回答編集履歴
6
調整
answer
CHANGED
|
@@ -210,7 +210,7 @@
|
|
|
210
210
|
and not t1.fid = t3.fid
|
|
211
211
|
and not t1.rid = t2.rid
|
|
212
212
|
and not exists(
|
|
213
|
-
select
|
|
213
|
+
select 1
|
|
214
214
|
from user_relation as t5
|
|
215
215
|
,user_relation as t6
|
|
216
216
|
where 1
|
5
追記
answer
CHANGED
|
@@ -190,4 +190,35 @@
|
|
|
190
190
|
|15|8|
|
|
191
191
|
|15|12|
|
|
192
192
|
|15|13|
|
|
193
|
-
|15|14|
|
|
193
|
+
|15|14|
|
|
194
|
+
|
|
195
|
+
# 参考
|
|
196
|
+
上記SQLはオプティマイザでは以下のように理解されているようです
|
|
197
|
+
```SQL
|
|
198
|
+
select distinct t1.uid AS self,t4.uid AS friendoffriend
|
|
199
|
+
from user_relation as t1
|
|
200
|
+
,user_relation as t2
|
|
201
|
+
,user_relation as t3
|
|
202
|
+
,user_relation as t4
|
|
203
|
+
where 1
|
|
204
|
+
and t1.sid = 1
|
|
205
|
+
and t2.fid = t1.fid
|
|
206
|
+
and t3.sid = 1
|
|
207
|
+
and t3.uid = t2.uid
|
|
208
|
+
and t4.fid = t3.fid
|
|
209
|
+
and not t2.uid = t4.uid
|
|
210
|
+
and not t1.fid = t3.fid
|
|
211
|
+
and not t1.rid = t2.rid
|
|
212
|
+
and not exists(
|
|
213
|
+
select t5.uid as self,t6.uid as friend
|
|
214
|
+
from user_relation as t5
|
|
215
|
+
,user_relation as t6
|
|
216
|
+
where 1
|
|
217
|
+
and t5.sid = 1
|
|
218
|
+
and t6.fid = t5.fid
|
|
219
|
+
and t4.uid = t6.uid
|
|
220
|
+
and t1.uid = t5.uid
|
|
221
|
+
and not t5.rid = t6.rid
|
|
222
|
+
)
|
|
223
|
+
order by t1.uid,t4.uid
|
|
224
|
+
```
|
4
chousei
answer
CHANGED
|
@@ -115,11 +115,11 @@
|
|
|
115
115
|
inner join user_relation as t2 on t1.fid=t2.fid and not t1.rid=t2.rid and t1.sid=1
|
|
116
116
|
inner join user_relation as t3 on t2.uid=t3.uid and not t2.fid=t3.fid and t3.sid=1
|
|
117
117
|
inner join user_relation as t4 on t3.fid=t4.fid and not t3.uid=t4.uid
|
|
118
|
-
) as
|
|
118
|
+
) as t7 where not exists(
|
|
119
119
|
select t5.uid as self,t6.uid as friend
|
|
120
120
|
from user_relation as t5
|
|
121
121
|
inner join user_relation as t6 on t5.fid=t6.fid and not t5.rid=t6.rid and t5.sid=1
|
|
122
|
-
where
|
|
122
|
+
where t7.friendoffriend=t6.uid and t7.self=t5.uid
|
|
123
123
|
)
|
|
124
124
|
order by self,friendoffriend
|
|
125
125
|
```
|
3
追記
answer
CHANGED
|
@@ -77,4 +77,117 @@
|
|
|
77
77
|
```SQL
|
|
78
78
|
insert into follow(userid,follow_userid,statusid) values(2,1,3);
