PostgreSQL 速查表

所有工具

免费

可搜索、可打印的 PostgreSQL 参考——psql 命令、JSONB、数组、CTE、窗口函数、索引、角色和 EXPLAIN。免费。

psql meta-commands

psql -U user -d mydb -h localhost

Connect to a database

\l

List all databases

\c mydb

Switch to another database

\dt

List tables in current schema

\d users

Describe a table (columns, indexes)

\d+ users

Describe with size and storage

\dn

List schemas

\df

List functions

\du

List roles and their attributes

\x

Toggle expanded (vertical) output

\timing

Toggle query execution timing

\e

Edit the query buffer in $EDITOR

\i script.sql

Run SQL from a file

\copy users TO 'users.csv' CSV HEADER

Export a table to CSV

\q

Quit psql

Querying

SELECT * FROM users LIMIT 10;

First 10 rows

WHERE name ILIKE '%ann%'

Case-insensitive pattern match

WHERE name ~* '^a.*z$'

Regex match (case-insensitive)

ORDER BY score DESC NULLS LAST

Control where nulls sort

SELECT DISTINCT ON (user_id) * FROM orders ORDER BY user_id, created_at DESC;

Latest row per group

LIMIT 10 OFFSET 20

Paginate results

FETCH FIRST 5 ROWS ONLY

SQL-standard limit

SELECT ... FOR UPDATE SKIP LOCKED

Queue-style row locking

TABLESAMPLE SYSTEM (1)

Sample ~1% of a table

LATERAL (SELECT ... LIMIT 3)

Per-row correlated subquery

JSONB

data->'user'

Get JSON field (as jsonb)

data->>'name'

Get JSON field as text

data#>>'{user,address,city}'

Get nested value by path as text

data @> '{"active": true}'

Contains — matches subdocument

data ? 'email'

Key exists at top level

data ?| array['a', 'b']

Any of the keys exist

jsonb_set(data, '{name}', '"Sam"')

Update a value by path

data - 'legacy_field'

Delete a key

data || '{"active": true}'

Merge/overwrite fields

jsonb_array_elements(data->'items')

Expand array to rows

jsonb_each(data)

Expand object to key/value rows

jsonb_build_object('id', id, 'name', name)

Build JSON from columns

jsonb_agg(row_to_json(t))

Aggregate rows into a JSON array

jsonb_pretty(data)

Pretty-print for debugging

Arrays & ranges

tags TEXT[]

Declare an array column

ARRAY['a', 'b', 'c']

Array literal

tags[1]

Element access (1-based!)

WHERE 'php' = ANY(tags)

Array contains a value

WHERE tags @> ARRAY['php','sql']

Array contains all values

WHERE tags && ARRAY['php','go']

Arrays overlap (share any)

array_agg(name ORDER BY name)

Aggregate values into an array

unnest(tags)

Expand array to rows

array_length(tags, 1)

Number of elements

string_to_array('a,b', ',')

Split text into an array

period tstzrange

Timestamp range column

WHERE period @> now()

Range contains a value

EXCLUDE USING gist (room WITH =, period WITH &&)

No overlapping bookings

CTEs & window functions

WITH recent AS (SELECT ...) SELECT * FROM recent;

Common table expression

WITH RECURSIVE tree AS (... UNION ALL ...)

Walk hierarchies/graphs

WITH del AS (DELETE ... RETURNING *) INSERT ...

Data-modifying CTE

ROW_NUMBER() OVER (ORDER BY score DESC)

Sequential rank, no ties

RANK() / DENSE_RANK() OVER (...)

Rankings with tie handling

OVER (PARTITION BY user_id ORDER BY at)

Window per group

LAG(price) OVER (ORDER BY day)

Previous row's value

LEAD(price, 2) OVER (...)

Value 2 rows ahead

SUM(amount) OVER (ORDER BY day)

Running total

AVG(x) OVER (ROWS BETWEEN 6 PRECEDING AND CURRENT ROW)

7-row moving average

NTILE(4) OVER (ORDER BY score)

Split rows into quartiles

FIRST_VALUE(name) OVER (...)

First value in the window

Upsert & RETURNING

INSERT ... ON CONFLICT (email) DO UPDATE SET name = EXCLUDED.name;

Upsert on a unique column

ON CONFLICT DO NOTHING

Silently skip duplicates

ON CONFLICT ON CONSTRAINT users_pkey DO UPDATE ...

