def __init__(self):
self.id2addr = {}
self.addr_id_set = set()
self.addr_array = []
self.place2xy = self.xyParser()
self.place2xy_set = set(self.place2xy.keys())
# 加载CBDB上的所有地址
addr_data = graph.run('MATCH (n:Addr_codes) RETURN id(n), n').data()
for data in addr_data:
addr_node = data['n']
self.createAddr(addr_node)
addr_belong_data = graph.run(
'MATCH (n1:Addr_codes)-->(:Addr_belongs_data)-->(n2:Addr_codes) RETURN n1.c_addr_id as son_id, n2.c_addr_id as parent_id'
).data()
for data in addr_belong_data:
son_id = data['son_id']
parent_id = data['parent_id']
son = self.getAddr(son_id)
parent = self.getAddr(parent_id)
parent.addSon(son)
son.addParent(parent)
for addr in self.addr_array:
if addr.x != 'None' and addr.x is not None: #如果原先已有纪录
xy = [int(value) / 1000000 for value in [addr.x, addr.y]]
# for value in [addr.x, addr.y]:
# if len(value)<9:
# for time in range(9-len(value)):
# value += '0'
# value = int(value)/1000000
# xy.append(value)
if xy[0] > 0:
while xy[0] > 180:
xy[0] /= 10
while xy[0] < 6:
xy[0] *= 10
if xy[1] > 0:
while xy[1] > 90:
xy[1] /= 10
while xy[1] < 6:
xy[1] *= 10
# print( [addr.x, addr.y], xy)
else:
# xy = [-1,-1]
xy = self._getXY(addr.name)
# print(addr.name,xy)
addr.x = xy[0]
addr.y = xy[1]
# 加载所有的地点关系的树形结构
print('加载地址管理器')
def loadTextEvents(self, LIMIT=1000000, SKIP=0, person_id=None):
print('开始加载文学事件')
if person_id is None:
query = 'MATCH (person:Biog_main)--(event:Text_data)--(text:Text_codes) RETURN person,event,id(event),text SKIP {} LIMIT {} '.format(
str(SKIP), str(LIMIT))
else:
query = 'MATCH (person:Biog_main{{c_personid:"{}"}})--(event:Text_data)--(text:Text_codes) RETURN person,event,id(event),text SKIP {} LIMIT {} '.format(
str(person_id), str(SKIP), str(LIMIT))
results = graph.run(query).data()
id_list = []
id2person = {}
for node_data in results:
event_id = str(node_data['id(event)'])
person = node_data['person']
event_node = node_data['event']
text = node_data['text'] #也有时间，未用
event = self.createEvents(event_id)
event.type = '文学事件'
field = 'c_year'
year = event_node[field]
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('text_data', field)
year_range = event_node[range_field]
event.addTimeAndRange(year, year_range)
id_list.append(event_id)
person = personManager.createPerson(person)
id2person[event_id] = person
if len(id_list) == 0:
return False
id_list_str = ','.join(id_list)
# print(id_list_str)
# 加载角色
results = graph.run(
'START event=node({}) MATCH (role:Text_role_codes)--(event) RETURN role.c_role_desc_chn AS role, id(event)'
.format(id_list_str))
results = results.data()
# print(results)
for result in results:
role = result['role']
event_id = str(result['id(event)'])
self.createEvents(event_id).addPerson(person, role)
event.setTrigger('文学作品' + role)
print('加载文学事件角色')
if len(id_list) < LIMIT:
return False
return True
def getPerson(self, person_id):
person_id = str(person_id)
if person_id in self.id_set:
return self.id2person[person_id]
else:
print('run')
data = graph.run(
'MATCH (n:Biog_main{{c_personid:"{}"}}) RETURN n'.format(
str(person_id))).data()
if len(data) != 0:
# print(data[0])
return self.createPerson(data[0]['n'])
return None
def __init__(self):
self.nian_hao = {}
data = graph.run('MATCH (n:Nian_hao) RETURN n').data()
for nian_hao in data:
nian_hao = nian_hao['n']
# print(nian_hao)
name = nian_hao['c_nianhao_chn']
id = nian_hao['c_nianhao_id']
start_year = nian_hao['c_firstyear']
end_year = nian_hao['c_lastyear']
range = [-9999, 9999]
def isYear(year):
return year is not None and year != 'None' and re.match(
'[-]*[0-9]+', str(year))
if isYear(start_year):
range[0] = int(start_year)
if isYear(end_year):
range[1] = int(end_year)
self.nian_hao[id] = {'name': name, 'id': id, 'time_range': range}
def loadRelationEvents(self, LIMIT=1000000, SKIP=0, person_id=None):
print('开始加载关系事件')
