def draw_graph(relation_list):
result = {}
key_dict = {
'User': '******',
'Event': 'event_id',
'Group': 'group',
'SpecialEvent': 'event'
}
map_eid = []
result_relation = [] #[[node1,relation,node2],...]
for i in relation_list:
# print i,'**********************'
this_relation = ['', '', ''] #[node1,relation,node2]
only_relation = [] #[node1,node2]
for m in walk(i):
try:
this_relation[1] = m.type()
# print m.type(),'!!!!!!!!!'
except:
aa = m.labels()
aa = [i for i in aa]
if len(aa) == 1:
try:
primary_key = key_dict[aa[0]]
except:
continue
primary_value = m[primary_key]
only_relation.append(primary_value)
if aa[0] == 'User':
eu_name = user_name_search(m['uid'])
elif aa[0] == 'Event':
# print m['event_id'].encode('utf-8'),'************'
if m['event_id'] in [u'徐玉玉事件', u'大学生失联']:
continue
eu_name = event_name_search(m['event_id'])
map_eid.append(m['event_id'])
else:
eu_name = m[primary_key]
if len(aa) > 1 or len(aa) < 1:
primary_key = 'User'
primary_value = m[primary_key]
eu_name = user_name_search(m['uid'])
only_relation.append(m['uid'])
try:
result[primary_key][primary_value] = eu_name
except:
result[primary_key] = {}
result[primary_key][primary_value] = eu_name
if len(only_relation) < 2:
continue
this_relation[0] = only_relation[0]
this_relation[2] = only_relation[1]
result_relation.append(this_relation)
return {
'result_relation': result_relation,
'node': result,
'map_eid': map_eid
}
def event_list_theme(event):
s_string = 'START s0 = node:special_event_index(event="%s")\
MATCH (s0)-[r]-(s:Event) RETURN s' % (event)
# print s_string
e_list = graph.run(s_string)
e_list_l = []
for i in e_list:
print i
e_this = dict(i)['s']['event_id']
e_name = event_name_search(e_this)
e_list_l.append([e_this, e_name])
return e_list_l
def search_way(node1, node2, node_type1, node_type2):
relation_type = [
'join', 'pusher', 'maker', 'other_relationship', 'friend', 'relative',
'colleague', 'user_tag'
]
index_type_dict = {'User': '******', 'Event': 'event_index'}
primary_idct = {'User': '******', 'Event': 'event'}
key_dict = {'User': '******', 'Event': 'event_id'}
origin_idlist = [node1, node2]
print origin_idlist
if node_type1 == 'User':
start_node_card = related_user_search([node1], 'activeness')[0]
else:
start_node_card = event_detail_search([node1], 'start_ts')[0]
if node_type2 == 'User':
end_node_card = related_user_search([node2], 'activeness')[0]
else:
end_node_card = event_detail_search([node2], 'start_ts')[0]
c_string = 'START node1 = node:' + index_type_dict[
node_type1] + '(' + primary_idct[node_type1] + '="' + node1 + '"),'
c_string += 'node2 = node:' + index_type_dict[
node_type2] + '(' + primary_idct[node_type2] + '="' + node2 + '") '
c_string += 'MATCH p = allShortestPaths(node1-[r*..5]-node2) return r'
print c_string
result = graph.run(c_string)
# uid_list = []#for card
# eid_list = []#for card
middle_card = [] #for card, middle nodes
uid_dict = {} #for graph
eid_dict = {} #for graph
relation_all = list(result)
# print relation_all,'!!!!!!!!!!!!!!!1'
relation_result = []
relation_result2 = []
# print [relation_all[0]['r'] , relation_all[1]['r']]
# if relation_all[0]['r'] == relation_all[1]['r']:
# return 'haha'
# else:
# return [relation_all[0]['r'] , relation_all[1]['r']]
length_relation = [0, 0]
length_relation[0] = len(relation_all)
for relation in relation_all:
# print list(relation['r']),'99999999999999999'
if len(list(relation['r'])) < 2:
return 0 #返回0 说明这两个节点有直接关系
print relation, 'relation'
length_relation[1] = len(list(relation['r']))
line_rel = []
for i in relation['r']:
# print i
a = walk(i)
# print a
this_relation = []
# aa = []
for m in a: #a=[node1,r,node2]
try:
m.type()
except:
aa = m.labels()
aa = [i for i in aa]
mm = dict(m)
# print mm,'========'
if mm.has_key('uid'):
# print mm
if m['uid'] == '1765891182':
print a, m, '----000000000000000000'
eu_name = user_name_search(m['uid'])
if uid_dict.has_key(m['uid']) == False:
uid_dict[m['uid']] = eu_name
if m['uid'] not in origin_idlist:
# print m['uid'], origin_idlist,'inininini'
mid_card = related_user_search([m['uid']],
'activeness')
if len(mid_card) == 0:
middle_card.append({'uid': m['uid']})
else:
# print len(mid_card), '!!!!!!!!!!!!!'
middle_card.append(mid_card[0])
this_relation.append([m['uid'], eu_name])
if [m['uid'], eu_name] not in line_rel:
line_rel.append([m['uid'], eu_name])
elif mm.has_key('event_id'):
eu_name = event_name_search(m['event_id'])
if eid_dict.has_key(m['event_id']) == False:
eid_dict[m['event_id']] = eu_name
if m['event_id'] not in origin_idlist:
mid_card = event_detail_search([m['event_id']],
'start_ts')
if len(mid_card) == 0:
middle_card.append(
{'event_id': m['event_id']})
else:
middle_card.append(mid_card[0])
this_relation.append([m['event_id'], eu_name])
if [m['event_id'], eu_name] not in line_rel:
line_rel.append([m['event_id'], eu_name])
else:
break
if len(this_relation) > 1:
if this_relation not in relation_result:
relation_result.append(this_relation)
# print len(line_rel), length_relation[1],'000000000'
# if len(line_rel) == length_relation[1]:
# relation_result2.append(line_rel)
return {'relation':relation_result, 'start_node_card':start_node_card, 'end_node_card':end_node_card,\
'user_nodes':uid_dict, 'event_nodes': eid_dict, 'middle_card':middle_card,'length_relation':length_relation}
def search_related_user(item):
query_body = {
"query": {
'bool': {
'should': [{
"wildcard": {
'uid': '*' + str(item.encode('utf-8')) + '*'
}
}, {
"wildcard": {
'uname': '*' + str(item.encode('utf-8')) + '*'
}
}]
}
},
'size': 10
}
only_uid = []
user_uid_list = []
u_nodes_list = {}
try:
name_results = es_user_portrait.search(index=portrait_name, doc_type=portrait_type, \
body=query_body, fields=['uid','uname'])['hits']['hits']
# print name_results,'@@@@@@@@@@@@@@@@@'
except:
return 'node does not exist'
if len(name_results) == 0:
return 'node does not exist'
for i in name_results:
# print i
uid = i['fields']['uid'][0]
uname = i['fields']['uname'][0]
only_uid.append(uid)
u_nodes_list[uid] = uname
user_uid_list.append([uid, uname])
print len(user_uid_list)
e_nodes_list = {}
user_relation = []
mid_uid_list = [] #存放第一层的数据，再以这些为起始点，扩展第二层
mid_eid_list = []
for uid_value in user_uid_list:
c_string = 'START s0 = node:node_index(uid="' + str(
uid_value[0]) + '") '
c_string += 'MATCH (s0)-[r1]-(s1) return s0,r1,s1 LIMIT 1'
result = graph.run(c_string)
# print list(result),'-----------------'
for i in list(result):
start_id = i['s0']['uid']
# # start_id = s0['uid']
relation1 = i['r1'].type()
m_id = dict(i['s1'])
if m_id.has_key('uid'):
middle_id = m_id['uid']
mid_uid_list.append(middle_id)
user_name = user_name_search(middle_id)
# print middle_id,'2222222222222222222'
u_nodes_list[str(middle_id)] = user_name
user_relation.append([start_id, relation1, middle_id])
if m_id.has_key('envent_id'):
middle_id = m_id['envent_id']
mid_eid_list.append(middle_id)
event_name = event_name_search(middle_id)
e_nodes_list[str(middle_id)] = event_name
user_relation.append([start_id, relation1, middle_id])
print len(mid_uid_list)
print len(mid_eid_list), '++++++++++++++++'
for mid_uid in mid_uid_list:
c_string = 'START s1 = node:node_index(uid="' + str(mid_uid) + '") '
c_string += 'MATCH (s1)-[r2]->(s2:User) return s1,r2,s2 LIMIT 5'
# print c_string
result = graph.run(c_string)
for i in result:
start_mid_id = i['s1']['uid']
relation2 = i['r2'].type()
end_id = dict(i['s2'])
if end_id.has_key('uid'):
user_name = user_name_search(end_id['uid'])
# print end_id['uid'],'333333333333333333333333'
u_nodes_list[end_id['uid']] = user_name
user_relation.append([start_mid_id, relation2, end_id['uid']])
if end_id.has_key('envent_id'):
event_name = event_name_search(end_id['event_id'])
e_nodes_list[end_id['event_id']] = event_name
user_relation.append(
[start_mid_id, relation2, end_id['envent_id']])
for mid_eid in mid_eid_list:
c_string = 'START s1 = node:event_index(event="' + str(mid_eid) + '") '
