( I think it could be a simple question for the most users here ..)
Short description:
I need a way (maybe with PL/SQL which I don't know ..) to "select defined data from all tables which contain this type of data"
Long description (example):
I have a different number of different tables. An often changing part of them - I don't know the number and the names - contains the column "FID". Now I need two steps:
a) Select all tables which contain the column "FID". ( I know how to do this as single step)
b) Select from all found tables the value FID and show it.
For me the problem is the step from a) to b). With known tables I would use UNION, but with a dynamic result of tables I have no idea ..
You could use a variation on an XML magic trick, by using dbms_xmlgen to get all the values into XML documents based on a query against user_tab_columns:
select dbms_xmlgen.getxmltype(
'select "' || column_name || '" from "' || table_name || '"')
from user_tab_columns
where upper(column_name) = 'FID'
and data_type = 'NUMBER';
... where I'm assuming FID is expected to be a numeric ID, so limiting only to numeric columns (and also allowing for mixed case/quoted identifiers for table and columns names, just in case). That gives one row per table, with an XML document listing the FID values in that table.
Then from that XML you can extract the individual values, again as numbers:
with cte (xml) as (
select dbms_xmlgen.getxmltype(
'select "' || column_name || '" as fid from "' || table_name || '"')
from user_tab_columns
where upper(column_name) = 'FID'
and data_type = 'NUMBER'
)
select x.fid
from cte
cross apply xmltable(
'/ROWSET/ROW'
passing cte.xml
columns fid number path 'FID'
) x;
Or if you want to see the table/column each value came from, just include those in the CTE and select list:
with cte (table_name, column_name, xml) as (
select table_name, column_name, dbms_xmlgen.getxmltype(
'select "' || column_name || '" as fid from "' || table_name || '"')
from user_tab_columns
where upper(column_name) = 'FID'
and data_type = 'NUMBER'
)
select cte.table_name, cte.column_name, x.fid
from cte
cross apply xmltable(
'/ROWSET/ROW'
passing cte.xml
columns fid number path 'FID'
) x;
If you want to search other schemas, then use all_tab_columns instead, and optionally include each table's owner:
with cte (owner, table_name, column_name, xml) as (
select owner, table_name, column_name, dbms_xmlgen.getxmltype(
'select "' || column_name || '" as fid from "' || owner || '"."' || table_name || '"')
from all_tab_columns
where upper(column_name) = 'FID'
and data_type = 'NUMBER'
)
select cte.owner, cte.table_name, cte.column_name, x.fid
from cte
cross apply xmltable(
'/ROWSET/ROW'
passing cte.xml
columns fid number path 'FID'
) x;
db<>fiddle
The basis for this trick goes back to at least 2007 but may be even older, from before getxmltype() existed (it seems to have been added in 10g); I'd originally used xmltype(getxml()):
select xmltype(dbms_xmlgen.getxml(
'select "' || column_name || '" from "' || table_name || '"'))
from user_tab_columns
where upper(column_name) = 'FID'
and data_type = 'NUMBER';
which works most of the time, but if any of the tables are empty throws "ORA-06502: PL/SQL: numeric or value error".
If you want to use pl/sql I really love pipelined functions:
create type result_type as Object ( text varchar2(2000) );
create type result_type_table as table of result_type;
create or replace function select_all( p_column_name in varchar2 )
return result_type_table
deterministic
pipelined
as
v_table_name varchar2(40);
v_result result_type := result_type('');
v_table_name_cursor sys_refcursor;
v_inner_cursor sys_refcursor;
begin
open v_table_name_cursor
for 'select a.table_name
from user_tab_cols a
, user_tables b
where a.column_name = :1
and a.table_name = b.table_name'
using upper(p_column_name);
loop
fetch v_table_name_cursor into v_table_name;
exit when v_table_name_cursor%notfound;
open v_inner_cursor
for 'select '||p_column_name||' from '||v_table_name;
loop
fetch v_inner_cursor into v_result.text;
exit when v_inner_cursor%notfound;
pipe row (v_result );
end loop;
close v_inner_cursor;
end loop;
close v_table_name_cursor;
end;
/
Using this function is simple:
select * from table( select_all('your_column_name') );
db<>fiddle
Is it possible to search every field of every table for a particular value in Oracle?
There are hundreds of tables with thousands of rows in some tables so I know this could take a very long time to query. But the only thing I know is that a value for the field I would like to query against is 1/22/2008P09RR8.
<
I've tried using this statement below to find an appropriate column based on what I think it should be named but it returned no results.