# 使用neo4j提取
# 获取事件
if person_id is None:
query = 'MATCH (person:Biog_main)--(n:Assoc_data) RETURN n, id(n) SKIP {} LIMIT {} '.format(
str(SKIP), str(LIMIT))
else:
query = 'MATCH (person:Biog_main{{c_personid:"{}"}})--(n:Assoc_data) RETURN n, id(n) SKIP {} LIMIT {} '.format(
str(person_id), str(SKIP), str(LIMIT))
results = graph.run(query).data()
csv_data = []
for node_data in results:
node_id = node_data['id(n)']
node_data = node_data['n']
# node_object_id = node_data['OBJECT_ID']
# 添加时间
# print(node_data['c_assoc_year']==None)
event = self.createEvents(node_id)
event.type = '关系事件'
if 'c_sequence' in node_data.keys():
sequence = node_data['c_sequence']
if sequence != 'None' and sequence is not None:
event.sequence = int(sequence)
c_assoc_year = node_data['c_assoc_year']
if c_assoc_year is not None and c_assoc_year != 0 and c_assoc_year != '0' and c_assoc_year != 'None':
c_assoc_year = int(c_assoc_year)
range_field = graph.getTableRange('assoc_data', 'c_assoc_year')
year_range = node_data[range_field]
event.addTimeAndRange(c_assoc_year, year_range)
# if 'c_assoc_nh_code' in node_data.keys():
# nh_code = node_data['c_assoc_nh_code']
# if nh_code != 'None' and nh_code is not None:
# nh_range = timeManager.getNianHaoRange(nh_code)
# event.addTimeAndRange(nh_range[0], '之后')
# event.addTimeAndRange(nh_range[1], '之前')
row = [node_id]
csv_data.append(row)
# print(csv_data)
if len(csv_data) == 0:
return False
id_list_str = ','.join([str(row[0]) for row in csv_data])
# 获得角色
# START n=node(282787) MATCH (person:Biog_main)-[r]-(n) RETURN id(person), r, n, person
results = graph.run(
'START n=node({}) MATCH (person:Biog_main)-[r]-(n) RETURN r, person, id(n)'
.format(id_list_str))
results = results.data()
for result in results:
person = personManager.createPerson(result['person'])
event_id = result['id(n)']
role = result['r']['RELATION_TYPE']
if role == '关系':
role = '主角'
self.createEvents(event_id).addPerson(person, role)
print('加载关系事件角色')
# 加载地点
results = graph.run(
'START n=node({}) MATCH (addr:Addr_codes)-[r]-(n) RETURN addr.c_addr_id as addr_id, id(n)'
.format(id_list_str))
results = results.data()
for result in results:
addr_id = result['addr_id']
event_id = result['id(n)']
self.createEvents(event_id).setAddr(addrManager.getAddr(addr_id))
print('加载关系数据地址')
# 加载触发词
results = graph.run(
'START n=node({}) MATCH (assoc:Assoc_codes)<-[r]-(n) RETURN assoc.c_assoc_desc_chn, id(n)'
.format(id_list_str))
results = results.data()
for result in results:
trigger_name = result['assoc.c_assoc_desc_chn']
event_id = result['id(n)']
self.createEvents(event_id).setTrigger(trigger_name)
# print(self.createEvents(event_id).trigger)
print('加载事件触发词')
# graph.runWithCsv(csv_data, 'START n=node(toInt(row[0])) MATCH (person:Biog_main)-[r]-(n) RETURN id(person), r, person, row[0]')
# print()
if len(csv_data) < LIMIT:
return False
return True
def loadEntryEvents(self, LIMIT=1000000, SKIP=0, person_id=None):
print('开始加载入仕数据')
if person_id is None:
query = 'MATCH (person:Biog_main)-[:参与人]->(event:Entry_data) RETURN person,event,id(event) SKIP {} LIMIT {} '.format(
str(SKIP), str(LIMIT))
else:
query = 'MATCH (person:Biog_main{{ c_personid:"{}" }})-[:参与人]->(event:Entry_data) RETURN person,event,id(event) SKIP {} LIMIT {} '.format(
str(person_id), str(SKIP), str(LIMIT))
results = graph.run(query).data()
id_list = []
for node_data in results:
event_id = str(node_data['id(event)'])
person = node_data['person']
event_node = node_data['event']
event = self.createEvents(event_id)
event.type = '入仕事件'
field = 'c_year'
year = event_node[field]
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('entry_data', field)
year_range = event_node[range_field]
event.addTimeAndRange(year, year_range)
if 'c_sequence' in node_data.keys():
sequence = node_data['c_sequence']
if sequence != 'None' and sequence is not None:
event.sequence = int(sequence)
id_list.append(event_id)
person = personManager.createPerson(person)
event.addPerson(person, '主角')
if len(id_list) == 0:
return False
id_list_str = ','.join(id_list)
# 入仕法
results = graph.run(
'START event=node({}) MATCH (method:Entry_codes)--(event) RETURN method.c_entry_desc_chn AS method, id(event)'
.format(id_list_str))
results = results.data()
# print(results)
for result in results:
method = result['method']
event_id = str(result['id(event)'])
self.createEvents(event_id).setTrigger('入仕')
self.createEvents(event_id).detail = str(method)
# self.createEvents(event_id).setTrigger(str(method))
print('加载入仕方式')
# 相关机构
if len(id_list) < LIMIT:
return False
# 分数
return True
def loadPostOfficeEvents(self, LIMIT=1000000, SKIP=0, person_id=None):
print('开始加载仕途事件')
# 使用neo4j提取
# 使用Posting_Data可能有问题，有些信息未连起来
if person_id is None:
query = 'MATCH (person:Biog_main)--(n1:Posted_to_office_data)--(n2:Posting_data)--(n3:Posted_to_addr_data) RETURN person,n1,n2,n3,id(n1),id(n2),id(n3) SKIP {} LIMIT {} '.format(
str(SKIP), str(LIMIT))
else:
query = 'MATCH (person:Biog_main{{ c_personid:"{}" }})--(n1:Posted_to_office_data)--(n2:Posting_data)--(n3:Posted_to_addr_data) RETURN person,n1,n2,n3,id(n1),id(n2),id(n3) SKIP {} LIMIT {} '.format(
str(person_id), str(SKIP), str(LIMIT))
results = graph.run(query).data()
id_list1 = []
id_list2 = []
id_list3 = []
id2id = {}
for node_data in results:
node_id1 = str(node_data['id(n1)'])
node_id2 = str(node_data['id(n2)'])
node_id3 = str(node_data['id(n3)'])
node_data1 = node_data['n1']
node_data2 = node_data['n2']
node_data3 = node_data['n3']
event1 = self.createEvents('担任' + node_id1) #担任
event2 = self.createEvents('卸任' + node_id1) #卸任
event1.type = '官职事件'
event2.type = '官职事件'
field = 'c_firstyear'
year = node_data1[field]
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('posted_to_office_data',
field)
year_range = node_data1[range_field]
event1.addTimeAndRange(year, year_range)
field = 'c_lastyear'
event2.addTimeAndRange(event1.time_range[0], '之后')
year = node_data1[field]
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('posted_to_office_data',
field)
year_range = node_data1[range_field]
event2.addTimeAndRange(year, year_range)
if 'c_sequence' in node_data1.keys():
sequence = node_data1['c_sequence']
if sequence != 'None' and sequence is not None:
event1.sequence = int(sequence)
event2.sequence = int(sequence)
id_list1.append(node_id1)
id_list2.append(node_id2)
id_list3.append(node_id3)
id2id[node_id1] = node_id1
id2id[node_id2] = node_id1
id2id[node_id3] = node_id1
person = personManager.createPerson(node_data['person'])
event1.addPerson(person, '主角')
event2.addPerson(person, '主角')
event1.setTrigger('担任')
event2.setTrigger('卸任')
if len(id_list1) == 0:
return False
id_list_str1 = ','.join(id_list1)
id_list_str2 = ','.join(id_list2)
id_list_str3 = ','.join(id_list3)
# print(id_list_str3)
# 加载地点
results = graph.run(
'START n=node({}) MATCH (addr:Addr_codes)-[r]-(n) RETURN addr.c_addr_id as addr_id, id(n)'
.format(id_list_str3))
results = results.data()
# print('START n=node({}) MATCH (addr:Addr_codes)-[r]-(n) RETURN addr.c_addr_id as addr_id, id(n)'.format(id_list_str3))
# print(results)
for result in results:
addr_id = result['addr_id']
event_id = str(result['id(n)'])
# print(addr_id)
self.createEvents('担任' + id2id[event_id]).setAddr(
addrManager.getAddr(addr_id))
self.createEvents('卸任' + id2id[event_id]).setAddr(
addrManager.getAddr(addr_id))
print('加载仕途数据地址')
# 加载职位
results = graph.run(
'START n=node({}) MATCH (office:Office_codes)-[r]-(n) RETURN office, id(n)'
.format(id_list_str1))
results = results.data()
# print('START n=node({}) MATCH (addr:Addr_codes)-[r]-(n) RETURN addr.c_addr_id as addr_id, id(n)'.format(id_list_str3))
# print(results)
for result in results:
office_code = result['office']
