从linux命令行在postgresql中为一个表生成create table语句:

为演示创建一个表:

CREATE TABLE your_table(
    thekey   integer NOT NULL,  
    ticker   character varying(10) NOT NULL,
    date_val date,
    open_val numeric(10,4) NOT NULL
); 

Pg_dump手动，可以输出表创建PSQL语句:

pg_dump -U your_user your_database -t your_table --schema-only

打印:

-- pre-requisite database and table configuration omitted
CREATE TABLE your_table (
    thekey integer NOT NULL, 
    ticker character varying(10) NOT NULL, 
    date_val date,  
    open_val numeric(10,4) NOT NULL 
); 
-- post-requisite database and table configuration omitted
  

解释:

Pg_dump帮助我们获得关于数据库本身的信息。-U代表用户名。我的pgadmin用户没有设置密码，所以我不需要输入密码。-t选项表示指定一个表。——schema-only表示只打印关于表的数据，而不是表中的数据。

pg_dump是优秀的C代码，它试图很好地处理不断发展的sql标准，并处理在postgresql的查询语言和它在磁盘上的表示之间产生的成千上万个细节。如果你想卷自己的“psql磁盘创建语句”的安排，你是龙:https://doxygen.postgresql.org/pg__dump_8c_source.html

另一个绕过pg_dump的选项是在创建表时保存表创建SQL语句。把它放在安全的地方，需要的时候拿来。

或者使用SQL从postgresql中获取表名、列名和数据类型信息:

CREATE TABLE your_table(  thekey integer NOT NULL,
                          ticker character varying(10) NOT NULL,
                          date_val date,
                          open_val numeric(10,4) NOT NULL
); 

SELECT table_name, column_name, data_type 
FROM information_schema.columns 
WHERE table_name = 'your_table'; 

打印:

┌────────────┬─────────────┬───────────────────┐ 
│ table_name │ column_name │     data_type     │ 
├────────────┼─────────────┼───────────────────┤ 
│ your_table │ thekey      │ integer           │ 
│ your_table │ ticker      │ character varying │ 
│ your_table │ date_val    │ date              │ 
│ your_table │ open_val    │ numeric           │ 
└────────────┴─────────────┴───────────────────┘ 

我能想到的最简单的方法是安装pgAdmin 3(在这里找到)，并用它来查看数据库。它将自动生成一个查询，该查询将创建所讨论的表。

pg_dump -t 'schema-name.table-name' --schema-only database-name

更多信息-在手册。

这是这个老问题的另一个解决方案。多年来，关于这个问题有许多优秀的答案，我的尝试在很大程度上借鉴了它们。

我使用安德烈·列别登科的解决方案作为起点，因为它的输出已经非常接近我的要求。

特点:

following common practice I have moved the foreign key constraints outside the table definition. They are now included as ALTER TABLE statements at the bottom. The reason is that a foreign key can also link to a column of the same table. In that fringe case the constraint can only be created after the table creation is completed. The create table statement would throw an error otherwise. The layout and indenting looks nicer now (at least to my eye) Drop command (commented out) in the header of the definition The solution is offered here as a plpgsql function. The algorithm does however not use any procedural language. The function just wraps one single query that can be used in a pure sql context as well. removed redundant subqueries Identifiers are now quoted if they are identical to reserved postgresql language elements replaced the string concatenation operator || with the appropriate string functions to improve performance, security and readability of the code. Note: the || operator produces NULL if one of the combined strings is NULL. It should only be used when that is the desired behaviour. (check out the usage in the code below for an example)

