Print values of TABLE OF RECORDS TYPE without knowing the COLUMN NAMES of record

Question

I have created a Procedure which uses a View to PRINT (into CSV ), all the records of that view. Any user can drop and recreate that view, so its column structure can change

Is there any way to FETCH all records of VIEW into a collection and PRINT elements of that collection without knowing ALL COLUMN NAMES/ without any dependency on columns.

If there is any way, then it can be used when our table has hundreds of columns too. THIS QUESTION HAS BEEN ASKED TO ME MANY TIMES but I cannot find the answer to it and now i have to edit my procedure every time i need to spool a different table into csv file

The code I use normally for spooling takes all the column names:-

cursor c1 is
select * 
from table1 nvr;
  
TYPE alldata_typ IS table of c1%rowtype;
    v_dta       alldata_typ;
  
BEGIN
  
  v_fname  := 'swe_WRF_.csv';

    v_file   := UTL_FILE.FOPEN('CDMP_OUT_DIR', v_fname, 'W');

    v_header :=  'FSA_CODE,FSA_NAME,PROVINCE,MAKE,SERIES,REPORT_YEAR,VEHICLE_COUNT';

    UTL_FILE.PUT_LINE(v_file, v_header);
  
  OPEN c1;
  loop
    FETCH c1 BULK COLLECT

     INTO v_dta LIMIT v_fetch_limit;

     exit when v_dta.count=0 ;

     
    FOR i IN 1..v_dta.count 
    LOOP

        UTL_FILE.PUT_LINE(v_file, 
                            v_dta(i).fsa_code ||', '||
                            v_dta(i).fsa_name ||', '||
                            v_dta(i).province ||', '||
                            v_dta(i).make ||', '||
                            v_dta(i).series ||', '||
                            v_dta(i).report_year ||','||
                            v_dta(i).vehicle_count
                            );

        v_count := v_count + 1;
end loop;
;

Answer 1

If you are using UTL_FILE, then you are writing the file out on the database server. If it just CSV you are after, then you can run SQL*Plus commands directly from the database by using the scheduler.

More details here https://connor-mcdonald.com/2018/06/20/the-death-of-utl_file/

but in essence, the scheduler can run a SQL*Plus session, which also has support for a CSV markup on the output. It will probably be a lot quicker than UTL_FILE for larger data volumes.

Answer 2

You need use dynamic SQL to generate all column to csv. I rewrited your code as below. Hope it useful for you.

declare
  v_cmd        varchar(4000);
  c1           SYS_REFCURSOR;
  v_tab_name   varchar2(100) := upper('TABLE_NAME');
  v_owner_name varchar2(100) := upper('OWNER_NAME');
  type record_type is record(
    line_data varchar2(4000));
  TYPE alldata_typ IS table of record_type;
  v_dta         alldata_typ;
  v_fname       varchar2(20);
  V_HEADER      varchar2(20);
  V_FETCH_LIMIT integer := 100;
  V_COUNT       integer;
BEGIN

  v_fname := 'swe_WRF_.csv';

  v_file   := UTL_FILE.FOPEN('CDMP_OUT_DIR', v_fname, 'W');

  select listagg(COLUMN_NAME, ',') within group(order by COLUMN_ID)
    into v_header --create header with all column
    from all_tab_columns
   where table_name = v_tab_name
     and OWNER = v_owner_name;

  UTL_FILE.PUT_LINE(v_file, v_header);

  select 'select ' || listagg(COLUMN_NAME, '||'', ''||') within group(order by COLUMN_ID) || ' as line_data from ' || v_tab_name
    into v_cmd --generate select statement
    from all_tab_columns
   where table_name = v_tab_name
     and OWNER = v_owner_name;

  OPEN c1 for v_cmd;
  loop
    FETCH c1 BULK COLLECT
    
      INTO v_dta LIMIT v_fetch_limit;
  
    exit when v_dta.count = 0;
  
    FOR i IN 1 .. v_dta.count LOOP
    
      UTL_FILE.PUT_LINE(v_file, 
                          v_dta(i).line_data
                         );
      --dbms_output.put_line(v_dta(i).line_data);
      v_count := v_count + 1;
    end loop;
  END LOOP;
end;