postgresql/src/test/regress/sql/text.sql
Tom Lane cc50080a82 Rearrange core regression tests to reduce cross-script dependencies.
The idea behind this patch is to make it possible to run individual
test scripts without running the entire core test suite.  Making all
the scripts completely independent would involve a massive rewrite,
and would probably be worse for coverage of things like concurrent DDL.
So this patch just does what seems practical with limited changes.

The net effect is that any test script can be run after running
limited earlier dependencies:
* all scripts depend on test_setup
* many scripts depend on create_index
* other dependencies are few in number, and are documented in
  the parallel_schedule file.

To accomplish this, I chose a small number of commonly-used tables
and moved their creation and filling into test_setup.  Later scripts
are expected not to modify these tables' data contents, for fear of
affecting other scripts' results.  Also, our former habit of declaring
all C functions in one place is now gone in favor of declaring them
where they're used, if that's just one script, or in test_setup if
necessary.

There's more that could be done to remove some of the remaining
inter-script dependencies, but significantly more-invasive changes
would be needed, and at least for now it doesn't seem worth it.

Discussion: https://postgr.es/m/1114748.1640383217@sss.pgh.pa.us
2022-02-08 15:30:38 -05:00

115 lines
4.0 KiB
SQL

--
-- TEXT
--
SELECT text 'this is a text string' = text 'this is a text string' AS true;
SELECT text 'this is a text string' = text 'this is a text strin' AS false;
-- text_tbl was already created and filled in test_setup.sql.
SELECT * FROM TEXT_TBL;
-- As of 8.3 we have removed most implicit casts to text, so that for example
-- this no longer works:
select length(42);
-- But as a special exception for usability's sake, we still allow implicit
-- casting to text in concatenations, so long as the other input is text or
-- an unknown literal. So these work:
select 'four: '::text || 2+2;
select 'four: ' || 2+2;
-- but not this:
select 3 || 4.0;
/*
* various string functions
*/
select concat('one');
select concat(1,2,3,'hello',true, false, to_date('20100309','YYYYMMDD'));
select concat_ws('#','one');
select concat_ws('#',1,2,3,'hello',true, false, to_date('20100309','YYYYMMDD'));
select concat_ws(',',10,20,null,30);
select concat_ws('',10,20,null,30);
select concat_ws(NULL,10,20,null,30) is null;
select reverse('abcde');
select i, left('ahoj', i), right('ahoj', i) from generate_series(-5, 5) t(i) order by i;
select quote_literal('');
select quote_literal('abc''');
select quote_literal(e'\\');
-- check variadic labeled argument
select concat(variadic array[1,2,3]);
select concat_ws(',', variadic array[1,2,3]);
select concat_ws(',', variadic NULL::int[]);
select concat(variadic NULL::int[]) is NULL;
select concat(variadic '{}'::int[]) = '';
--should fail
select concat_ws(',', variadic 10);
/*
* format
*/
select format(NULL);
select format('Hello');
select format('Hello %s', 'World');
select format('Hello %%');
select format('Hello %%%%');
-- should fail
select format('Hello %s %s', 'World');
select format('Hello %s');
select format('Hello %x', 20);
-- check literal and sql identifiers
select format('INSERT INTO %I VALUES(%L,%L)', 'mytab', 10, 'Hello');
select format('%s%s%s','Hello', NULL,'World');
select format('INSERT INTO %I VALUES(%L,%L)', 'mytab', 10, NULL);
select format('INSERT INTO %I VALUES(%L,%L)', 'mytab', NULL, 'Hello');
-- should fail, sql identifier cannot be NULL
select format('INSERT INTO %I VALUES(%L,%L)', NULL, 10, 'Hello');
-- check positional placeholders
select format('%1$s %3$s', 1, 2, 3);
select format('%1$s %12$s', 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12);
-- should fail
select format('%1$s %4$s', 1, 2, 3);
select format('%1$s %13$s', 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12);
select format('%0$s', 'Hello');
select format('%*0$s', 'Hello');
select format('%1$', 1);
select format('%1$1', 1);
-- check mix of positional and ordered placeholders
select format('Hello %s %1$s %s', 'World', 'Hello again');
select format('Hello %s %s, %2$s %2$s', 'World', 'Hello again');
-- check variadic labeled arguments
select format('%s, %s', variadic array['Hello','World']);
select format('%s, %s', variadic array[1, 2]);
select format('%s, %s', variadic array[true, false]);
select format('%s, %s', variadic array[true, false]::text[]);
-- check variadic with positional placeholders
select format('%2$s, %1$s', variadic array['first', 'second']);
select format('%2$s, %1$s', variadic array[1, 2]);
-- variadic argument can be array type NULL, but should not be referenced
select format('Hello', variadic NULL::int[]);
-- variadic argument allows simulating more than FUNC_MAX_ARGS parameters
select format(string_agg('%s',','), variadic array_agg(i))
from generate_series(1,200) g(i);
-- check field widths and left, right alignment
select format('>>%10s<<', 'Hello');
select format('>>%10s<<', NULL);
select format('>>%10s<<', '');
select format('>>%-10s<<', '');
select format('>>%-10s<<', 'Hello');
select format('>>%-10s<<', NULL);
select format('>>%1$10s<<', 'Hello');
select format('>>%1$-10I<<', 'Hello');
select format('>>%2$*1$L<<', 10, 'Hello');
select format('>>%2$*1$L<<', 10, NULL);
select format('>>%2$*1$L<<', -10, NULL);
select format('>>%*s<<', 10, 'Hello');
select format('>>%*1$s<<', 10, 'Hello');
select format('>>%-s<<', 'Hello');
select format('>>%10L<<', NULL);
select format('>>%2$*1$L<<', NULL, 'Hello');
select format('>>%2$*1$L<<', 0, 'Hello');