From unix to Sql - sql

I am making a shell script where I am reading inputs from one file. File contains data
123
1234
121
I am reading the inputs from this file using while read line do condition and putting all inputs in SQL statements.Now in my shell script i am going on SQL Prompt and running some queries. In one condition, I am using EXECUTE IMMEDIATE STATEMENT in SQL.
as
EXECUTE IMMEDIATE 'CREATE TABLE BKP_ACDAGENT4 as SELECT * FROM BKP_ACDAGENT WHERE DATASOURCEAGENTID IN ('123','1234','121')';
I want this to be execute, but somehow its not working.
Can anyone help me in executing it?

You need to escape the single quotes which you have used for the predicates in your IN list, that is the single quotes in
WHERE DATASOURCEAGENTID IN ('123','1234','121')';
are causing the issue here. You need to escape the single quotes using two single quotes
EXECUTE IMMEDIATE 'CREATE TABLE BKP_ACDAGENT4 as SELECT * FROM BKP_ACDAGENT WHERE DATASOURCEAGENTID IN (''123'',''1234'',''121'')';
The above will work on all Oracle version.
If you're one Oracle 10g or above, you can use q keyword
EXECUTE IMMEDIATE q'[CREATE TABLE BKP_ACDAGENT4 as SELECT * FROM BKP_ACDAGENT WHERE DATASOURCEAGENTID IN ('123','1234','121')]';

Related

Pasting multi-line queries into BigQuery SQL shell

I'm running the BigQuery command line shell and I'm not able to successfully run multi-line queries (aka queries with line breaks) because whenever I paste the query into the shell each line gets run individually instead of the whole thing together.
For example,
select * from table
works fine because it is in one line but if I try to run
select
*
from
table
it does not work because each line gets run separately.
Is there any way to get this to work?
The query command creates a query job that runs the supplied SQL query. in the documentation Using the bq command-line tool you can find some examples like:
bq query --nouse_legacy_sql \
'SELECT
COUNT(*)
FROM
`bigquery-public-data`.samples.shakespeare'
Honestly I too couldn't find decent help on this, so I created a bit of a workaround using the CONCAT function. Notice that I have a space before the FROM clause. Hope his helps.
DECLARE table_name STRING DEFAULT '20220214';
DECLARE query STRING;
SET query = CONCAT("""select user_pseudo_id, user_id""",
""" FROM `app.analytics.events_""" || table_name || """` where user_id IS NOT NULL group by user_pseudo_id, user_id""");
EXECUTE IMMEDIATE query;

Define a VIEW in Oracle without using CREATE

I do not have sufficient privileges to use a CREATE statement, I can only SELECT. I have a script that has three distinct parts and calculations that need to run and each one references the same complicated WITH statement selection that redundantly clutters the code and is a pain to maintain in three separate locations.
I have tried creating temp tables and views, but again, privileges do not support. Is there a way using either SQL or PL/SQL syntax to define my WITH statement ONCE without using CREATE, and then reference it like I would any other table? Example:
--Define the temp table
WITH tempview AS (SELECT .... FROM ...);
--First query
SELECT ... FROM tempview;
/
--Second query
SELECT ... FROM tempview;
/
--Third query
SELECT ... FROM tempview;
/
Getting the correct permissions and creating permanent objects is the best approach. It sounds like this view would only be used in a single script, which doesn't necessarily make it any less valid to create it, but you might find it harder to justify depending on your DBA and policies. It's certainly worth trying that approach, as #DCookie suggested.
If that fails then there may be hacky workarounds, depending on the client you will run this script in.
For instance, in SQL*Plus it's possible to abuse substitution variables to get something close to what you describe. This uses the define command to create a substitution variable that contains the 'view' query, and then uses that variable inside a WITH clause. (You can't replace the entire with like this, but it's maybe clearer like this anyway). I'm used a trivial dummy query:
define tempview_query = 'SELECT * -
FROM dual -
UNION ALL -
SELECT * -
FROM dual'
WITH tempview AS (&tempview_query)
SELECT * FROM tempview;
WITH tempview AS (&tempview_query)
SELECT * FROM tempview;
When the script is run the output produced is:
D
-
X
X
2 rows selected.
D
-
X
X
2 rows selected.
I've also executed set verify off to hide the substitutions, but turning it on might be instructive to see what's happening.
Notice the dashes at the end of each line of the query; that's the continuation character, and as the define docs mention:
If the value of a defined variable extends over multiple lines (using the SQL*Plus command continuation character), SQL*Plus replaces each continuation character and carriage return with a space.
so the 'new' query shown by set verify on will have your entire view query on a single line (if you display it). It's feasible that with a long enough query you'd hit some line length limit but hopefully you won't reach that point (except you did; see below).
You can do the same thing in SQL Developer, but there the continuation needs to use two dashes, so:
define tempview_query = 'SELECT * --
FROM dual --
UNION ALL --
SELECT * --
FROM dual'
except it isn't quite the same as the continuation in SQL*Plus; here the define has to end with a dash, but it is not replaced in the way the SQL*Plus docs describe - so with a single dash the define works but the query ends up invalid. (At least in 4.2.0; possibly a bug...) By using two dashes the multi-line define still works, the dashes remain part of the query, but they're treated as comment markers; so they make the substituted query look odd (again, if you display it) but don't stop it working. You won't notice with set verify off unless someone looks in v$sql.
If your query exceeds 240 characters - which is rather likely unless it's trivial enough to repeat anyway - you'll hit something like:
string beginning "'SELECT * ..." is too long. maximum size is 240 characters.
Both SQL*Plus and SQL Developer allow you to set a substitution variable from a query, using the column ... new_value command:
column tempalias new_value tempview_query
set termout off
select q'[SELECT *
FROM dual
UNION ALL
SELECT *
FROM dual]'
FROM dual;
set termout on
The query selects the text of your view query as a string; I've used the alternative quoting mechanism, with [] as the delimiters, so you don't have to escape any single quotes in the view query. (You need to pick a delimiter that can't appear in the query too, of course). Also note that you don't need the line continuation character any more.
The text literal that query generates is aliased as tempalias. The column command sets the tempview_query substitution variable to whatever that aliased column expression contains. Using the substitution variable is then the same as in the previous examples.
WITH tempview AS (&tempview_query)
SELECT * FROM tempview;
The set termout lines just hide that generating query; you can temporarily omit the off line to see what the query produces, and that it does exactly match the view query you expected.
Other clients might have similar mechanisms, but those are the only two I'm really familiar with. I should probably also reiterate that this is a bit of a hack, and not something I'd necessarily recommend...
Another trick with SQL*Plus is to import code from a second SQL script. You can do this with the # command (making sure to put the # at the very start of a line), e.g.:
tempview.sql
WITH tempview AS (SELECT .... FROM ...)
(notice there is no ending semicolon ; here, and make sure you either don't have a blank line at the end of the file or set sqlblanklines on)
main.sql
--First query
#tempview.sql
SELECT ... FROM tempview
/
--Second query
#tempview.sql
SELECT ... FROM tempview
/
--Third query
#tempview.sql
SELECT ... FROM tempview
/