Target a named constraint

EXCLUDED.column

The value that failed to insert

DO UPDATE ... WHERE users.updated_at < EXCLUDED.updated_at

Conditional upsert

INSERT ... RETURNING id;

Get generated ids back

UPDATE ... RETURNING *;

Return the updated rows

DELETE ... RETURNING id, email;

Return what was deleted

MERGE INTO t USING src ON ... (PG 15+)

SQL-standard merge

Data types

id BIGINT GENERATED ALWAYS AS IDENTITY

Modern auto-increment (prefer)

SERIAL / BIGSERIAL

Legacy auto-increment (sequence)

UUID DEFAULT gen_random_uuid()

Native UUID with generator

TEXT

Unlimited string — no VARCHAR(n) needed

NUMERIC(10, 2)

Exact decimal (money math)

TIMESTAMPTZ

Timestamp with time zone (prefer)

INTERVAL '2 days 3 hours'

Duration type, date arithmetic

BOOLEAN

True/false/null (real type)

JSONB

Binary JSON — indexable, prefer over JSON

BYTEA

Raw binary data

INET / CIDR / MACADDR

Network address types

CREATE TYPE status AS ENUM ('new', 'done');

Custom enum type

col TEXT GENERATED ALWAYS AS (lower(email)) STORED

Generated column

Indexes

CREATE INDEX idx_users_email ON users (email);

Default B-tree index

CREATE UNIQUE INDEX ... ON users (lower(email));

Unique expression index

CREATE INDEX ... ON orders (user_id) WHERE status = 'open';

Partial index

CREATE INDEX ... USING gin (data);

GIN — jsonb/array containment

CREATE INDEX ... USING gin (to_tsvector('english', body));

Full-text search index

CREATE INDEX ... USING gist (location);

GiST — geometry, ranges, KNN

CREATE INDEX ... USING brin (created_at);

BRIN — huge append-only tables

CREATE INDEX CONCURRENTLY ...

Build without locking writes

CREATE INDEX ... ON t (a, b DESC);

Composite with sort direction

CREATE INDEX ... ON t (a) INCLUDE (b);

Covering index (index-only scan)

REINDEX INDEX CONCURRENTLY idx_name;

Rebuild a bloated index

DROP INDEX CONCURRENTLY idx_name;

Remove without blocking

Roles & privileges

CREATE ROLE app LOGIN PASSWORD 'secret';

Create a login role (user)

CREATE ROLE readonly NOLOGIN;

Create a group role

GRANT readonly TO app;

Add a role to a group

GRANT SELECT ON ALL TABLES IN SCHEMA public TO readonly;

Bulk grant reads

ALTER DEFAULT PRIVILEGES IN SCHEMA public GRANT SELECT ON TABLES TO readonly;

Auto-grant future tables

GRANT USAGE ON SCHEMA public TO app;

Allow schema access

REVOKE INSERT ON users FROM app;

Remove a privilege

ALTER ROLE app SET statement_timeout = '30s';

Per-role setting

ALTER TABLE users OWNER TO app;

Change object owner

ALTER TABLE docs ENABLE ROW LEVEL SECURITY;

Turn on row-level security

CREATE POLICY own ON docs USING (owner_id = current_user_id());

RLS policy

Maintenance & EXPLAIN

EXPLAIN SELECT ...;

Show the planned query plan

EXPLAIN (ANALYZE, BUFFERS) SELECT ...;

Run it and show real timings

VACUUM (VERBOSE) users;

Reclaim dead-row space

VACUUM FULL users;

Rewrite table (locks it!)