c_string += 'MATCH (s1)-[r2]->(s2:User) return s1,r2,s2 LIMIT 3'
event_result = graph.run(c_string)
for i in event_result:
relation2 = i['r2'].type()
end_id = dict(i['s2'])
if end_id.has_key('uid'):
# print end_id['uid'],'44444444444444444444444'
user_name = user_name_search(end_id['uid'])
u_nodes_list[end_id['uid']] = user_name
user_relation.append([mid_eid, relation2, end_id['uid']])
if end_id.has_key('envent_id'):
event_name = event_name_search(end_id['event_id'])
e_nodes_list[end_id['event_id']] = event_name
user_relation.append([mid_eid, relation2, end_id['envent_id']])
return {'total_user':len(user_uid_list),'user_nodes':u_nodes_list,'event_nodes':e_nodes_list,\
'relation':user_relation,'draw_nodes_length':len(u_nodes_list)}
def compare_graph_theme(theme_name1, theme_name2, layer, diff):
s_string1 = 'START s0 = node:special_event_index(event="%s")\
MATCH (s0)-[r]-(s) RETURN s.event_id' % theme_name1
theme_result1 = graph.run(s_string1)
eid_list1 = []
for i in theme_result1:
e_dict = dict(i)
print e_dict, '&&&&&&&&&&&&'
esd = e_dict['s.event_id']
eid_list1.append(esd)
print len(eid_list1)
s_string2 = 'START s0 = node:special_event_index(event="%s")\
MATCH (s0)-[r]-(s) RETURN s.event_id' % theme_name2
theme_result2 = graph.run(s_string2)
eid_list2 = []
for i in theme_result2:
e_dict = dict(i)
esd = e_dict['s.event_id']
eid_list2.append(esd)
print len(eid_list2)
relation_1 = get_graph(eid_list1, layer)
relation_2 = get_graph(eid_list2, layer)
if diff == '0':
e1 = draw_graph(relation_1)
e2 = draw_graph(relation_2)
if diff == '1':
same_relation = set(relation_1) & set(relation_2)
same_relation = [i for i in same_relation]
e1 = draw_graph(same_relation)
e2 = e1
if diff == '2':
same_relation = set(relation_1) & set(relation_2)
only1_relation = set(relation_1) - same_relation
only2_relation = set(relation_2) - same_relation
e1 = draw_graph(only1_relation)
e2 = draw_graph(only2_relation)
for i in eid_list1:
e_name = event_name_search(i)
try:
e1['node']['event_id'][i] = e_name
except:
e1['node']['event_id'] = {}
e1['node']['event_id'][i] = e_name
try:
e1['map_eid'].append(i)
except:
e1['map_eid'] = []
e1['map_eid'].append(i)
for i in eid_list2:
e_name = event_name_search(i)
try:
e2['node']['event_id'][i] = e_name
# e2[]
except:
e2['node']['event_id'] = {}
e2['node']['event_id'][i] = e_name
try:
e2['map_eid'].append(i)
except:
e2['map_eid'] = []
e2['map_eid'].append(i)
return {'e1': e1, 'e2': e2}
def search_related_event_f(item):
query_body = {
"query": {
'bool': {
'should': [{
"wildcard": {
'keywords': '*' + str(item.encode('utf-8')) + '*'
}
}, {
"wildcard": {
'en_name': '*' + str(item.encode('utf-8')) + '*'
}
}, {
"wildcard": {
'name': '*' + str(item.encode('utf-8')) + '*'
}
}]
}
},
'size': 10
}
only_eid = []
event_id_list = []
u_nodes_list = {}
e_nodes_list = {}
event_relation = []
try:
name_results = es_event.search(index=event_name, doc_type=event_type, \
body=query_body, fields=['name','en_name'])['hits']['hits']
except:
return 'node does not exist'
if len(name_results) == 0:
return 'node does not exist'
print name_results, '*********************'
for i in name_results:
name = i['fields']['name'][0]
en_name = i['fields']['en_name'][0]
only_eid.append(en_name)
e_nodes_list[en_name] = name
event_id_list.append([en_name, name])
for event_value in event_id_list:
c_string = 'START s0 = node:event_index(event="' + str(
event_value[0]) + '") '
c_string += 'MATCH (s0)-[r1]-(s1) return s0,r1,s1 LIMIT 10'
# print c_string,'==========='
mid_eid_list = [] #存放第一层的数据，再以这些为起始点，扩展第二层
mid_uid_list = []
result = graph.run(c_string)
# print list(result),'-----------------'
for i in list(result):
print i
start_id = i['s0']['event_id']
# start_id = s0['event']
relation1 = i['r1'].type()
m_id = dict(i['s1'])
if m_id.has_key('uid'):