|
|
79
79
|
```
|
|
80
|
-
user_relationを確認するとfid=1のデータが削除され、新たにケツにデータが2件登録されます
|
|
80
|
+
user_relationを確認するとfid=1のデータが削除され、新たにケツにデータが2件登録されます
|
|
81
|
+
|
|
82
|
+
# 友達の友達
|
|
83
|
+
- 上記testをした場合は１度データは元にもどしておいて下さい
|
|
84
|
+
```SQL
|
|
85
|
+
truncate follow;
|
|
86
|
+
truncate user_relation;
|
|
87
|
+
insert into follow(userid,follow_userid,statusid) values
|
|
88
|
+
(1,2,1),(1,3,1),(1,7,1),(1,8,1),(4,3,1),
|
|
89
|
+
(4,8,1),(4,7,1),(6,10,1),(7,10,2),(19,20,2),
|
|
90
|
+
(8,10,1),(2,3,1),(15,1,1),(14,1,1),(13,1,1),
|
|
91
|
+
(12,1,1),(11,1,2),(1,9,2),(3,20,2);
|
|
92
|
+
```
|
|
93
|
+
|
|
94
|
+
- 友達の友達を表示します
|
|
95
|
+
```SQL
|
|
96
|
+
select * from (
|
|
97
|
+
select t1.uid as self,t2.uid as friend,t4.uid as friendoffriend from user_relation as t1
|
|
98
|
+
inner join user_relation as t2 on t1.fid=t2.fid and not t1.rid=t2.rid and t1.sid=1
|
|
99
|
+
inner join user_relation as t3 on t2.uid=t3.uid and not t2.fid=t3.fid and t3.sid=1
|
|
100
|
+
inner join user_relation as t4 on t3.fid=t4.fid and not t3.uid=t4.uid
|
|
101
|
+
) as s1 where not exists(
|
|
102
|
+
select t5.uid as self,t6.uid as friend
|
|
103
|
+
from user_relation as t5
|
|
104
|
+
inner join user_relation as t6 on t5.fid=t6.fid and not t5.rid=t6.rid and t5.sid=1
|
|
105
|
+
where s1.friendoffriend=t6.uid and s1.self=t5.uid
|
|
106
|
+
)
|
|
107
|
+
order by self,friend,friendoffriend;
|
|
108
|
+
```
|
|
109
|
+
|
|
110
|
+
- (応用)selfに対してfriendoffirendをユニークに表示します
|
|
111
|
+
|
|
112
|
+
```SQL
|
|
113
|
+
select distinct self,friendoffriend from (
|
|
114
|
+
select t1.uid as self,t2.uid as friend,t4.uid as friendoffriend from user_relation as t1
|
|
115
|
+
inner join user_relation as t2 on t1.fid=t2.fid and not t1.rid=t2.rid and t1.sid=1
|
|
116
|
+
inner join user_relation as t3 on t2.uid=t3.uid and not t2.fid=t3.fid and t3.sid=1
|
|
117
|
+
inner join user_relation as t4 on t3.fid=t4.fid and not t3.uid=t4.uid
|
|
118
|
+
) as s1 where not exists(
|
|
119
|
+
select t5.uid as self,t6.uid as friend
|
|
120
|
+
from user_relation as t5
|
|
121
|
+
inner join user_relation as t6 on t5.fid=t6.fid and not t5.rid=t6.rid and t5.sid=1
|
|
122
|
+
where s1.friendoffriend=t6.uid and s1.self=t5.uid
|
|
123
|
+
)
|
|
124
|
+
order by self,friendoffriend