ANALYZE users;

Refresh planner statistics

SELECT * FROM pg_stat_activity;

List running queries

SELECT pg_cancel_backend(pid);

Cancel a running query

SELECT pg_terminate_backend(pid);

Kill a connection

SELECT pg_size_pretty(pg_total_relation_size('users'));

Table size incl. indexes

SELECT pg_size_pretty(pg_database_size('mydb'));

Whole database size

SELECT * FROM pg_stat_user_indexes;

Spot unused indexes

SELECT * FROM pg_locks WHERE NOT granted;

Find blocked lock waits

Backup & restore

pg_dump mydb > mydb.sql

Plain-SQL dump of a database

pg_dump -Fc mydb > mydb.dump

Custom format (compressed)

pg_dump -Fd mydb -j 4 -f dumpdir/

Directory format, 4 parallel jobs

pg_dump -t users -t orders mydb

Dump specific tables only

pg_dump --schema-only mydb

Structure without data

pg_dump --data-only mydb

Data without structure

psql mydb < mydb.sql

Restore a plain-SQL dump

pg_restore -d mydb -j 4 mydb.dump

Restore custom format in parallel

pg_restore -l mydb.dump

List dump contents

pg_restore -t users -d mydb mydb.dump

Restore a single table

pg_dumpall --globals-only > roles.sql

Back up roles and tablespaces

pg_basebackup -D /backup -Fp -Xs -P

Physical base backup (PITR)

Extensions

CREATE EXTENSION IF NOT EXISTS pgcrypto;

Hashing, encryption functions

CREATE EXTENSION "uuid-ossp";

uuid_generate_v4() and friends

CREATE EXTENSION pg_stat_statements;

Track query performance stats

CREATE EXTENSION pg_trgm;

Trigram fuzzy search + LIKE index

CREATE EXTENSION postgis;

Geospatial types and queries

CREATE EXTENSION vector;

pgvector — embeddings & ANN search

CREATE EXTENSION hstore;

Key/value pairs in a column

CREATE EXTENSION postgres_fdw;

Query remote Postgres servers

CREATE EXTENSION citext;

Case-insensitive text type

SELECT * FROM pg_available_extensions;

List installable extensions

\dx

List installed extensions (psql)

没有条目匹配“:q”。

关于 PostgreSQL 速查表

这份 PostgreSQL 速查表将你实际输入的 Postgres 命令汇集到一个可搜索的参考页面上：psql 元命令、查询、JSONB、数组和范围类型、CTE 和窗口函数、Upsert 和 RETURNING、数据类型、索引、角色和权限、维护和 EXPLAIN、备份和恢复，以及扩展。

它侧重于 Postgres 与其他数据库的不同之处——JSONB 运算符、数组和范围类型、带 RETURNING 的 ON CONFLICT upsert、GIN 和 GiST 索引，以及查询慢时你需要读懂的 EXPLAIN ANALYZE 输出——同时仍涵盖了日常的 SELECT、JOIN 和 DDL 语法。

本速查表免费且在客户端运行：使用搜索框实时过滤行，通过固定目录在章节间跳转，点击即可复制任何语句，还可以打印页面作为桌边参考。

如何使用 PostgreSQL 速查表

浏览各章节，从“psql 元命令”和“查询”到“索引”和“扩展”。
搜索关键词如 JSONB、ON CONFLICT 或 EXPLAIN 以实时过滤每一行。
需要存储和性能细节而非语法时跳转到“数据类型”或“索引”。
点击语句或其复制图标，将 SQL 复制到剪贴板。
打印本速查表获取离线 PostgreSQL 参考。

常见问题

十二个章节：psql 元命令、查询、JSONB、数组和范围类型、CTE 和窗口函数、Upsert 和 RETURNING、数据类型、索引、角色和权限、维护和 EXPLAIN、备份和恢复，以及扩展。

是的。专门的章节列出了 -> 和 ->> 访问器、@> 包含运算符、jsonb_set、jsonb_array_elements 以及使这些查询快速的 GIN 索引。

它们排在速查表的最前面——\l、\dt、\d、\du、\timing、\x 和其他你在写 SQL 之前就需要的反斜杠元命令。

可以。点击 SQL 代码片段或其行上的复制图标，即可立即复制到剪贴板。

是的，完全免费——可搜索、可打印，在浏览器中渲染，无需注册。

PostgreSQL 速查表

psql meta-commands

Querying

JSONB

Arrays & ranges

CTEs & window functions

Upsert & RETURNING

Data types

Indexes

Roles & privileges

Maintenance & EXPLAIN

Backup & restore

Extensions

关于 PostgreSQL 速查表

如何使用 PostgreSQL 速查表

常见问题

PostgreSQL 速查表包含哪些内容？

它涵盖 JSONB 吗？

包含 psql 命令吗？

可以一键复制查询吗？

PostgreSQL 参考是免费的吗？

相关工具

热门搜索

嵌入此工具