SELECT * from dba_objects
WHERE object_name like '%DTN%'
There is absolutely no documentation on this database and I have no idea where this field is being pulled from.
Any thoughts?
Quote:
I've tried using this statement below
to find an appropriate column based on
what I think it should be named but it
returned no results.*
SELECT * from dba_objects WHERE
object_name like '%DTN%'
A column isn't an object. If you mean that you expect the column name to be like '%DTN%', the query you want is:
SELECT owner, table_name, column_name FROM all_tab_columns WHERE column_name LIKE '%DTN%';
But if the 'DTN' string is just a guess on your part, that probably won't help.
By the way, how certain are you that '1/22/2008P09RR8' is a value selected directly from a single column? If you don't know at all where it is coming from, it could be a concatenation of several columns, or the result of some function, or a value sitting in a nested table object. So you might be on a wild goose chase trying to check every column for that value. Can you not start with whatever client application is displaying this value and try to figure out what query it is using to obtain it?
Anyway, diciu's answer gives one method of generating SQL queries to check every column of every table for the value. You can also do similar stuff entirely in one SQL session using a PL/SQL block and dynamic SQL. Here's some hastily-written code for that:
SET SERVEROUTPUT ON SIZE 100000
DECLARE
match_count INTEGER;
BEGIN
FOR t IN (SELECT owner, table_name, column_name
FROM all_tab_columns
WHERE owner <> 'SYS' and data_type LIKE '%CHAR%') LOOP
EXECUTE IMMEDIATE
'SELECT COUNT(*) FROM ' || t.owner || '.' || t.table_name ||
' WHERE '||t.column_name||' = :1'
INTO match_count
USING '1/22/2008P09RR8';
IF match_count > 0 THEN
dbms_output.put_line( t.table_name ||' '||t.column_name||' '||match_count );
END IF;
END LOOP;
END;
/
There are some ways you could make it more efficient too.
In this case, given the value you are looking for, you can clearly eliminate any column that is of NUMBER or DATE type, which would reduce the number of queries. Maybe even restrict it to columns where type is like '%CHAR%'.
Instead of one query per column, you could build one query per table like this:
SELECT * FROM table1
WHERE column1 = 'value'
OR column2 = 'value'
OR column3 = 'value'
...
;
I did some modification to the above code to make it work faster if you are searching in only one owner.
You just have to change the 3 variables v_owner, v_data_type and v_search_string to fit what you are searching for.
SET SERVEROUTPUT ON SIZE 100000
DECLARE
match_count INTEGER;
-- Type the owner of the tables you are looking at
v_owner VARCHAR2(255) :='ENTER_USERNAME_HERE';
-- Type the data type you are look at (in CAPITAL)
-- VARCHAR2, NUMBER, etc.
v_data_type VARCHAR2(255) :='VARCHAR2';
-- Type the string you are looking at
v_search_string VARCHAR2(4000) :='string to search here...';
BEGIN
FOR t IN (SELECT table_name, column_name FROM all_tab_cols where owner=v_owner and data_type = v_data_type) LOOP
EXECUTE IMMEDIATE
'SELECT COUNT(*) FROM '||t.table_name||' WHERE '||t.column_name||' = :1'
INTO match_count
USING v_search_string;
IF match_count > 0 THEN
dbms_output.put_line( t.table_name ||' '||t.column_name||' '||match_count );
END IF;
END LOOP;
END;
/
I know this is an old topic. But I see a comment to the question asking if it could be done in SQL rather than using PL/SQL. So thought to post a solution.
The below demonstration is to Search for a VALUE in all COLUMNS of all TABLES in an entire SCHEMA:
Search a CHARACTER type
Let's look for the value KING in SCOTT schema.
SQL> variable val varchar2(10)
SQL> exec :val := 'KING'
PL/SQL procedure successfully completed.
SQL> SELECT DISTINCT SUBSTR (:val, 1, 11) "Searchword",
2 SUBSTR (table_name, 1, 14) "Table",
3 SUBSTR (column_name, 1, 14) "Column"
4 FROM cols,
5 TABLE (xmlsequence (dbms_xmlgen.getxmltype ('select '
6 || column_name
7 || ' from '
8 || table_name
9 || ' where upper('
10 || column_name
11 || ') like upper(''%'
12 || :val
13 || '%'')' ).extract ('ROWSET/ROW/*') ) ) t
14 ORDER BY "Table"
15 /
Searchword Table Column
----------- -------------- --------------
KING EMP ENAME
SQL>
Search a NUMERIC type
Let's look for the value 20 in SCOTT schema.