|
|
125
|
+
```
|
|
126
|
+
- 結果
|
|
127
|
+
|
|
128
|
+
|self|friendoffriend|
|
|
129
|
+
|--:|--:|
|
|
130
|
+
|1|4|
|
|
131
|
+
|1|10|
|
|
132
|
+
|2|4|
|
|
133
|
+
|2|7|
|
|
134
|
+
|2|8|
|
|
135
|
+
|2|12|
|
|
136
|
+
|2|13|
|
|
137
|
+
|2|14|
|
|
138
|
+
|2|15|
|
|
139
|
+
|3|7|
|
|
140
|
+
|3|8|
|
|
141
|
+
|3|12|
|
|
142
|
+
|3|13|
|
|
143
|
+
|3|14|
|
|
144
|
+
|3|15|
|
|
145
|
+
|4|1|
|
|
146
|
+
|4|2|
|
|
147
|
+
|4|10|
|
|
148
|
+
|6|8|
|
|
149
|
+
|7|2|
|
|
150
|
+
|7|3|
|
|
151
|
+
|7|8|
|
|
152
|
+
|7|12|
|
|
153
|
+
|7|13|
|
|
154
|
+
|7|14|
|
|
155
|
+
|7|15|
|
|
156
|
+
|8|2|
|
|
157
|
+
|8|3|
|
|
158
|
+
|8|6|
|
|
159
|
+
|8|7|
|
|
160
|
+
|8|12|
|
|
161
|
+
|8|13|
|
|
162
|
+
|8|14|
|
|
163
|
+
|8|15|
|
|
164
|
+
|10|1|
|
|
165
|
+
|10|4|
|
|
166
|
+
|12|2|
|
|
167
|
+
|12|3|
|
|
168
|
+
|12|7|
|
|
169
|
+
|12|8|
|
|
170
|
+
|12|13|
|
|
171
|
+
|12|14|
|
|
172
|
+
|12|15|
|
|
173
|
+
|13|2|
|
|
174
|
+
|13|3|
|
|
175
|
+
|13|7|
|
|
176
|
+
|13|8|
|
|
177
|
+
|13|12|
|
|
178
|
+
|13|14|
|
|
179
|
+
|13|15|
|
|
180
|
+
|14|2|
|
|
181
|
+
|14|3|
|
|
182
|
+
|14|7|
|
|
183
|
+
|14|8|
|
|
184
|
+
|14|12|
|
|
185
|
+
|14|13|
|
|
186
|
+
|14|15|
|
|
187
|
+
|15|2|
|
|
188
|
+
|15|3|
|
|
189
|
+
|15|7|
|
|
190
|
+
|15|8|
|
|
191
|
+
|15|12|
|
|
192
|
+
|15|13|
|
|
193
|
+
|15|14|
|
2
sample
answer
CHANGED
|
@@ -3,4 +3,78 @@
|
|
|
3
3
|
検索の際はunionして処理するのでパフォーマンスもSQLの視認性も期待できません。
|
|
4
4
|
|
|
5
5
|
followテーブルからstatusを切り離し、triggerでデータ投入・削除時に
|
|
6
|
-
別テーブルで管理すると良いかもしれません
|
|
6
|
+
別テーブルで管理すると良いかもしれません
|
|
7
|
+
|
|
8
|
+
# sample
|
|
9
|
+
私の方からの説明が不足しているので以下サンプルをあげておきます
|
|
10
|
+
- テーブル作成
|
|
11
|
+
```SQL
|
|
12
|
+
create table follow(
|
|
13
|
+
fid int primary key auto_increment,
|
|
14
|
+
userid int not null,
|
|
15
|
+
follow_userid int not null,
|
|
16
|
+
statusid tinyint not null
|
|
17
|
+
);
|
|
18
|
+
create table user_relation (
|
|
19
|
+
rid int primary key auto_increment,/*リレーションテーブル用のid*/
|
|
20
|
+
fid int not null, /*followテーブルのid*/
|
|
21
|
+
uid int not null, /*ユーザーid*/
|
|
22
|
+
sid int not null, /*ステータスid*/
|
|
23
|
+
flg tinyint default 0 /*削除用フラグ*/
|
|
24
|
+
);
|
|
25
|
+
```
|
|
26
|
+
|
|
27
|
+
- followテーブルにtriggerを仕込みます
|
|
28
|
+
```SQL
|
|
29
|
+
drop trigger if exists trg_bef_insert_follow;
