']['证书信息'])
ctfs = Certificate.create_from_dict(data)
for ctf in ctfs:
c = ctf.pop('certificate')
c_n = self.get_neo_node(c)
if c_n is not None:
nodes.append(c_n)
relationships.append(Have(etp_n, c_n, **ctf))
pass
if '专利信息' in etp['content'].keys():
data = self.get_format_dict(etp['content']['专利信息'])
pats = Patent.create_from_dict(data)
for pat in pats:
p = pat.pop('patent')
p_n = self.get_neo_node(p)
if p_n is not None:
nodes.append(p_n)
relationships.append(Have(etp_n, p_n, **pat))
pass
if '商标信息' in etp['content'].keys():
data = self.get_format_dict(etp['content']['商标信息'])
tms = Trademark.create_from_dict(data)
for tm in tms:
t = tm.pop('trademark')
t_n = self.get_neo_node(t)
if t_n is not None:
nodes.append(t_n)
relationships.append(Have(etp_n, t_n, **tm))
pass
if '软件著作权' in etp['content'].keys():
data = self.get_format_dict(etp['content']['软件著作权'])
scrs = SoftCopyRight.create_from_dict(data)
for scr in scrs:
s = scr.pop('softcopyright')
s_n = self.get_neo_node(s)
if s_n is not None:
nodes.append(s_n)
relationships.append(Have(etp_n, s_n, **scr))
pass
if '作品著作权' in etp['content'].keys():
data = self.get_format_dict(etp['content']['作品著作权'])
wcrs = WorkCopyRight.create_from_dict(data)
for wcr in wcrs:
w = wcr.pop('workcopyright')
w_n = self.get_neo_node(w)
if w_n is not None:
nodes.append(w_n)
relationships.append(Have(etp_n, w_n, **wcr))
pass
if '微博' in etp['content'].keys():
data = self.get_format_dict(etp['content']['微博'])
wbs = Weibo.create_from_dict(data)
for wb in wbs:
w = wb.pop('weibo')
w_n = self.get_neo_node(w)
if w_n is not None:
nodes.append(w_n)
relationships.append(Have(etp_n, w_n, **wb))
pass
if '微信公众号' in etp['content'].keys():
data = self.get_format_dict(etp['content']['微信公众号'])
oas = OfficialAccount.create_from_dict(data)
for oa in oas:
woa = oa.pop('WeChat')
woa_n = self.get_neo_node(woa)
if woa_n is not None:
nodes.append(woa_n)
relationships.append(Have(etp_n, woa_n, **oa))
pass
if '小程序' in etp['content'].keys():
data = self.get_format_dict(etp['content']['小程序'])
alts = Applets.create_from_dict(data)
for alt in alts:
a = alt.pop('applets')
a_n = self.get_neo_node(a)
if a_n is not None:
nodes.append(a_n)
relationships.append(Have(etp_n, a_n, **alt))
pass
if 'APP' in etp['content'].keys():
data = self.get_format_dict(etp['content']['APP'])
aps = App.create_from_dict(data)
for ap in aps:
a = ap.pop('app')
a_n = self.get_neo_node(a)
if a_n is not None:
nodes.append(a_n)
relationships.append(Have(etp_n, a_n, **ap))
pass
return nodes, relationships