SQL> variable val NUMBER
SQL> exec :val := 20
PL/SQL procedure successfully completed.
SQL> SELECT DISTINCT SUBSTR (:val, 1, 11) "Searchword",
2 SUBSTR (table_name, 1, 14) "Table",
3 SUBSTR (column_name, 1, 14) "Column"
4 FROM cols,
5 TABLE (xmlsequence (dbms_xmlgen.getxmltype ('select '
6 || column_name
7 || ' from '
8 || table_name
9 || ' where upper('
10 || column_name
11 || ') like upper(''%'
12 || :val
13 || '%'')' ).extract ('ROWSET/ROW/*') ) ) t
14 ORDER BY "Table"
15 /
Searchword Table Column
----------- -------------- --------------
20 DEPT DEPTNO
20 EMP DEPTNO
20 EMP HIREDATE
20 SALGRADE HISAL
20 SALGRADE LOSAL
SQL>
Yes you can and your DBA will hate you and will find you to nail your shoes to the floor because that will cause lots of I/O and bring the database performance really down as the cache purges.
select column_name from all_tab_columns c, user_all_tables u where c.table_name = u.table_name;
for a start.
I would start with the running queries, using the v$session and the v$sqlarea. This changes based on oracle version. This will narrow down the space and not hit everything.
Here is another modified version that will compare a lower substring match. This works in Oracle 11g.
DECLARE
match_count INTEGER;
-- Type the owner of the tables you are looking at
v_owner VARCHAR2(255) :='OWNER_NAME';
-- Type the data type you are look at (in CAPITAL)
-- VARCHAR2, NUMBER, etc.
v_data_type VARCHAR2(255) :='VARCHAR2';
-- Type the string you are looking at
v_search_string VARCHAR2(4000) :='%lower-search-sub-string%';
BEGIN
FOR t IN (SELECT table_name, column_name FROM all_tab_cols where owner=v_owner and data_type = v_data_type) LOOP
EXECUTE IMMEDIATE
'SELECT COUNT(*) FROM '||t.table_name||' WHERE lower('||t.column_name||') like :1'
INTO match_count
USING v_search_string;
IF match_count > 0 THEN
dbms_output.put_line( t.table_name ||' '||t.column_name||' '||match_count );
END IF;
END LOOP;
END;
/
I modified Flood's script to execute once for each table rather than for every column of each table for faster execution. It requires Oracle 11g or greater.
set serveroutput on size 100000
declare
v_match_count integer;
v_counter integer;
-- The owner of the tables to search through (case-sensitive)
v_owner varchar2(255) := 'OWNER_NAME';
-- A string that is part of the data type(s) of the columns to search through (case-insensitive)
v_data_type varchar2(255) := 'CHAR';
-- The string to be searched for (case-insensitive)
v_search_string varchar2(4000) := 'FIND_ME';
-- Store the SQL to execute for each table in a CLOB to get around the 32767 byte max size for a VARCHAR2 in PL/SQL
v_sql clob := '';
begin
for cur_tables in (select owner, table_name from all_tables where owner = v_owner and table_name in
(select table_name from all_tab_columns where owner = all_tables.owner and data_type like '%' || upper(v_data_type) || '%')
order by table_name) loop
v_counter := 0;
v_sql := '';
for cur_columns in (select column_name from all_tab_columns where
owner = v_owner and table_name = cur_tables.table_name and data_type like '%' || upper(v_data_type) || '%') loop
if v_counter > 0 then
v_sql := v_sql || ' or ';
end if;
v_sql := v_sql || 'upper(' || cur_columns.column_name || ') like ''%' || upper(v_search_string) || '%''';
v_counter := v_counter + 1;
end loop;
v_sql := 'select count(*) from ' || cur_tables.table_name || ' where ' || v_sql;
execute immediate v_sql
into v_match_count;
if v_match_count > 0 then
dbms_output.put_line('Match in ' || cur_tables.owner || ': ' || cur_tables.table_name || ' - ' || v_match_count || ' records');
end if;
end loop;
exception
when others then
dbms_output.put_line('Error when executing the following: ' || dbms_lob.substr(v_sql, 32600));
end;
/
I was having following issues for #Lalit Kumars answer,
ORA-19202: Error occurred in XML processing
ORA-00904: "SUCCESS": invalid identifier
ORA-06512: at "SYS.DBMS_XMLGEN", line 288
ORA-06512: at line 1
19202. 00000 - "Error occurred in XML processing%s"
*Cause: An error occurred when processing the XML function
*Action: Check the given error message and fix the appropriate problem
Solution is:
WITH char_cols AS
(SELECT /*+materialize */ table_name, column_name
FROM cols
WHERE data_type IN ('CHAR', 'VARCHAR2'))
SELECT DISTINCT SUBSTR (:val, 1, 11) "Searchword",
SUBSTR (table_name, 1, 14) "Table",
SUBSTR (column_name, 1, 14) "Column"
FROM char_cols,
TABLE (xmlsequence (dbms_xmlgen.getxmltype ('select "'
|| column_name
|| '" from "'
|| table_name
|| '" where upper("'
|| column_name
|| '") like upper(''%'
|| :val
|| '%'')' ).extract ('ROWSET/ROW/*') ) ) t
ORDER BY "Table"
/
I would do something like this (generates all the selects you need).