event_id = str(result['id(n)'])
event1 = self.createEvents('担任' + id2id[event_id])
event2 = self.createEvents('卸任' + id2id[event_id])
event1.detail = '担任' + str(office_code['c_office_chn'])
event2.detail = '卸任' + str(office_code['c_office_chn'])
# self.createEvents('担任'+id2id[event_id]).setTrigger('担任'+str(office_code['c_office_chn']))
# self.createEvents('卸任'+id2id[event_id]).setTrigger('卸任'+str(office_code['c_office_chn']))
print('加载仕途官职')
# 加载官职的授予方式
results = graph.run(
'START n=node({}) MATCH (office:Appointment_type_codes)-[r]-(n) RETURN office, id(n)'
.format(id_list_str1))
results = results.data()
# print('START n=node({}) MATCH (addr:Addr_codes)-[r]-(n) RETURN addr.c_addr_id as addr_id, id(n)'.format(id_list_str3))
# print(results)
for result in results:
office_code = result['office']['c_appt_type_desc_chn']
event_id = str(result['id(n)'])
self.createEvents(
'担任' + id2id[event_id]).detail += '授予方式' + str(office_code)
# self.createEvents('担任'+id2id[event_id]).setTrigger('担任'+str(office_code['c_office_chn']))
# self.createEvents('卸任'+id2id[event_id]).setTrigger('卸任'+str(office_code['c_office_chn']))
print('加载仕途官职')
if len(id_list1) < LIMIT:
return False
return True
def loadAddrEvents(self, LIMIT=1000000, SKIP=0, person_id=None):
print('开始加载迁移事件')
if person_id is None:
query = 'MATCH (person:Biog_main)--(event:Biog_addr_data) RETURN person, event, id(event) SKIP {} LIMIT {} '.format(
str(SKIP), str(LIMIT))
else:
query = 'MATCH (person:Biog_main{{c_personid:"{}"}})-->(event:Biog_addr_data) RETURN person, event, id(event) SKIP {} LIMIT {} '.format(
str(person_id), str(SKIP), str(LIMIT))
# print(query)
results = graph.run(query).data()
id_list = []
for node_data in results:
# print(node_data)
node_id = node_data['id(event)']
person = node_data['person']
node_data = node_data['event']
event = self.createEvents(node_id)
event.type = '前往'
event.addPerson(personManager.createPerson(person), '主角')
event.setTrigger('前往')
event.type = '前往'
event.detail = '前往某地'
field = 'c_firstyear'
year = node_data[field]
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('Biog_addr_data', field)
year_range = node_data[range_field]
event.addTimeAndRange(year, year_range)
event = self.createEvents('离开' + str(node_id))
event.addPerson(personManager.createPerson(person), '主角')
event.setTrigger('离开')
event.type = '离开'
event.detail = '离开某地'
field = 'c_lastyear'
if field in node_data.keys():
year = node_data[field]
# print(year)
if year is not None and year != 0 and year != '0' and year != 'None':
year = int(year)
range_field = graph.getTableRange('Biog_addr_data', field)
year_range = node_data[range_field]
# print(year)
event.addTimeAndRange(year, year_range)
id_list.append(str(node_id))
if len(id_list) == 0:
return False
# print(id_list)
id_list_str = ','.join(id_list)
# 加载地点
results = graph.run(
'START n=node({}) MATCH (addr:Addr_codes)--(n) RETURN addr.c_addr_id as addr_id, id(n)'
.format(id_list_str))
results = results.data()
# print(results)
for result in results:
addr_id = result['addr_id']
event_id = str(result['id(n)'])
# print(addr_id)
event = self.createEvents(event_id)
addr = addrManager.getAddr(addr_id)
event.setAddr(addr)
event.detail = '前往' + str(addr.name)
event = self.createEvents('离开' + event_id)
addr = addrManager.getAddr(addr_id)
event.setAddr(addr)
event.detail = '离开' + str(addr.name)
# 加载原因
print('加载迁移数据地址')
results = graph.run(
'START n=node({}) MATCH (addr:Biog_addr_codes)--(n) RETURN addr.c_addr_desc_chn as desc, id(n)'
.format(id_list_str))
results = results.data()
# print(results)
for result in results:
desc = str(result['desc'])
event_id = str(result['id(n)'])
# print(addr_id)
event = self.createEvents(event_id)
event.setTrigger(desc)
event.detail = event.detail + '由于' + desc
print('加载迁移原因')
if len(id_list) < LIMIT:
return False
return True
