Searching for content in stored webpages?

Question

I am a student and Oracle Text is new to me. I am using isqlPlus with Oracle10g
I am trying to search the content of external websites I upload using a procedure. The files are uploaded as BLOBs. I know UltraSearch is out there, but I´m trying to make an easy example using only SQL syntax. So far, I am able to load the html documents into a directory, but the search results are always 0. I don´t know what I am missing (sorry for the ignorance)! Can anyone give me some tips?
I am not a DBA. This are the rights I have:
CONNECT
CREATE SESSION
CREATE ANY DIRECTORY
CTXAPP
RESOURCE
CREATE CLUSTER
CREATE INDEXTYPE
CREATE OPERATOR
CREATE PROCEDURE
CREATE SEQUENCE
CREATE TABLE
CREATE TRIGGER
CREATE TYPE
UNLIMITED TABLESPACE
--First I create a table
DROP TABLE t03_sorbaugh.CONTENT_INVENTORY;
CREATE TABLE t03_sorbaugh.CONTENT_INVENTORY (
CONTENT_INVENTORY_ID NUMBER NOT NULL,
FILE_NAME VARCHAR2(1024),
KEYWORDS VARCHAR2(2048),
URL SYS.HTTPURITYPE,
TEXT BLOB DEFAULT empty_blob(),
TEXT_LOADED CHAR(1) DEFAULT 'N' NOT NULL,
CONSTRAINT UNQ_CONTENT_INVENTORY
UNIQUE(CONTENT_INVENTORY_ID) USING INDEX
TABLESPACE USERS
--Then I load an external Website
INSERT INTO t03_sorbaugh.CONTENT_INVENTORY
(CONTENT_INVENTORY_ID, URL, KEYWORDS)
VALUES ('1', sys.httpuritype.createuri('http://en.wikipedia.org/wiki/Art'), '');
COMMIT;
--The Procedure looks like this (sorry for not keeping it short)
CREATE OR REPLACE PROCEDURE loadLOBFromBFILE IS
Dest_loc BLOB;
Src_loc BFILE := BFILENAME('DIR_TESTCASE', '');
Amount INTEGER := 0;
cursor ContentInv_cursor is
--create a recordset of the file names
--that we need to load from the filesystem into the
--Oracle database.
SELECT
CONTENT_INVENTORY_ID,
FILE_NAME
FROM t03_sorbaugh.CONTENT_INVENTORY
WHERE FILE_NAME IS NOT NULL;
varContentInv_id CONTENT_INVENTORY.CONTENT_INVENTORY_ID%TYPE;
varFileName CONTENT_INVENTORY.FILE_NAME%TYPE;
BEGIN
open ContentInv_cursor;
loop
varFileName := NULL;
varContentInv_id := NULL;
fetch ContentInv_cursor into varContentInv_id, varFileName;
exit when ContentInv_cursor%notfound;
Src_loc := BFILENAME('DIR_TESTCASE', varFileName);
IF DBMS_LOB.FILEEXISTS (Src_loc) = 1 THEN
SELECT text INTO Dest_loc FROM CONTENT_INVENTORY
WHERE CONTENT_INVENTORY_ID = varContentInv_id FOR UPDATE;
/* Opening the LOB is mandatory */
DBMS_LOB.OPEN(Src_loc, DBMS_LOB.LOB_READONLY);
DBMS_LOB.LOADFROMFILE(Dest_loc, Src_loc, dbms_lob.getlength(Src_loc));
/* Closing the LOB is mandatory */
DBMS_LOB.CLOSE(Src_loc);
/* Optional update statement */
UPDATE t03_sorbaugh.CONTENT_INVENTORY
SET TEXT_LOADED = 'Y'
WHERE CONTENT_INVENTORY_ID = varContentInv_id;
COMMIT;
END IF;
end loop;
EXCEPTION
WHEN OTHERS THEN
RAISE_APPLICATION_ERROR (-20001, 'Error occurred while loading LOBs.');
close ContentInv_cursor;
END;
-- I upload the file to the directory 'd:\daten\oracletextpraktikum'
set serveroutput on;
EXECUTE loadLOBFromBFILE;
-- PL/SQL-Prozedur erfolgreich abgeschlossen.
-- Now the preferences and the index
begin
ctx_ddl.create_preference('doc_lexer', 'BASIC_LEXER');
ctx_ddl.set_attribute('doc_lexer', 'printjoins', '_-');
end;
drop index idxContentMgmtURL;
create index idxContentMgmtURL on content_inventory(url) indextype is ctxsys.context;
-- Finally, I run an example where I search for the word "Art" ... the only Webpage uploaded to my directory contains the Art definition from Wikipedia, but still, I get no search results :(
SET LINESIZE 5000
SET PAGESIZE 5000
select content_inventory_id, url
from t03_sorbaugh.content_inventory
where contains(url,'&WORD_TO_FIND') > 0;
--Type "Art"
neu 3: where contains(url,'art') > 0
Es wurden keine Zeilen ausgewählt
Is isqlPlus somehow limited in using Oracle Text? I can´t seem to find the Problem!

Answer 1

Okay, but I'm afarid I can't help much more than that, as I'm still new at Oracle Text myself. From my limited knowledge, it all looks fine, but like you, I must be missing something.
You might try asking the DBA if the TXTSUP_UTIL package is installed in the CTXSYS schema. If it isn't, he can download it from Metalink or I can send it via email.
While it won't fix the problem, it does offer some insight into the index preferences and storage, etc.
BIll Ferguson