How to execute multiple Oracle sql queries from VBscript?

I have a database object called objdb. From this object it seems that I can only execute one sql query at a time. if I insert a semi-colon in the end,it gives me a
ORA-0911 invalid character error.
For example, if I run
objdb.executeSQl("select * from table1; select * from table2")
This gives me a error.
How can I run multiple queries from a single ExecuteSQl method ?

Nested sql statement in talend (updated)

I am trying to use nested sql in postgresql tPostgresqlRow_1.
But I receive an error.
The following sql runs okay if I run it in PgAdmin.
But in Talend I receive an error.
I am getting the max date from one table and updating the column in another table.
update "STG_magento_de"."configuration_table"
set created_at=(select MAX(created_at) from "STG_magento_de"."sales_flat_order_test")
where table_name='sales_flat_order_test'
The tPostgresqlRow component expects a Java string containing the SQL statement.
The most likely problem is that you have unescaped quotes in the statement. That works fine in pgAdmin because it is valid. To pass the same statement from Talend, you'll have to escape all the quotes in the statement itself. Alternatively, you could try removing the double quotes from the SQL statement.
Remember to enclose the whole thing in quotes, so that it is a proper Java string.
create or replace function eliminarComillas (text) returns text AS $$
select replace(replace(replace($1,'**\"**','"'),'‘','‘'),'’','’');
$$ language sql;"
Following sql with escape quotes worked.
"update \"STG_magento_de\".\"configuration_table\"
set created_at=(select MAX(created_at) from \"STG_magento_de\".\"sales_flat_order_test\")
where table_name='sales_flat_order_test'"

Regex to split SQL script but ignore splitting GO under commented script

I'm trying to parse a sort of big SQL script containing commands like create functions and store procedures. I want to split the file in strings whenever I find a GO statement (I want to execute it using ordinary ado.net instead of SMO).
The problem is that I haven't find a suitable regular expression for that so far. Using a simple \bGO\b ignoring case will split it. But will also split all go(s) inside a comment like
/*****************************\
sql statement 1
GO
sql statement 2
GO
\****************************/
My requirement is : Do not split the script if it is under comment even though the script contains GO statement. Suppose my entire script is as below :
sql statement 1
GO
/*****************************\
sql statement 2
GO
sql statement 3
GO
\****************************/
sql statement 4
Expected output should be like
First command :
sql statement 1
Second command :
/*****************************\
sql statement 2
GO
sql statement 3
GO
\****************************/
sql statement 4
Have any idea on this ?
Thanks in advance. :)
You can remove all the comments and then split by GO:
/\/\**\\[^\\]*?\\\**\// # match the comment
\/\**\\ # matches /*****\
[^\\]*? # any text within the comments
\\\**\/ # matches \*****/
Remove above finds and split the result by GO.