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 $$ declare chars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}'; result text := ''; i integer := 0; begin if length < 0 then raise exception 'Given length cannot be less than 0'; end if; for i in 1..length loop result := result || chars[1+random()*(array_length(chars, 1)-1)]; end loop; return result; end; $$ 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。
CREATE 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。
CREATE 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。
CREATE 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 $$ declare chars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}'; result text := ''; i integer := 0; begin if length < 0 then raise exception 'Given length cannot be less than 0'; end if; for i in 1..length loop result := result || chars[1+random()*(array_length(chars, 1)-1)]; end loop; return result; end; $$ 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
CREATE 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, name Recheck Cond: ((users.name)::text ~~ '%pf%'::text) Rows Removed by Index Recheck: 49807 Heap Blocks: exact=319 Buffers: 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=653 Planning time: 0.188 ms Execution time: 53.231 ms (11 rows)
以上为个人经验,希望能给大家一个参考,也希望大家多多支持。如有错误或未考虑完全的地方,望不吝赐教。
免责声明:本站文章均来自网站采集或用户投稿,网站不提供任何软件下载或自行开发的软件!
如有用户或公司发现本站内容信息存在侵权行为,请邮件告知! 858582#qq.com
暂无“PostgreSQL模糊匹配走索引的操作”评论...
RTX 5090要首发 性能要翻倍!三星展示GDDR7显存
三星在GTC上展示了专为下一代游戏GPU设计的GDDR7内存。
首次推出的GDDR7内存模块密度为16GB,每个模块容量为2GB。其速度预设为32 Gbps(PAM3),但也可以降至28 Gbps,以提高产量和初始阶段的整体性能和成本效益。
据三星表示,GDDR7内存的能效将提高20%,同时工作电压仅为1.1V,低于标准的1.2V。通过采用更新的封装材料和优化的电路设计,使得在高速运行时的发热量降低,GDDR7的热阻比GDDR6降低了70%。
更新动态
2024年11月23日
2024年11月23日
- 凤飞飞《我们的主题曲》飞跃制作[正版原抓WAV+CUE]
- 刘嘉亮《亮情歌2》[WAV+CUE][1G]
- 红馆40·谭咏麟《歌者恋歌浓情30年演唱会》3CD[低速原抓WAV+CUE][1.8G]
- 刘纬武《睡眠宝宝竖琴童谣 吉卜力工作室 白噪音安抚》[320K/MP3][193.25MB]
- 【轻音乐】曼托凡尼乐团《精选辑》2CD.1998[FLAC+CUE整轨]
- 邝美云《心中有爱》1989年香港DMIJP版1MTO东芝首版[WAV+CUE]
- 群星《情叹-发烧女声DSD》天籁女声发烧碟[WAV+CUE]
- 刘纬武《睡眠宝宝竖琴童谣 吉卜力工作室 白噪音安抚》[FLAC/分轨][748.03MB]
- 理想混蛋《Origin Sessions》[320K/MP3][37.47MB]
- 公馆青少年《我其实一点都不酷》[320K/MP3][78.78MB]
- 群星《情叹-发烧男声DSD》最值得珍藏的完美男声[WAV+CUE]
- 群星《国韵飘香·贵妃醉酒HQCD黑胶王》2CD[WAV]
- 卫兰《DAUGHTER》【低速原抓WAV+CUE】
- 公馆青少年《我其实一点都不酷》[FLAC/分轨][398.22MB]
- ZWEI《迟暮的花 (Explicit)》[320K/MP3][57.16MB]