You can later on feed them to sqlplus:
echo "select table_name from user_tables;" | sqlplus -S user/pwd | grep -v "^--" | grep -v "TABLE_NAME" | grep "^[A-Z]" | while read sw;
do echo "desc $sw" | sqlplus -S user/pwd | grep -v "\-\-\-\-\-\-" | awk -F' ' '{print $1}' | while read nw;
do echo "select * from $sw where $nw='val'";
done;
done;
It yields:
select * from TBL1 where DESCRIPTION='val'
select * from TBL1 where ='val'
select * from TBL2 where Name='val'
select * from TBL2 where LNG_ID='val'
And what it does is - for each table_name from user_tables get each field (from desc) and create a select * from table where field equals 'val'.
if we know the table and colum names but want to find out the number of times string is appearing for each schema:
Declare
owner VARCHAR2(1000);
tbl VARCHAR2(1000);
cnt number;
ct number;
str_sql varchar2(1000);
reason varchar2(1000);
x varchar2(1000):='%string_to_be_searched%';
cursor csr is select owner,table_name
from all_tables where table_name ='table_name';
type rec1 is record (
ct VARCHAR2(1000));
type rec is record (
owner VARCHAR2(1000):='',
table_name VARCHAR2(1000):='');
rec2 rec;
rec3 rec1;
begin
for rec2 in csr loop
--str_sql:= 'select count(*) from '||rec.owner||'.'||rec.table_name||' where CTV_REMARKS like '||chr(39)||x||chr(39);
--dbms_output.put_line(str_sql);
--execute immediate str_sql
execute immediate 'select count(*) from '||rec2.owner||'.'||rec2.table_name||' where column_name like '||chr(39)||x||chr(39)
into rec3;
if rec3.ct <> 0 then
dbms_output.put_line(rec2.owner||','||rec3.ct);
else null;
end if;
end loop;
end;
Procedure to Search Entire Database:
CREATE or REPLACE PROCEDURE SEARCH_DB(SEARCH_STR IN VARCHAR2, TAB_COL_RECS OUT VARCHAR2) IS
match_count integer;
qry_str varchar2(1000);
CURSOR TAB_COL_CURSOR IS
SELECT TABLE_NAME,COLUMN_NAME,OWNER,DATA_TYPE FROM ALL_TAB_COLUMNS WHERE DATA_TYPE in ('NUMBER','VARCHAR2') AND OWNER='SCOTT';
BEGIN
FOR TAB_COL_REC IN TAB_COL_CURSOR
LOOP
qry_str := 'SELECT COUNT(*) FROM '||TAB_COL_REC.OWNER||'.'||TAB_COL_REC.TABLE_NAME||
' WHERE '||TAB_COL_REC.COLUMN_NAME;
IF TAB_COL_REC.DATA_TYPE = 'NUMBER' THEN
qry_str := qry_str||'='||SEARCH_STR;
ELSE
qry_str := qry_str||' like '||SEARCH_STR;
END IF;
--dbms_output.put_line( qry_str );
EXECUTE IMMEDIATE qry_str INTO match_count;
IF match_count > 0 THEN
dbms_output.put_line( qry_str );
--dbms_output.put_line( TAB_COL_REC.TABLE_NAME ||' '||TAB_COL_REC.COLUMN_NAME ||' '||match_count);
TAB_COL_RECS := TAB_COL_RECS||'##'||TAB_COL_REC.TABLE_NAME||'##'||TAB_COL_REC.COLUMN_NAME;
END IF;
END LOOP;
END SEARCH_DB;
Execute Statement
DECLARE
SEARCH_STR VARCHAR2(200);
TAB_COL_RECS VARCHAR2(200);
BEGIN
SEARCH_STR := 10;
SEARCH_DB(
SEARCH_STR => SEARCH_STR,
TAB_COL_RECS => TAB_COL_RECS
);
DBMS_OUTPUT.PUT_LINE('TAB_COL_RECS = ' || TAB_COL_RECS);
END;
Sample Results
Connecting to the database test.
