【greenplum】 获取表结构，实现类似mysql show create table 功能

转载于：http://www.cnblogs.com/gomysql/p/6007013.html

    大家都知道在MySQL里面是非常easy的，show create table table_name 就搞定了，在gpdb里面就没这么容易，在查询大量资料以后终于找到了方法。那就是自己定义一个函数去获取，函数中可以嵌套python代码，非常的方便。但是资料中的代码有大量错误，在经过几番调试以后终于可以使用了。

1. 创建语言

CREATE PROCEDURAL LANGUAGE plpythonu; -- 已经创建了可忽略

2. 创建函数（代码如下）：

create or replace function get_table_structure(tablename text)    # 表名格式为 tableschema.table_name
    returns text
as $$
    try:
        table_name = tablename.lower().split('.')[1]
        talbe_schema=tablename.lower().split('.')[0]
    except (IndexError):
        return 'Please in put "tableschema.table_name"'
    get_table_oid="select oid,reloptions,relkind from pg_class where oid='%s'::regclass"%(tablename)
    try:
        rv_oid=plpy.execute(get_table_oid,5)      #plpy.execute(query [, max-rows])
        if not rv_oid:
            return 'Did not find any relation named"'+tablename +'".'
    except (Error):
        return 'Did not find any relation named"'+tablename +'".'
    table_oid=rv_oid[0]['oid']
    rv_reloptions=rv_oid[0]['reloptions'] # 访问方法特定的选项，使用"keyword=value"格式的字符串
    rv_relkind=rv_oid[0]['relkind']
    create_sql="";
    table_kind='table';
    if rv_relkind !='r' and rv_relkind !='v':  # r:普通表， v:视图， c:复合类型， t:表， o:内部 AO 节点文件， 
        plpy.error('%s is not table or view'%(tablename));
    elif rv_relkind=='v':
        get_view_def="select pg_get_viewdef(%s,'t') as viewdef;" % (table_oid)
        rv_viewdef=plpy.execute(get_view_def);
        create_sql='create view %s as \n' % (tablename)
        create_sql += rv_viewdef[0]['viewdef']+'\n';
        table_kind='view'
    else:
       #获取行
      get_columns="select a.attname,pg_catalog.format_type(a.atttypid,a.atttypmod),\
       (select substring(pg_catalog.pg_get_expr(d.adbin,d.adrelid) for 128) \
        from pg_catalog.pg_attrdef d where d.adrelid=a.attrelid and d.adnum=a.attnum and a.atthasdef) \
        as default,a.attnotnull as isnull from pg_catalog.pg_attribute \
        a where a.attrelid= %s and a.attnum >0 and not a.attisdropped order by a.attnum;" % (table_oid);
        rv_columns=plpy.execute(get_columns)
        
        #获取分布键
        get_table_distribution1="select attrnums from pg_catalog.gp_distribution_policy t where localoid = '" + table_oid + "' "
        rv_distribution1=plpy.execute(get_table_distribution1,500)
        rv_distribution2=''
        if rv_distribution1 and rv_distribution1[0]['attrnums']:
            get_table_distribution2="select attname from pg_attribute where attrelid='"+table_oid+"' and attnum in (" + str(rv_distribution1[0]['attrnums']).strip('{').strip('}').strip('[').strip(']')+")"
            rv_distribution2=plpy.execute(get_table_distribution2,500)
       
       
        create_sql='create table %s (\n' % (tablename)
        #获取索引
        get_index="select pg_get_indexdef(indexrelid) as indexdef from pg_index where indrelid=%s" % (table_oid);
        rv_index=plpy.execute(get_index);