CREATE OR REPLACE FUNCTION public.wmv_get_table_definition (
    p_schema_name character varying,
    p_table_name character varying
)
    RETURNS SETOF TEXT
    AS $BODY$
BEGIN
    RETURN query 
    WITH table_rec AS (
        SELECT
            c.relname, n.nspname, c.oid
        FROM
            pg_catalog.pg_class c
            LEFT JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace
        WHERE
            relkind = 'r'
            AND n.nspname = p_schema_name
            AND c.relname LIKE p_table_name
        ORDER BY
            c.relname
    ),
    col_rec AS (
        SELECT
            a.attname AS colname,
            pg_catalog.format_type(a.atttypid, a.atttypmod) AS coltype,
            a.attrelid AS oid,
            ' DEFAULT ' || (
                SELECT
                    pg_catalog.pg_get_expr(d.adbin, d.adrelid)
                FROM
                    pg_catalog.pg_attrdef d
                WHERE
                    d.adrelid = a.attrelid
                    AND d.adnum = a.attnum
                    AND a.atthasdef) AS column_default_value,
            CASE WHEN a.attnotnull = TRUE THEN
                'NOT NULL'
            ELSE
                'NULL'
            END AS column_not_null,
            a.attnum AS attnum
        FROM
            pg_catalog.pg_attribute a
        WHERE
            a.attnum > 0
            AND NOT a.attisdropped
        ORDER BY
            a.attnum
    ),
    con_rec AS (
        SELECT
            conrelid::regclass::text AS relname,
            n.nspname,
            conname,
            pg_get_constraintdef(c.oid) AS condef,
            contype,
            conrelid AS oid
        FROM
            pg_constraint c
            JOIN pg_namespace n ON n.oid = c.connamespace
    ),
    glue AS (
        SELECT
            format( E'-- %1$I.%2$I definition\n\n-- Drop table\n\n-- DROP TABLE IF EXISTS %1$I.%2$I\n\nCREATE TABLE %1$I.%2$I (\n', table_rec.nspname, table_rec.relname) AS top,
            format( E'\n);\n\n\n-- adempiere.wmv_ghgaudit foreign keys\n\n', table_rec.nspname, table_rec.relname) AS bottom,
            oid
        FROM
            table_rec
    ),
    cols AS (
        SELECT
            string_agg(format('    %I %s%s %s', colname, coltype, column_default_value, column_not_null), E',\n') AS lines,
            oid
        FROM
            col_rec
        GROUP BY
            oid
    ),
    constrnt AS (
        SELECT
            string_agg(format('    CONSTRAINT %s %s', con_rec.conname, con_rec.condef), E',\n') AS lines,
            oid
        FROM
            con_rec
        WHERE
            contype <> 'f'
        GROUP BY
            oid
    ),
    frnkey AS (
        SELECT
            string_agg(format('ALTER TABLE %I.%I ADD CONSTRAINT %s %s', nspname, relname, conname, condef), E';\n') AS lines,
            oid
        FROM
            con_rec
        WHERE
            contype = 'f'
        GROUP BY
            oid
    )
    SELECT
        concat(glue.top, cols.lines, E',\n', constrnt.lines, glue.bottom, frnkey.lines, ';')
    FROM
        glue
        JOIN cols ON cols.oid = glue.oid
        LEFT JOIN constrnt ON constrnt.oid = glue.oid
        LEFT JOIN frnkey ON frnkey.oid = glue.oid;
END;
$BODY$
LANGUAGE plpgsql;

与前面提到的其他答案一样，没有内置函数可以做到这一点。

下面是一个函数，它试图获取复制表或比较ddl中部署和检查所需的所有信息。

这个函数输出:

列(w/ precision, null/not-null，默认值) 约束 索引

CREATE OR REPLACE FUNCTION public.show_create_table(
  in_schema_name varchar,
  in_table_name varchar
)
RETURNS text
LANGUAGE plpgsql VOLATILE
AS
$$
  DECLARE
    -- the ddl we're building
    v_table_ddl text;

    -- data about the target table
    v_table_oid int;

    -- records for looping
    v_column_record record;
    v_constraint_record record;
    v_index_record record;
  BEGIN
    -- grab the oid of the table; https://www.postgresql.org/docs/8.3/catalog-pg-class.html
    SELECT c.oid INTO v_table_oid
    FROM pg_catalog.pg_class c
    LEFT JOIN pg_catalog.pg_namespace n ON n.oid = c.relnamespace
    WHERE 1=1
      AND c.relkind = 'r' -- r = ordinary table; https://www.postgresql.org/docs/9.3/catalog-pg-class.html
      AND c.relname = in_table_name -- the table name
      AND n.nspname = in_schema_name; -- the schema

    -- throw an error if table was not found
    IF (v_table_oid IS NULL) THEN
      RAISE EXCEPTION 'table does not exist';
    END IF;