SELECT COUNT(*) FROM SCOTT.EMP WHERE DEPTNO=10
SELECT COUNT(*) FROM SCOTT.DEPT WHERE DEPTNO=10
TAB_COL_RECS = ##EMP##DEPTNO##DEPT##DEPTNO
Process exited.
Disconnecting from the database test.
I don't of a simple solution on the SQL promprt. Howeve there are quite a few tools like toad and PL/SQL Developer that have a GUI where a user can input the string to be searched and it will return the table/procedure/object where this is found.
There are some free tools that make these kind of search, for example, this one works fine and source code is available:
https://sites.google.com/site/freejansoft/dbsearch
You'll need the Oracle ODBC driver and a DSN to use this tool.
Modifying the code to search case-insensitively using a LIKE query instead of finding exact matches...
DECLARE
match_count INTEGER;
-- Type the owner of the tables you want to search.
v_owner VARCHAR2(255) :='USER';
-- Type the data type you're looking for (in CAPS). Examples include: VARCHAR2, NUMBER, etc.
v_data_type VARCHAR2(255) :='VARCHAR2';
-- Type the string you are looking for.
v_search_string VARCHAR2(4000) :='Test';
BEGIN
dbms_output.put_line( 'Starting the search...' );
FOR t IN (SELECT table_name, column_name FROM all_tab_cols where owner=v_owner and data_type = v_data_type) LOOP
EXECUTE IMMEDIATE
'SELECT COUNT(*) FROM '||t.table_name||' WHERE LOWER('||t.column_name||') LIKE :1'
INTO match_count
USING LOWER('%'||v_search_string||'%');
IF match_count > 0 THEN
dbms_output.put_line( t.table_name ||' '||t.column_name||' '||match_count );
END IF;
END LOOP;
END;
I found the best solution but it's a little slow. (It will work perfectly with all SQL IDE's.)
SELECT DISTINCT table_name, column_name, data_type
FROM user_tab_cols,
TABLE (xmlsequence (dbms_xmlgen.getxmltype ('select '
|| column_name
|| ' from '
|| table_name
|| ' where lower('
|| column_name
|| ') like lower(''%'
|| 'your_text_here'
|| '%'')' ).extract ('ROWSET/ROW/*') ) ) a
where table_name not in (
select distinct table_name
from user_tab_cols where data_type like 'SDO%'
or data_type like '%LOB') AND DATA_TYPE = 'VARCHAR2'
order by table_name, column_name;
--it run completed -- no error
SET SERVEROUTPUT ON SIZE 100000
DECLARE
v_match_count INTEGER;
v_counter INTEGER;
v_owner VARCHAR2 (255) := 'VASOA';
v_search_string VARCHAR2 (4000) := '99999';
v_data_type VARCHAR2 (255) := 'CHAR';
v_sql CLOB := '';
BEGIN
FOR cur_tables
IN ( SELECT owner, table_name
FROM all_tables
WHERE owner = v_owner
AND table_name IN (SELECT table_name
FROM all_tab_columns
WHERE owner = all_tables.owner
AND data_type LIKE
'%'
|| UPPER (v_data_type)
|| '%')
ORDER BY table_name)
LOOP
v_counter := 0;
v_sql := '';
FOR cur_columns
IN (SELECT column_name, table_name
FROM all_tab_columns
WHERE owner = v_owner
AND table_name = cur_tables.table_name
AND data_type LIKE '%' || UPPER (v_data_type) || '%')
LOOP
IF v_counter > 0
THEN
v_sql := v_sql || ' or ';
END IF;
IF cur_columns.column_name is not null
THEN
v_sql :=
v_sql
|| 'upper('
|| cur_columns.column_name
|| ') ='''
|| UPPER (v_search_string)||'''';
v_counter := v_counter + 1;
END IF;
END LOOP;
IF v_sql is null
THEN
v_sql :=
'select count(*) from '
|| v_owner
|| '.'
|| cur_tables.table_name;
END IF;
IF v_sql is not null
THEN
v_sql :=
'select count(*) from '
|| v_owner
|| '.'
