10G-form: How to do MULTIPLE WORD OR SINGLE WORD SEARCH at ITEM?

Question

I am using 10G DB + 10G Form Builder.
I've 'database_item' where I am storing multiline text data.
I want to do MULTIPLE WORD OR SINGLE WORD SEARCH at this item through FORM.
I've tried by creating the following NON Database Items on the form:
- multiline_search_text_item, and
- multiline_result__text_item
And then writing execute_query in KEY-NEXT-ITEM trigger.
I've also tried using this in the POST-TEXT-ITEM at multiline_search_text_item:
select table.database_item into :multiline_result__text_item from table where multiline_search_text_item = :multiline_search_text_item;
Pl help me asap.
Gaurav

Answer 1

What you want to do is not clear.
The query you wrote will select records where the table contains exactly what has been written in the search item. You can use LIKE and a wildcard search to find records which contain the search text:
select table.database_item into :multiline_result__text_item
from table
where multiline_search_text_item LIKE '%'||:multiline_search_text_item||'%';
You can use UPPER to make this case insensitive:
select table.database_item into :multiline_result__text_item
from table
where Upper(multiline_search_text_item) LIKE Upper('%'||:multiline_search_text_item||'%');
But I suspect you want either to match the individual words in the search text to individual words in the database multiline field, or find records where the search words appear (not necessarily as whole words). In that case, check out the following:
-- set up a table (multiline and various whitespaces)
DROP TABLE t;
CREATE TABLE t AS
SELECT
  ROWNUM rn,
  owner||Chr(9)||object_name||Chr(160)||subobject_name||'    '||object_id||' '||
  data_object_id||Chr(10)||object_type||' '||created||' '||last_ddl_time||' '||
  timestamp||' '||status||' '||temporary||' '||generated||' '||secondary AS col
FROM all_objects;
-- check the format of the multiline text item (col)
SELECT * FROM t WHERE ROWNUM < 4;
-- a type for the function below
CREATE TYPE string_tab AS TABLE OF VARCHAR2(255);
-- this function takes a string and cuts out each word, idetifying words
-- as being separated by any whitespace.  it returns a collection
CREATE OR REPLACE FUNCTION string_to_tab(
  p_string IN VARCHAR2) RETURN string_tab IS
  l_string LONG DEFAULT
    RTrim(regexp_replace(p_string,'[[:space:]]+',' ')) || ' ';
  l_data string_tab := string_tab();
  n NUMBER;
BEGIN
  LOOP
    EXIT WHEN l_string IS NULL;
    n := InStr(l_string, ' ');
    l_data.extend;
    l_data(l_data.Count) := LTrim(RTrim(SubStr(l_string, 1, n - 1)));
    l_string := SubStr(l_string, n + 1);
  END LOOP;
  RETURN l_data;
END string_to_tab;
-- selecting from t where ANY of the words in the search text has
-- a match in the multiline field (as a word or part of a word), so SYS
-- matches to MDSYS and SYSTEM
SELECT DISTINCT
  t.rn, t.col
FROM
  t,
  (SELECT column_value
   FROM TABLE(CAST(string_to_tab('SYS   INDEX') AS string_tab))) x
WHERE InStr(t.col,x.column_value) > 0;
-- selecting from t where ALL of the words in the search text has
-- a match in the multiline field (as a word or part of a word), so SYS
-- matches to MDSYS and SYSTEM
SELECT rn, col FROM(
SELECT
  t.rn, t.col, cnt_x, Count(*) cnt
FROM
  t,
  (SELECT column_value , Count(1) over(PARTITION BY 1)cnt_x
   FROM TABLE(CAST(string_to_tab('SYS   INDEX') AS string_tab))) x
WHERE InStr(t.col,x.column_value) > 0
GROUP BY t.rn, t.col, cnt_x
WHERE cnt = cnt_x;
-- selecting from t where ANY of the words in the search text
-- match a word in the multiline field, so SYS matches only to SYS
SELECT DISTINCT
  t.rn, t.col
FROM
  t,
  (TABLE(CAST(string_to_tab(t.col) AS string_tab))) t2,
  (SELECT column_value
   FROM TABLE(CAST(string_to_tab('SYS   INDEX') AS string_tab))) x
WHERE t2.column_value = x.column_value;
-- selecting from t where ALL of the words in the search text
-- match a word in the multiline field, so SYS matches only to SYS
SELECT rn, col FROM(
SELECT
  t.rn, t.col, cnt_x, Count(*) cnt
FROM
  t,
  (TABLE(CAST(string_to_tab(t.col) AS string_tab))) t2,
  (SELECT column_value , Count(1) over(PARTITION BY 1)cnt_x
   FROM TABLE(CAST(string_to_tab('SYS   INDEX') AS string_tab))) x
WHERE t2.column_value = x.column_value
GROUP BY t.rn, t.col, cnt_x
WHERE cnt = cnt_x;For your application you would replace 'SYS INDEX' with a variable (the search field). You can use upper() and wildcards for case insensitive & 'fuzzy' searches. You might need to modify the function so it removes other delimiters such as commas and colons.