    -- start the create definition
    v_table_ddl := 'CREATE TABLE ' || in_schema_name || '.' || in_table_name || ' (' || E'\n';

    -- define all of the columns in the table; https://stackoverflow.com/a/8153081/3068233
    FOR v_column_record IN
      SELECT
        c.column_name,
        c.data_type,
        c.character_maximum_length,
        c.is_nullable,
        c.column_default
      FROM information_schema.columns c
      WHERE (table_schema, table_name) = (in_schema_name, in_table_name)
      ORDER BY ordinal_position
    LOOP
      v_table_ddl := v_table_ddl || '  ' -- note: two char spacer to start, to indent the column
        || v_column_record.column_name || ' '
        || v_column_record.data_type || CASE WHEN v_column_record.character_maximum_length IS NOT NULL THEN ('(' || v_column_record.character_maximum_length || ')') ELSE '' END || ' '
        || CASE WHEN v_column_record.is_nullable = 'NO' THEN 'NOT NULL' ELSE 'NULL' END
        || CASE WHEN v_column_record.column_default IS NOT null THEN (' DEFAULT ' || v_column_record.column_default) ELSE '' END
        || ',' || E'\n';
    END LOOP;

    -- define all the constraints in the; https://www.postgresql.org/docs/9.1/catalog-pg-constraint.html && https://dba.stackexchange.com/a/214877/75296
    FOR v_constraint_record IN
      SELECT
        con.conname as constraint_name,
        con.contype as constraint_type,
        CASE
          WHEN con.contype = 'p' THEN 1 -- primary key constraint
          WHEN con.contype = 'u' THEN 2 -- unique constraint
          WHEN con.contype = 'f' THEN 3 -- foreign key constraint
          WHEN con.contype = 'c' THEN 4
          ELSE 5
        END as type_rank,
        pg_get_constraintdef(con.oid) as constraint_definition
      FROM pg_catalog.pg_constraint con
      JOIN pg_catalog.pg_class rel ON rel.oid = con.conrelid
      JOIN pg_catalog.pg_namespace nsp ON nsp.oid = connamespace
      WHERE nsp.nspname = in_schema_name
      AND rel.relname = in_table_name
      ORDER BY type_rank
    LOOP
      v_table_ddl := v_table_ddl || '  ' -- note: two char spacer to start, to indent the column
        || 'CONSTRAINT' || ' '
        || v_constraint_record.constraint_name || ' '
        || v_constraint_record.constraint_definition
        || ',' || E'\n';
    END LOOP;

    -- drop the last comma before ending the create statement
    v_table_ddl = substr(v_table_ddl, 0, length(v_table_ddl) - 1) || E'\n';

    -- end the create definition
    v_table_ddl := v_table_ddl || ');' || E'\n';

    -- suffix create statement with all of the indexes on the table
    FOR v_index_record IN
      SELECT indexdef
      FROM pg_indexes
      WHERE (schemaname, tablename) = (in_schema_name, in_table_name)
    LOOP
      v_table_ddl := v_table_ddl
        || v_index_record.indexdef
        || ';' || E'\n';
    END LOOP;

    -- return the ddl
    RETURN v_table_ddl;
  END;
$$;

例子

SELECT * FROM public.show_create_table('public', 'example_table');

生产

CREATE TABLE public.example_table (
  id bigint NOT NULL DEFAULT nextval('test_tb_for_show_create_on_id_seq'::regclass),
  name character varying(150) NULL,
  level character varying(50) NULL,
  description text NOT NULL DEFAULT 'hello there!'::text,
  CONSTRAINT test_tb_for_show_create_on_pkey PRIMARY KEY (id),
  CONSTRAINT test_tb_for_show_create_on_level_check CHECK (((level)::text = ANY ((ARRAY['info'::character varying, 'warn'::character varying, 'error'::character varying])::text[])))
);
CREATE UNIQUE INDEX test_tb_for_show_create_on_pkey ON public.test_tb_for_show_create_on USING btree (id);

使用它并在ddl中获得输出。出文件

~/bin/pg_dump -p 30000 -d <db_name> -U <db_user> --schema=<schema_name> -t <table_name> --schema-only >> /tmp/ddl.out

因此这将在路径:/tmp/ DDL .out中生成DDL