|| cur_tables.table_name
|| ' where '
|| v_sql;
END IF;
--v_sql := 'select count(*) from ' ||v_owner||'.'|| cur_tables.table_name ||' where '|| v_sql;
--dbms_output.put_line(v_sql);
--DBMS_OUTPUT.put_line (v_sql);
EXECUTE IMMEDIATE v_sql INTO v_match_count;
IF v_match_count > 0
THEN
DBMS_OUTPUT.put_line (v_sql);
dbms_output.put_line('Match in ' || cur_tables.owner || ': ' || cur_tables.table_name || ' - ' || v_match_count || ' records');
END IF;
END LOOP;
EXCEPTION
WHEN OTHERS
THEN
DBMS_OUTPUT.put_line (
'Error when executing the following: '
|| DBMS_LOB.SUBSTR (v_sql, 32600));
END;
/
Borrowing, slightly enhancing and simplifying from this Blog post the following simple SQL statement seems to do the job quite well:
SELECT DISTINCT (:val) "Search Value", TABLE_NAME "Table", COLUMN_NAME "Column"
FROM cols,
TABLE (XMLSEQUENCE (DBMS_XMLGEN.GETXMLTYPE(
'SELECT "' || COLUMN_NAME || '" FROM "' || TABLE_NAME || '" WHERE UPPER("'
|| COLUMN_NAME || '") LIKE UPPER(''%' || :val || '%'')' ).EXTRACT ('ROWSET/ROW/*')))
ORDER BY "Table";
The Oracle LIKE condition allows wildcards to be used in the WHERE clause of a SELECT, INSERT, UPDATE, or DELETE statement.
%: to match any string of any length
Eg-
SELECT last_name
FROM customer_tab
WHERE last_name LIKE '%A%';
-: to match on a single character
Eg-
SELECT last_name
FROM customer_tab
WHERE last_name LIKE 'A_t';
Lets say I have a table called student:
When I select from that table the result is:
SQL> select * from student;
NAME ROLL_NO
-------------------- --------------------
gokul 3
gokul1 34
Instead can we get the output like this? without using PIVOT keyword
SQL> select * from student;
NAME gokul gokul1
ROLL_NO 3 34
The simple answer is: no, you cannot.
In Oracle, it is possible to do a dynamic pivot on an unknown number of columns but it is not something that you can do using a simple SELECT * FROM student. You will need to use PL/SQL to build some dynamic SQL:
VARIABLE cur REFCURSOR;
DECLARE
p_sql CLOB;
BEGIN
SELECT 'SELECT ''ROLL_NO'' AS "NAME", '
|| LISTAGG(
'MAX( CASE name WHEN '''
|| name
|| ''' THEN roll_no END ) AS "'
|| name || '"',
','
) WITHIN GROUP ( ORDER BY ROWNUM )
|| ' FROM students'
INTO p_sql
FROM students;
OPEN :cur FOR p_sql;
END;
/
PRINT cur;
Output:
NAME gokul gokul1
------- ----------- -----------
ROLL_NO 3 34
Edit
The above code will work where the LISTAGG output is under 4000 bytes but if you want to go over that then you need a slightly different method:
VARIABLE cur REFCURSOR;
DECLARE
TYPE name_t IS TABLE OF students.name%type;
p_sql CLOB;
names name_t;
BEGIN
SELECT DISTINCT name
BULK COLLECT INTO names
FROM students;
p_sql := EMPTY_CLOB() || 'SELECT ''ROLL_NO'' AS "NAME"';
FOR i IN 1 .. names.COUNT LOOP
p_sql := p_sql || ', MAX( CASE name WHEN '''
|| names(i)
|| ''' THEN roll_no END ) AS "'
|| names(i)
|| '"';
END LOOP;
p_sql := p_sql || ' FROM students';
OPEN :cur FOR p_sql;
END;
/
PRINT cur;
In Oracle 10g, is there a way to do the following in PL/SQL?
for each table in database
for each row in table
for each column in row
if column is of type 'varchar2'
column = trim(column)
Thanks!
Of course, doing large-scale dynamic updates is potentially dangerous and time-consuming. But here's how you can generate the commands you want. This is for a single schema, and will just build the commands and output them. You could copy them into a script and review them before running. Or, you could change dbms_output.put_line( ... ) to EXECUTE IMMEDIATE ... to have this script execute all the statements as they are generated.
