Guess table definition. The function will create sql create table statement from sample data.

gtd

create or replace function gtd(sample text)
   returns varchar as $$
   declare outsql varchar;
begin
select tbl_definition into outsql from (

with 

source as (select $1::varchar instr)

,elems as (

select elem[en], rn, en
   from (
      select elem, rn, generate_subscripts(elem,1) en
         from(
            select string_to_array(arr[rn],';') as elem, rn
               from  (
                  select *, generate_subscripts(arr, 1) as rn
                     from  (
                        select regexp_split_to_array(instr, E'\\n') as arr from source
                           ) w
                     ) x
            ) y
         ) z
)

, type_tests as (

select
   *,
   case
      when elem ~ '^$' then 'empty'
      when elem ~ '^[-+]?[0-9]+$' then 'integer'
      when elem ~ '^[-+]?[0-9]*\.?[0-9]*([eE][-+]?[0-9]+)?$' then 'numeric'
      when elem ~ '^[0-9]{4}-[0-9]{1,2}-[0-9]{1,2} [0-9]{1,2}:[0-9]{1,2}:[0-9]{1,2}(\.[0-9]*)?$' or
           elem ~ '^[0-9]{4}/[0-9]{1,2}/[0-9]{1,2} [0-9]{1,2}:[0-9]{1,2}:[0-9]{1,2}(\.[0-9]*)?$' then 'timestamp'
      when elem ~ '^[0-9]{1,2}:[0-9]{1,2}(:[0-9]{1,2}(\.[0-9]+)?)?$' then 'time'
      when elem ~ '^[0-9]{4}/[0-9]{1,2}/[0-9]{1,2}$' or elem ~ '^[0-9]{4}-[0-9]{1,2}-[0-9]{1,2}$' then 'date'
      else 'varchar' end as dtype 
from 
   elems
)

, results_agg as (
   select
      tt2.en,
      tt2.elem,
      count(*),
      array_agg(tt.dtype order by tt.rn) as dtypes,
      array_agg(tt.elem order by tt.rn) tmp
   from
      type_tests tt
      join type_tests tt2 on (tt.en = tt2.en and tt2.rn=1 and tt.rn <> 1)
   group by
      tt2.en,
      tt2.elem
   order by
      tt2.en   
      )

, type_determination as (
   select 
      *,
      case
         when dtypes @> array['integer']    and not(dtypes && array['numeric','date','time','timestamp','varchar']) then 'integer'
         when dtypes @> array['timestamp']  and not(dtypes && array['numeric','time','varchar']) then 'timestamp'
         when dtypes @> array['date']       and not(dtypes && array['numeric','time','varchar']) then 'date'
         when dtypes @> array['time']       and not(dtypes && array['numeric','date','varchar','timestamp']) then 'time'
         when dtypes @> array['numeric']    and not(dtypes && array['date','time','timestamp','varchar']) then 'numeric'
         else 'varchar' end as result_type
   from 
      results_agg
)


select
   E'create table (\n' || array_to_string(array_agg(E'\t' || replace(lower(elem),' ','_') || 
                  ' ' || result_type order by en),E',\n') || E'\n);' as tbl_definition
from
   type_determination
) a;

        return outsql;
end;
$$ LANGUAGE plpgsql
;

select gtd('a a;b;c c c;d;e;f;g;h;i i;j;k
2012-01-01 13:04:04.321;test;2012-01-01;2012/01/01;abcd;321;-324;.1;-5.5;0;14:54
2012/01/01 13:04:04.321;test;2012-01-01;2012/01/01;abcd;321;-324;.1;-5;0;14:54
2012-01-01 13:04:04.321;test;2012-01-01;2012/01/01;.1;321;-324;.1;-5.5;0;14:54
2012-01-01 13:04:04.321;test;2012-01-01;2012/01/01;abcd;321;-324;.1;-5.5;0;14:54
2012-01-01 13:04:04.321;;2012-01-01;2012/01/01;1;321;-324;.1;-5.5;0;14:54');