        
        get_parinfo1="select attname as columnname from pg_attribute where attnum =(select paratts[0] from pg_partition where parrelid=%s) and attrelid=%s;"%(table_oid,table_oid);
        get_parinfo2=""" select pp.parrelid,prl.parchildrelid,case when pp.parkind='h'::"char" then 'hash'::text when pp.parkind='r'::"char" then 'range'::text when pp.parkind='l'::"char" then 'list'::text else null::text end as partitiontype,pg_get_partition_rule_def(prl.oid,true) as partitionboundary from pg_partition pp,pg_partition_rule prl where pp.paristemplate=false and pp.parrelid = %s and prl.paroid = pp.oid order by prl.parname; """ % (table_oid)
        v_par_parent=plpy.execute(get_parinfo1);
        v_par_info=plpy.execute(get_parinfo2);
        max_column_len=10
        max_type_len=4
        max_modifiers_len=4
        max_default_len=4
        for i in rv_columns:
            if i['attname']:
                if max_column_len < i['attname'].__len__():
                    max_column_len=i['attname'].__len__()
            if i['format_type']:
                if max_type_len < i['format_type'].__len__():
                    max_type_len=i['format_type'].__len__()
            if i['default']:
                if max_type_len < i['default'].__len__():
                    max_default_len=i['default'].__len__()
        first=True
        for i in rv_columns:
            if first==True:
                split_char=' ';
                first=False
            else:
                split_char=',';
            if i['attname']:
                create_sql += " " + split_char + i['attname'].ljust(max_column_len+6)+''
            else:
                create_sql += "" + split_char + ' '.ljust(max_column_len+6)
            if i['format_type']:
                create_sql += ' ' + i['format_type'].ljust(max_type_len +2)
            else:
                create_sql += ' ' + ' '.ljust(max_type_len+2)
            if i['isnull'] and i['isnull']:
                create_sql += ' ' + ' not null '.ljust(8)
            if i['default']:
                create_sql += ' default ' + i['default'].ljust(max_default_len+6)
            create_sql += "\n"
        create_sql += ")"
 
        if rv_reloptions:
            create_sql +=" with ("+str(rv_reloptions).strip('{').strip('}').strip('[').strip(']') +")\n"
            create_sql = create_sql.replace("'",'')
        if rv_distribution2:
            create_sql += 'Distributed by ('
            for i in rv_distribution2:
                create_sql += i['attname'] + ','
            create_sql =create_sql.strip(',')+')'
        elif rv_distribution1:
            create_sql += 'Distributed randomly\n'
        if v_par_parent:
            partitiontype=v_par_info[0]['partitiontype'];
            create_sql +='\nPARTITION BY '+ partitiontype + "("+v_par_parent[0]['columnname']+")\n(\n";
            for i in v_par_info:
                create_sql +=" " +i['partitionboundary']+',\n';
            create_sql=create_sql.strip(',\n');
            create_sql+="\n)"
        create_sql+=";\n\n"
        for i in rv_index:
            create_sql += i['indexdef']+';\n'

        
        get_table_comment="select 'comment on %s %s is '''|| COALESCE (description,'')|| '''' as comment from pg_description where objoid=%s and objsubid=0;" % (table_kind,tablename,table_oid)
        get_column_comment="select 'comment on column %s.'||b.attname ||' is ''' || COALESCE(a.description,'')|| ''' ' as comment from pg_catalog.pg_description a,pg_catalog.pg_attribute b where objoid=%s and a.objoid=b.attrelid and a.objsubid=b.attnum;" % (tablename,table_oid)
        rv_table_comment=plpy.execute(get_table_comment);
        rv_column_comment=plpy.execute(get_column_comment);


        for i in rv_table_comment:
            create_sql += i['comment']+';\n'
        for i in rv_column_comment:
            create_sql +=i['comment']+';\n'
        return create_sql;

$$ LANGUAGE plpythonu;

3. 进行测试

testdb=# SELECT get_table_structure('public.tb1_partition_range_yyyymmdd');
                                                get_table_structure                                                
-------------------------------------------------------------------------------------------------------------------
 create table public.tb1_partition_range_yyyymmdd (                                                                
   id               numeric                                                                                        
  ,yyyymmdd         date                                                                                           
 ) with (appendonly=true, compresslevel=5)                                                                         
 Distributed by (id)                                                                                               
 PARTITION BY range(yyyymmdd)                                                                                      
 (                                                                                                                 
  PARTITION p20120811 START ('2012-08-11'::date) END ('2012-08-12'::date) WITH (appendonly=true, compresslevel=5), 
  PARTITION p20120812 START ('2012-08-12'::date) END ('2012-08-13'::date) WITH (appendonly=true, compresslevel=5)  
 );                                                                                                                
                                                                                                                   
 CREATE INDEX idx_yyyymmdd ON tb1_partition_range_yyyymmdd USING btree (yyyymmdd);                                 
 
(1 row)