SET SERVEROUTPUT ON
BEGIN
FOR c IN
(SELECT t.table_name, c.column_name
FROM user_tables t, user_tab_columns c
WHERE c.table_name = t.table_name
AND data_type='VARCHAR2')
LOOP
dbms_output.put_line(
'UPDATE '||c.table_name||
' SET '||c.column_name||' = TRIM('||c.column_name||') WHERE '||
c.column_name||' <> TRIM('||c.column_name||') OR ('||
c.column_name||' IS NOT NULL AND TRIM('||c.column_name||') IS NULL)'
);
END LOOP;
END;
Presumably you want to do this for every column in a schema, not in the database. Trying to do this to the dictionary tables would be a bad idea...
declare
v_schema varchar2(30) := 'YOUR_SCHEMA_NAME';
cursor cur_tables (p_schema_name varchar2) is
select owner, table_name, column_name
from all_tables at,
inner join all_tab_columns atc
on at.owner = atc.owner
and at.table_name = atc.table_name
where atc.data_type = 'VARCHAR2'
and at.owner = p_schema;
begin
for r_table in cur_tables loop
execute immediate 'update ' || r.owner || '.' || r.table_name
|| ' set ' || r.column_name || ' = trim(' || r.column_name ||');';
end loop;
end;
This will only work for fields that are VARCHAR2s in the first place. If your database contains CHAR fields, then you're out of luck, because CHAR fields are always padded to their maximum length.
I have "read only" access to a few tables in an Oracle database. I need to get schema information on some of the columns. I'd like to use something analogous to MS SQL's sp_help.
I see the table I'm interested in listed in this query:
SELECT * FROM ALL_TABLES
When I run this query, Oracle tells me "table not found in schema", and yes the parameters are correct.
SELECT
DBMS_METADATA.GET_DDL('TABLE', 'ITEM_COMMIT_AGG', 'INTAMPS') AS DDL
FROM DUAL;
After using my Oracle universal translator 9000 I've surmised this doesn't work because I don't have sufficient privileges. Given my constraints how can I get the datatype and data length of a column on a table I have read access to with a PL-SQL statement?
ALL_TAB_COLUMNS should be queryable from PL/SQL. DESC is a SQL*Plus command.
SQL> desc all_tab_columns;
Name Null? Type
----------------------------------------- -------- ----------------------------
OWNER NOT NULL VARCHAR2(30)
TABLE_NAME NOT NULL VARCHAR2(30)
COLUMN_NAME NOT NULL VARCHAR2(30)
DATA_TYPE VARCHAR2(106)
DATA_TYPE_MOD VARCHAR2(3)
DATA_TYPE_OWNER VARCHAR2(30)
DATA_LENGTH NOT NULL NUMBER
DATA_PRECISION NUMBER
DATA_SCALE NUMBER
NULLABLE VARCHAR2(1)
COLUMN_ID NUMBER
DEFAULT_LENGTH NUMBER
DATA_DEFAULT LONG
NUM_DISTINCT NUMBER
LOW_VALUE RAW(32)
HIGH_VALUE RAW(32)
DENSITY NUMBER
NUM_NULLS NUMBER
NUM_BUCKETS NUMBER
LAST_ANALYZED DATE
SAMPLE_SIZE NUMBER
CHARACTER_SET_NAME VARCHAR2(44)
CHAR_COL_DECL_LENGTH NUMBER
GLOBAL_STATS VARCHAR2(3)
USER_STATS VARCHAR2(3)
AVG_COL_LEN NUMBER
CHAR_LENGTH NUMBER
CHAR_USED VARCHAR2(1)
V80_FMT_IMAGE VARCHAR2(3)
DATA_UPGRADED VARCHAR2(3)
HISTOGRAM VARCHAR2(15)
You can use the desc command.
desc MY_TABLE
This will give you the column names, whether null is valid, and the datatype (and length if applicable)
The best solution that I've found for such case is
select column_name, data_type||
case
when data_precision is not null and nvl(data_scale,0)>0 then '('||data_precision||','||data_scale||')'
when data_precision is not null and nvl(data_scale,0)=0 then '('||data_precision||')'
when data_precision is null and data_scale is not null then '(*,'||data_scale||')'
when char_length>0 then '('||char_length|| case char_used
when 'B' then ' Byte'
when 'C' then ' Char'
else null
end||')'
end||decode(nullable, 'N', ' NOT NULL')
from user_tab_columns
where table_name = 'TABLE_NAME'
and column_name = 'COLUMN_NAME';
#Aaron Stainback, thank you for correction!
