PostgreSQL模糊匹配走索引的操作
项目招商找A5 快速获取精准代理名单
这篇文章主要介绍了PostgreSQL模糊匹配走索引的操作，具有很好的参考价值，希望对大家有所帮助。一起跟随小编过来看看吧。
场景 lower(name) like 'pf%'
create table users (id int primary key, name varchar(255));Create or replace function random_string(length integer) returns text as$$declarechars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}';result text := '';i integer := 0;beginif length < 0 thenraise exception 'Given length cannot be less than 0';end if;for i in 1..length loopresult := result || chars[1+random()*(array_length(chars, 1)-1)];end loop;return result;end;$$ language plpgsql;insert into users values(generate_series(1,50000), random_string(15));
普通bt：不走索引
pg_trgm模块提供函数和操作符测定字母数字文本基于三元模型匹配的相似性，还有支持快速搜索相似字符串的索引操作符类。三元模型是一组从一个字符串中获得的三个连续的字符。我们可以通过计数两个字符串共享的三元模型的数量来测量它们的相似性。这个简单的想法证明在测量许多自然语言词汇的相似性时是非常有效的。
1CREATE INDEX users_idx0 ON users (name);
全字匹配查询（走索引）
explain select * from users where name='pfDNQVmhqDrF1EY'; QUERY PLAN-------------------------------------------------------------------------Index Scan using users_idx0 on users (cost=0.29..8.31 rows=1 width=20)Index Cond: ((name)::text = 'pfDNQVmhqDrF1EY'::text)(2 rows)
加函数全字匹配（不走索引）
explain select * from users where lower(name)='pfDNQVmhqDrF1EY'; QUERY PLAN-----------------------------------------------------------Seq Scan on users (cost=0.00..1069.00 rows=250 width=20)Filter: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)(2 rows)
模糊匹配（不走索引）
explain select * from users where name like 'pf%'; QUERY PLAN--------------------------------------------------------Seq Scan on users (cost=0.00..944.00 rows=5 width=20)Filter: ((name)::text ~~ 'pf%'::text)
explain select * from users where name like 'pf_'; QUERY PLAN--------------------------------------------------------Seq Scan on users (cost=0.00..944.00 rows=5 width=20)Filter: ((name)::text ~~ 'pf_'::text)
字段带函数的bt索引：函数走索引
drop index users_idx0;CREATE INDEX users_dex1 ON users (lower(name));
加函数全字匹配（走索引）
explain select * from users where lower(name)='pfDNQVmhqDrF1EY'; QUERY PLAN---------------------------------------------------------------------------Bitmap Heap Scan on users (cost=6.23..324.34 rows=250 width=20)Recheck Cond: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)-> Bitmap Index Scan on users_dex1 (cost=0.00..6.17 rows=250 width=0) Index Cond: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)(4 rows)
模糊匹配（不走索引）
explain select * from users where lower(name) like 'pf%'; QUERY PLAN-----------------------------------------------------------Seq Scan on users (cost=0.00..1069.00 rows=250 width=20)Filter: (lower((name)::text) ~~ 'pf%'::text)(2 rows)
声明操作符类的bt索引：like走索引
定义索引的同时可以为索引的每个字段声明一个操作符类。
1CREATE INDEX name ON table (column opclass [sort options] [, …]);
这个操作符类指明该索引用于该字段时要使用的操作符。
1CREATE INDEX users_dex2 ON users (lower(name) varchar_pattern_ops);
模糊匹配（走索引）
explain select * from users where lower(name) like 'pf%'; QUERY PLAN------------------------------------------------------------------------------------------------------Bitmap Heap Scan on users (cost=4.82..144.00 rows=5 width=20)Filter: (lower((name)::text) ~~ 'pf%'::text)-> Bitmap Index Scan on users_dex2 (cost=0.00..4.82 rows=53 width=0) Index Cond: ((lower((name)::text) ~>=~ 'pf'::text) AND (lower((name)::text) ~<~ 'pg'::text))(4 rows)
场景2 name like '%pf%'
Create or replace function random_string(length integer) returns text as$$declarechars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}';result text := '';i integer := 0;beginif length < 0 thenraise exception 'Given length cannot be less than 0';end if;for i in 1..length loopresult := result || chars[1+random()*(array_length(chars, 1)-1)];end loop;return result;end;$$ language plpgsql;create table users (id int primary key, name varchar(255));insert into users values(generate_series(1,50000), random_string(15));
声明操作符bt：不走索引
1CREATE INDEX idx_name ON users USING btree (lower(name) varchar_pattern_ops);
explain (analyze true,format yaml, verbose true, buffers true) select * from users where lower(name) like '%pf%';\ QUERY PLAN------------------------------------------------------------ Plan: + Node Type: "Seq Scan" + Parallel Aware: false + Relation Name: "users" + Schema: "public" + Alias: "users" + Startup Cost: 0.00 + Total Cost: 1069.00 + Plan Rows: 5 + Plan Width: 20 + Actual Startup Time: 0.320 + Actual Total Time: 86.841 + Actual Rows: 710 + Actual Loops: 1 + Output: + - "id" + - "name" + Filter: "(lower((users.name)::text) ~~ '%pf%'::text)"+ Rows Removed by Filter: 49290 + Shared Hit Blocks: 319 + Shared Read Blocks: 0 + Shared Dirtied Blocks: 0 + Shared Written Blocks: 0 + Local Hit Blocks: 0 + Local Read Blocks: 0 + Local Dirtied Blocks: 0 + Local Written Blocks: 0 + Temp Read Blocks: 0 + Temp Written Blocks: 0 +Planning Time: 0.188 +Triggers: +Execution Time: 86.975
声明pg_trgm操作符bt：可以走索引
CREATE EXTENSION pg_trgm;CREATE INDEX idx_users_name_trgm_gist ON users USING gist (name gist_trgm_ops);
explain (analyze true, verbose true, buffers true) select * from users where name like '%pf%'; QUERY PLAN------------------------------------------------------------------------------------------------------------------------------------------Bitmap Heap Scan on public.users (cost=32.19..371.08 rows=505 width=20) (actual time=19.314..53.132 rows=193 loops=1)Output: id, nameRecheck Cond: ((users.name)::text ~~ '%pf%'::text)Rows Removed by Index Recheck: 49807Heap Blocks: exact=319Buffers: shared hit=972-> Bitmap Index Scan on idx_users_name_trgm_gist (cost=0.00..32.06 rows=505 width=0) (actual time=19.175..19.175 rows=50000 loops=1) Index Cond: ((users.name)::text ~~ '%pf%'::text) Buffers: shared hit=653Planning time: 0.188 msExecution time: 53.231 ms(11 rows)
文章来源：脚本之家
来源地址：https://www.jb51.net/article/204764.htm
尊敬的看官您对PostgreSQL模糊匹配走索引的操作有什么看法呢？互联网品牌制作专家愿与您共同探讨！版权声明：本文内容由互联网用户自发贡献，该文观点仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容， 请加微信号oem365 举报，一经查实，本站将立刻删除。上一篇: 什么是SEO数据化SEO优化技术数据化有价值吗 返 回 下一篇:PostgreSQL忘记postgres账号密码的解决方法