|
|
30
|
+
drop trigger if exists trg_aft_insert_follow;
|
|
31
|
+
drop trigger if exists trg_delete_follow;
|
|
32
|
+
delimiter //
|
|
33
|
+
create trigger trg_bef_insert_follow before insert on follow
|
|
34
|
+
for each row begin
|
|
35
|
+
update user_relation as t1,user_relation as t2
|
|
36
|
+
set t1.flg=1,t2.flg=1 where t1.fid=t2.fid and
|
|
37
|
+
(t1.uid=new.userid and t2.uid=new.follow_userid or
|
|
38
|
+
t2.uid=new.userid and t1.uid=new.follow_userid);
|
|
39
|
+
end
|
|
40
|
+
//
|
|
41
|
+
create trigger trg_aft_insert_follow after insert on follow
|
|
42
|
+
for each row begin
|
|
43
|
+
insert user_relation(fid,uid,sid) values
|
|
44
|
+
(new.fid,new.userid,new.statusid),
|
|
45
|
+
(new.fid,new.follow_userid,new.statusid);
|
|
46
|
+
delete from user_relation where flg>0;
|
|
47
|
+
end
|
|
48
|
+
//
|
|
49
|
+
create trigger trg_delete_follow after delete on follow
|
|
50
|
+
for each row begin
|
|
51
|
+
delete from user_relation where fid=old.fid;
|
|
52
|
+
end
|
|
53
|
+
//
|
|
54
|
+
delimiter ;
|
|
55
|
+
```
|
|
56
|
+
|
|
57
|
+
- テストデータ
|
|
58
|
+
```SQL
|
|
59
|
+
insert into follow(userid,follow_userid,statusid) values
|
|
60
|
+
(1,2,1),(1,3,1),(1,7,1),(1,8,1),(4,3,1),
|
|
61
|
+
(4,8,1),(4,7,1),(6,10,1),(7,10,2),(19,20,2),
|
|
62
|
+
(8,10,1),(2,3,1),(15,1,1),(14,1,1),(13,1,1),
|
|
63
|
+
(12,1,1),(11,1,2),(1,9,2),(3,20,2);
|
|
64
|
+
```
|
|
65
|
+
※ここまでは命題のまま
|
|
66
|
+
|
|
67
|
+
# test
|
|
68
|
+
- 仮にfollowからデータを削除します
|
|
69
|
+
```SQL
|
|
70
|
+
delete from follow where fid=3;
|
|
71
|
+
```
|
|
72
|
+
このときuser_relationテーブルを確認するとfid=3のデータが削除されていることがわかります
|
|
73
|
+
|
|
74
|
+
- userid,follow_useridをひっくり返して投入
|
|
75
|
+
テストとして(2,1,3)のデータを投入します
|
|
76
|
+
このデータはfid=1のデータのuserid,follow_useridがひっくり返っているものです
|
|
77
|
+
```SQL
|
|
78
|
+
insert into follow(userid,follow_userid,statusid) values(2,1,3);
|
|
79
|
+
```
|
|
80
|
+
user_relationを確認するとfid=1のデータが削除され、新たにケツにデータが2件登録されます
|
1
調整
answer
CHANGED
|
@@ -1,3 +1,6 @@
|
|
|
1
1
|
繰り返しになりますが、今のままでは相当効率がわるいです。
|
|
2
2
|
１レコードでユーザーidを２個管理するとどうしても主従関係が発生します
|
|
3
|
-
検索の際はunionして処理するのでパフォーマンスもSQLの視認性も期待できません。
|
|
3
|
+
検索の際はunionして処理するのでパフォーマンスもSQLの視認性も期待できません。
|
|
4
|
+
|
|
5
|
+
followテーブルからstatusを切り離し、triggerでデータ投入・削除時に
|
|
6
|
+
別テーブルで管理すると良いかもしれません
|