pablormira / sql_formatter Goto Github PK

something like

select asdf
from table; /* some comment */

would be wrongly separated from the next query like

select asdf
from table; 

/* some comment */

Currently, case when will be formatted without breaklines like

SELECT field1,
       case when field2 < 1 then 1 when field2 >=1 and field2 < 5 then 2 else 3 end as field2_cat
... 

and I would like to have some indentation like

SELECT field1,
       case when field2 < 1 then 1
            when field2 >=1 and field2 < 5 then 2 
            else 3 end as field2_cat
... 

Add validation for case when ... end correctly written. Sometimes I forgot the end

Test interaction to pre-commit to check that the CLI can be used to automatically format SQL files with pre-commit

Up to now we also format SQL comments with --

Release first version of the package and publish to pipy and anaconda

All done! seems misleading because it could also mean everything is already formatted before program execution

Describe the bug

There is no formatting for PARTITION BY clausels, e.g.

SELECT asdf,
       row_number() OVER (PARTITI
       ON BY qwer
ORDER BY asdf) 

The newline for ON in PARITITION is rather a general problem as in #26

Expected behavior

SELECT asdf,
       row_number() OVER (PARTITION BY qwer
                          ORDER BY asdf) 

Specially something like

select asdf, qwer
/* bla bla */
/* bla bla 2 */
asdf2
from table1

will be marked as

select asdf, qwer
[CS]/* bla bla */[C]  [CI] /* bla bla 2 */[C]
asdf2
from table1

This is wrong: [CI] is only for newlines within multiline comments

Allow the user to mark queries somewhat to not format them, e.g. using something like /*formatter-skip*/

This is the case if there is some special keyword in between like COMMENT, e.g.:

CREATE OR REPLACE VIEW my_view COMMENT='blabla' AS

in that case we wrongly get

CREATE OR REPLACE VIEW MY_VIEW comment='blabla' AS

Expected behavior: Untouched

Something like

sql-formatter -r *.sql

with default False

e.g.

ON a.field1 = b.field1 and a.field and a.field2 = b.field2

to

ON a.field1 = b.field1 and 
   a.field2 = b.field2

Describe the bug

sql-formatter sql_file.sql introduces 3 newlines at the beginning of file after formatting

Specially if somebody wants to install from Source:

• Installing first fastcore
• Then latest release

via pip

Describe the bug

sql-formatter sql_file.sql separates function arguments in SELECT as if they were normal fields, e.g.

SELECT asdf,
       substr(asdf,
       3,
       1) as substr_asdf  

Expected behavior

SELECT asdf,
       substr(asdf, 3, 1) as substr_asdf  

Setup: how to install development environment and so on

Describe the bug

ON formatting function formats also words containing "on", like function. Consider the following query:

SELECT asdf,
               my_function
FROM   ...

The formatter then does

SELECT asdf,
               my_functi
ON
FROM   ...

Expected behavior

SELECT asdf,
               my_function
FROM    ...

Closely related to #23

Describe the bug

ORDER BY is not being formatted

Expected behavior

ORDER BY should be formatted like GROUP BY

check_sql_query() should identify create .* table

If there are no balanced parenthesis then abort formatting

Related to #52

E.g.

select asdf,
       case when asdf >= 1 and asdf <= 10 and substr(qwer, 1, 2) = 'abc' and substr(qwer, 3, 2) = 'qwerty' then 1
            else 0 end as case_field,
       asdf2
from table1

should be formatted as

select asdf,
       case when asdf >= 1 and 
                 asdf <= 10 and 
                 substr(qwer, 1, 2) = 'abc' and 
                 substr(qwer, 3, 2) = 'qwerty' then 1
            else 0 end as case_field,
       asdf2
from table1

• Add validation for overlapping queries:

CREATE TABLE asdf AS
SELECT qwer
FROM table1 <- no semicolon

CREATE TABLE asdf2 AS
...

Describe the bug

Only SELECT is formatted. SELECT DISTINCT does not get proper indentation

Something like

CREATE TASK bla bla
CREATE TABLE my_table AS

should not return a warning.

Only CREATE TABLE / VIEW twice should return a warning

something like

select asdf,
case when asdf = 1 and -- some comment
asdf != 2 then 1 else 0 end as some_field
from table1

would be formatted like

select asdf,
       case when asdf = 1 and -- some comment
            asdf != 2 then 1 else 0 end as some_field
from table1

and not better like

select asdf,
       case when asdf = 1 and -- some comment
                 asdf != 2 then 1 else 0 end as some_field
from table1

E.g.

format_where() is not robust against keywords and and or in comments

Kind of reformatting afterwards

Specially useful for SELECT and ON lines