Note: if you are trying to get this information for tables that are in a different SCHEMA use the all_tab_columns view, we have this problem as our Applications use a different SCHEMA for security purposes.
use the following:
EG:
SELECT
data_length
FROM
all_tab_columns
WHERE
upper(table_name) = 'MY_TABLE_NAME' AND upper(column_name) = 'MY_COL_NAME'
select t.data_type
from user_tab_columns t
where t.TABLE_NAME = 'xxx'
and t.COLUMN_NAME='aaa'
Oracle 11.2: Get a list of the full datatype in your table:
create table SOMETABLE (foo integer, bar varchar(300));
select data_type || '(' || data_length || ')' thetype
from user_tab_columns where TABLE_NAME = 'SOMETABLE';
Prints:
NUMBER(22)
VARCHAR(300)
Screenshot:
Documentation:
https://docs.oracle.com/cd/B19306_01/server.102/b14237/statviews_4462.htm#REFRN26277
select column_name, data_type || '(' || data_length || ')' as datatype
from all_tab_columns
where TABLE_NAME = upper('myTableName')
Quick and dirty way (e.g. to see how data is stored in oracle)
SQL> select dump(dummy) dump_dummy, dummy
, dump(10) dump_ten
from dual
DUMP_DUMMY DUMMY DUMP_TEN
---------------- ----- --------------------
Typ=1 Len=1: 88 X Typ=2 Len=2: 193,11
1 row selected.
will show that dummy column in table sys.dual has typ=1 (varchar2), while 10 is Typ=2 (number).
You can try this.
SELECT *
FROM (SELECT column_name,
data_type,
data_type
|| CASE
WHEN data_precision IS NOT NULL
AND NVL (data_scale, 0) > 0
THEN
'(' || data_precision || ',' || data_scale || ')'
WHEN data_precision IS NOT NULL
AND NVL (data_scale, 0) = 0
THEN
'(' || data_precision || ')'
WHEN data_precision IS NULL AND data_scale IS NOT NULL
THEN
'(*,' || data_scale || ')'
WHEN char_length > 0
THEN
'(' || char_length
|| CASE char_used
WHEN 'B' THEN ' Byte'
WHEN 'C' THEN ' Char'
ELSE NULL
END
|| ')'
END
|| DECODE (nullable, 'N', ' NOT NULL')
DataTypeWithLength
FROM user_tab_columns
WHERE table_name = 'CONTRACT')
WHERE DataTypeWithLength = 'CHAR(1 Byte)';
To see the internal representation size in bytes you can use:
REGEXP_SUBSTR(DUMP(your_column_name), 'Len=(\d+)\:', 1, 1, 'c', 1 )
DECLARE
c NUMBER;
d NUMBER;
col_cnt INTEGER;
f BOOLEAN;
rec_tab DBMS_SQL.DESC_TAB;
col_num NUMBER;
PROCEDURE print_rec(rec in DBMS_SQL.DESC_REC) IS
BEGIN
DBMS_OUTPUT.NEW_LINE;
DBMS_OUTPUT.PUT_LINE('col_type = '
|| rec.col_type);
DBMS_OUTPUT.PUT_LINE('col_maxlen = '
|| rec.col_max_len);
DBMS_OUTPUT.PUT_LINE('col_name = '
|| rec.col_name);
DBMS_OUTPUT.PUT_LINE('col_name_len = '
|| rec.col_name_len);
DBMS_OUTPUT.PUT_LINE('col_schema_name = '
|| rec.col_schema_name);
DBMS_OUTPUT.PUT_LINE('col_schema_name_len = '
|| rec.col_schema_name_len);
DBMS_OUTPUT.PUT_LINE('col_precision = '
|| rec.col_precision);
DBMS_OUTPUT.PUT_LINE('col_scale = '
|| rec.col_scale);
DBMS_OUTPUT.PUT('col_null_ok = ');
IF (rec.col_null_ok) THEN
DBMS_OUTPUT.PUT_LINE('true');
ELSE
DBMS_OUTPUT.PUT_LINE('false');
END IF;
END;
BEGIN
c := DBMS_SQL.OPEN_CURSOR;
-- YOUR SELECT HERE
DBMS_SQL.PARSE(c, '
SELECT *
FROM table1 a
bable2 b
table3 c
where a.id = b.id
and b.id2 = c.id
', DBMS_SQL.NATIVE);
d := DBMS_SQL.EXECUTE(c);
DBMS_SQL.DESCRIBE_COLUMNS(c, col_cnt, rec_tab);
col_num := rec_tab.first;
IF (col_num IS NOT NULL) THEN
LOOP
print_rec(rec_tab(col_num));
col_num := rec_tab.next(col_num);
EXIT WHEN (col_num IS NULL);
END LOOP;
END IF;
DBMS_SQL.CLOSE_CURSOR(c);
END;
/