Control Characters in Data

Question

I've created a simple table with 2 rows. When I query the table for an individual row (where id = 1), my return set has control characters in between 2 columns...sometimes.
If I select * from the table, I get a core dump, then a Oracle sqlplus error message that wants to send a report to Microsoft. Once I decide to send/not send the message, sqlplus blows away.
These 2 rows are actually 2 rows in a much larger table. I encountered this problem in the larger table, thought my data was corrupted, so I created the exact same table, in a totally different schema, and pasted the data in the new table. When I received the same error, I dropped the table, recreated it, and hand entered all the data, with the same results.
Table structure:
id NUMBER(9) PRIMARY KEY
TABLE_ID NUMBER(9)
FILTER_ONE VARCHAR2(800)
FILTER_TWO VARCHAR2(100)
CLASS_FILTER VARCHAR2(1)
If I select* FROM test_table WHERE ID = 27, my result set looks like this:
'blah blah (' ☻ '))' ☺ 'Y'
The 1st open parenthesis is the end of the filter_one column, the double parenthesis is the entire filter_two, and the Y is the class filter. The control characters come from I know not where, but my app cannot function - this is a dynamically built query and the query of course fails because sqlplus can't interpret the control characters.
Anyone ever heard of this? I should also mention that this query has worked for the past year, and has not been changed - it just suddenly stopped working.
Thanks!
Dave

Answer 1

Sure -
CREATE TABLE TEST (
ID NUMBER(9) PRIMARY KEY,
TABLE_ID NUMBER(9),
FILTER_ONE VARCHAR2(800),
FILTER_TWO VARCHAR2(100),
CLASS_FILTER VARCHAR2(1) )
INSERT INTO test VALUES(
27,38,'where id in (select decode(pci.config_comp_int_id,null,decode(pci.component_id,null,pci.spirit_comp_id, pci.component_id),
decode(cci.component_id,null,cci.component_id, cci.spirit_comp_id)) from c3net_units u
join (select uc.*,connect_by_root(unit_id) top_unit_id
from c3net_unit_config_int uc start with uc.unit_id is not null connect by prior uc.id = uc.parent_id) uci
on (u.id = uci.top_unit_id)
left join (c3net_unit_plat_int upi join c3net_plat_comp_int pci on
( upi.id = pci.unit_plat_int_id)) on ( uci.id = upi.unit_config_int_id)
left join (c3net_config_plat_int cpi join c3net_config_comp_int cci on (cpi.id = cci.config_plat_int_id))
on (uci.config_id = cpi.config_id) where u.version_id in (',
'))','Y');
INSERT INTO test VALUES(
28,39,'where id in (select decode(upi.config_plat_int_id,null,decode(upi.platform_id,null,upi.spirit_plat_id,
upi.platform_id),
decode(cpi.platform_id,null,cpi.spirit_plat_id, cpi.platform_id)) from c3net_units u
join (select uc.*,connect_by_root(unit_id) top_unit_id
from c3net_unit_config_int uc start with uc.unit_id is not null connect by prior uc.id = uc.parent_id) uci
on (u.id = uci.top_unit_id)
left join c3net_unit_plat_int upi on ( uci.id = upi.unit_config_int_id)
left join c3net_config_plat_int cpi on (uci.config_id = cpi.config_id) where u.version_id in (',
'))','Y');
When I then "select * from test where id = 27;" I get strange data - looks like control characters - in the filter_one and filter_two fields
When I change the id to 28 and re-query, it gives me the core dump and sqlplus error.
Thanks for your time guys