Complicated pivot example

Question

I work in the transportation industry. We haul a lot of product in a given day from a variety of sources. My bosses would like a report that shows the origins in columns with row data containing the date and number of loads hauled based on the destination. We shift from origin to origin on a daily basis, so I will have some origins that only have a couple of days worth of loads and others that will show loads from several weeks in a row.
From everything I can see, I'm going to need to dynamically create a report that first finds all the origins that were hauled from in a given date range, then generates an aggregating report that shows loads on a given day.
Has anyone got an example of something like this they could point me to?
Here's a visual example of what I'm after:
DATE Origin1 Origin2 Origin3 Origin5 Origin8 ...
9/1/2009 10 2 - - -
9/2/2009 12 16 - - -
9/3/2009 - 15 - - 10
9/4/2009 - - 54 - 10
9/5/2009 - 10 - 15 12
...

Answer 1

Okay, I've come back to this issue and I had an idea that was starting to work until I ran into some problems with sheer size. From various documentation sources, it looks like the largest variable I can declare is a VARCHAR2(4000). I figured out a way to dynamically build the SQL statement I need in PL/SQL, but my estimates for the amount of space I need are coming up significantly short.
What I did was build a procedure that attempts to tell me how many columns I need, then loops through the columns and starts generating the SQL in pieces. At the end, my intention was to concatenate the SELECT, FROM, WHERE, and ORDER BY clauses together into one query, then execute the query and see my nice (but very wide and huge) report. I counted the characters I'm going to need, and if I handle up to 30 origins, my SELECT statement is going to be 500 characters by itself. Each of the FROM clauses tops 250 characters (x30 columns = >7500 characters). Then I need another 1000 characters for the WHERE clause and only 30 for the ORDER BY clause.
All in all, I'm estimating almost 10,000 characters for the entire statement. Can I use a CLOB for this? Do I need to write to file and call it as I would a script? Just looking for options.
Here's the start I've got so far:
CREATE OR REPLACE PROCEDURE PROC_BWSUM(f NUMBER, dtst DATE, dted DATE)
AS
  --Passed in parameters
  v_fact        NUMBER      := f;
  v_from_dt     DATE        := dtst;
  v_to_dt       DATE        := dted;
  --query construction variables
  qry_col_ct    INTEGER     := 0;
  qry_stn_ct    INTEGER     := 0;
  qry_stn_id    NUMBER;
  qry_stn_nm    VARCHAR2(27);
  qry_stn_key   NUMBER;
  qry_select    VARCHAR2(505);  -- will accommodate up to 30 stations
  qry_from      VARCHAR2(4000);
  qry_where     VARCHAR2(100);
  qry_orderby   VARCHAR2(100);
  cursor c_cols is select j.STATION_ID, s.SHORT_NM, COUNT(j.LOAD_JOB_ID)
        from LOAD_RATES r, STATIONS s, TC c, TC_LOAD_JOBS j
       where c.TC_ID = j.TC_ID and j.LOAD_RATE_ID = r.LOAD_RATE_ID
         and FN_STN_KEY(j.FACTORY_ID, j.STATION_ID) = s.KEY_ID
         and j.FACTORY_ID = v_fact and r.ACTIVE = 1
         and c.DATE_INDEX between to_date('09/01/2009','MM/DD/YYYY')
              and to_date('10/01/2009','MM/DD/YYYY')
       group by j.STATION_ID, s.SHORT_NM
       having COUNT(j.LOAD_JOB_ID) > 0
       order by j.STATION_ID;
  --value variables
  v_lds         NUMBER;
  v_tons        NUMBER;
begin
  htp.p('<table>');
  qry_select    := 'select base.DAY';
  qry_from      := ' from (select c.DATE_INDEX as DAY, COUNT(j.LOAD_JOB_ID) as LDS
           from TC c, TC_LOAD_JOBS j
          where c.TC_ID = j.TC_ID
            and c.DATE_INDEX between v_from_dt and v_to_dt
            and j.FACTORY_ID = v_fact
          group by c.DATE_INDEX) base';
  qry_where     := '';
  qry_orderby   := ' order by base.DAY;';
  htp.p('<tr>');
  select COUNT(r.LOAD_RATE_ID) into qry_col_ct
    from LOAD_RATES r where r.FACTORY_ID = v_fact and r.ACTIVE = 1;
  open c_cols;
    LOOP
    FETCH c_cols into qry_stn_id, qry_stn_nm, qry_stn_key;
    EXIT WHEN c_cols%NOTFOUND;
      qry_stn_ct    := qry_stn_ct + 1;
      htp.p('<th colspan=2>'||qry_stn_nm||' ('||qry_stn_id||')</th>');
      qry_select    := qry_select ||', a'||qry_stn_ct||'.LDS, a'||qry_stn_ct||'.TNS';
      IF qry_stn_ct = 1 THEN
        qry_where     := qry_where  ||'where base.DAY = a'||qry_stn_ct||'.DAY (+)';
      ELSE
        qry_where     := qry_where  ||'and base.DAY = a'||qry_stn_ct||'.DAY (+)';
      END IF;
    END LOOP;
  close c_cols;
  htp.p('</tr>');
  htp.p('<tr>');
      htp.p('<td colspan = '||qry_col_ct*2||'>'||qry_select||'</td>');
  htp.p('</tr>');
  htp.p('<tr>');
      htp.p('<td colspan = '||qry_col_ct*2||'>'||qry_from||'</td>');
  htp.p('</tr>');
  htp.p('<tr>');
      htp.p('<td colspan = '||qry_col_ct*2||'>'||qry_where||'</td>');
  htp.p('</tr>');
  htp.p('<tr>');
      htp.p('<td colspan = '||qry_col_ct*2||'>'||qry_orderby||'</td>');
  htp.p('</tr>');
  htp.p('</table>');
end;
/