middle_id = m_id['uid']
mid_uid_list.append(middle_id)
user_name = user_name_search(middle_id)
u_nodes_list[middle_id] = user_name
event_relation.append([start_id, relation1, middle_id])
if m_id.has_key('envent_id'):
middle_id = m_id['envent_id']
mid_eid_list.append(middle_id)
event_name2 = event_name_search(middle_id)
e_nodes_list[middle_id] = event_name2
event_relation.append([start_id, relation1, middle_id])
# print mid_uid_list
# print mid_eid_list,'++++++++++++++++'
for mid_uid in mid_uid_list:
c_string = 'START s1 = node:node_index(uid="' + str(mid_uid) + '") '
c_string += 'MATCH (s1)-[r2]->(s2:Event) return s1,r2,s2 LIMIT 5'
uid_result = graph.run(c_string)
for i in uid_result:
relation2 = i['r2'].type()
end_id = dict(i['s2'])
if end_id.has_key('uid'):
user_name = user_name_search(end_id['uid'])
u_nodes_list[end_id['uid']] = user_name
event_relation.append([mid_uid, relation2, end_id['uid']])
if end_id.has_key('envent_id'):
event_name2 = event_name_search(end_id['envent_id'])
e_nodes_list[end_id['envent_id']] = event_name2
event_relation.append(
[mid_uid, relation2, end_id['envent_id']])
for mid_eid in mid_eid_list:
c_string = 'START s1 = node:event_index(event="' + str(mid_eid) + '") '
c_string += 'MATCH (s1)-[r2]->(s2:Event) return s1,r2,s2 LIMIT 5'
eid_result = graph.run(c_string)
for i in eid_result:
relation2 = i['r2'].type()
end_id = dict(i['s2'])
if end_id.has_key('uid'):
user_name = user_name_search(end_id['uid'])
u_nodes_list[end_id['uid']] = user_name
event_relation.append([mid_eid, relation2, end_id['uid']])
if end_id.has_key('envent_id'):
event_name2 = event_name_search(end_id['envent_id'])
e_nodes_list[end_id['envent_id']] = event_name2
event_relation.append(
[mid_eid, relation2, end_id['envent_id']])
return {'total_event':len(event_id_list),'user_nodes':u_nodes_list,'event_nodes':e_nodes_list,\
'relation':event_relation}
def theme_tab_graph(theme_name, node_type, relation_type, layer):
s_string = 'START s0 = node:special_event_index(event="%s")\
MATCH (s0)-[r]-(s) RETURN s.event_id as event' % theme_name
# print s_string
all_event_id = [] #for map,event_id
event_result = graph.run(s_string)
s_string2 = 'START s0 = node:special_event_index(event="%s")\
MATCH (s0)-[r]-(s) RETURN r' % theme_name
event_result_r = graph.run(s_string2)
relation_list_o = []
for r in event_result_r:
r1 = dict(r)['r']
relation_list_o.append(r1)
# print event_list
# b = Node("Group", group=group_name)
# print g.degree(b),'-=-=-=-=-=----------------'
if node_type != '':
node_type = ':' + node_type
# if relation_type!='':
# relation_type = ':' + relation_type
event_relation = []
only_event_o = []
# total_event = len(list(event_list))
event_list = []
e_nodes_list = {} #all event nodes
u_nodes_list = {} #all user nodes
for event in event_result:
event_value = event['event']
only_event_o.append(event_value)
event_name = event_name_search(event_value)
event_list.append([event_value, event_name]) #取event
e_nodes_list[event_value] = event_name
all_event_id.extend(event_list)
relation = get_graph_single(only_event_o, node_type, relation_type, layer)
relation.extend(relation_list_o)
relation = [i for i in set(relation)]
result = draw_graph(relation)
# try:
# result['node']['event'][theme_name] = theme_name
# except:
# result['node']['event'] = {}
# result['node']['event'][theme_name] = theme_name
for i in only_event_o:
# e_name = event_name_search(i)
# try:
# result['node']['event_id'][i] = e_name
# except:
# result['node']['event_id'] ={}
# result['node']['event_id'][i] = e_name
try:
result['map_eid'].append(i)
except:
result['map_eid'] = []
result['map_eid'].append(i)
result['map_eid'] = [i for i in set(result['map_eid'])]
# print len(result['node']['event_id']), len(result['map_eid'])
return result