paetzke/format-sql

Makes your SQL readable.

format-sql's Issues

TypeError: write() argument must be str, not None

I run the following command

format-sql a_sql_file.sql

but got the following error

Traceback (most recent call last):
  File "/usr/local/bin/format-sql", line 11, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.5/dist-packages/format_sql/main.py", line 83, in main
    _write_back(filename, lines, args.dry_run)
  File "/usr/local/lib/python3.5/dist-packages/format_sql/main.py", line 148, in _write_back
    f.write(lines)
TypeError: write() argument must be str, not None

not WITH token

exception when using WITH token in function
def _parse(toks):

raising InvalidSQL()

TypeError: expected a character buffer object

  File "/usr/local/bin/format-sql", line 11, in <module>                                                                                                      
    sys.exit(main())                                                                                                                                          
  File "/usr/local/lib/python2.7/dist-packages/format_sql/main.py", line 83, in main                                                                          
    _write_back(filename, lines, args.dry_run)                                                                                                                
  File "/usr/local/lib/python2.7/dist-packages/format_sql/main.py", line 148, in _write_back                                                                  
    f.write(lines)                                                                                                                                            
TypeError: expected a character buffer object       

I've tried this with several files, including one ruby file with embedded SQL and then just with the SQL itself. Any insight into this? I'm just surprised that others aren't getting the same error.

Support psycopg2 SQL `{name}` format placeholders

Currently these are being unwrapped, so FROM {schema_name}.table_name becomes schema_name table_name.

AND (and more) on their own line

Would it possible to have a switch to have the example in the README parsed like this

    SELECT
        country,
        product,
        SUM(profit)
    FROM
        sales
        LEFT JOIN x ON  -- note the difference here
            x.id = sales.k  -- possibly with more AND's, each AND in its line
    GROUP BY
        country,
        product
    HAVING
        f > 7
        AND      -- and here
        fk = 9
    LIMIT 5;

RuntimeError: generator raised StopIteration

I want to format this sql query but it's showing runtime error
=> https://github.com/yugyesh/song_play_analysis/blob/dev/sql_queries.py

:: cast replaced by empty space

See example, where did the :: go?

from format_sql import format_sql
print('\n'.join(format_sql('SELECT cast::jsonb')))
>>> SELECT
>>>    cast jsonb

Can't parse quoted fields (postgres)

The parser will fail to parse a query like select "id" from table. This is DBMS specific, so this may actually be a parse error for other databases, but its how postgres does quoting of DDL-level stuff like columns and table names.

Parse error in where clausel when using operations on a term

See this example snippet:

>>> format_sql.format_sql('''select a.a from aaa a where ((a.b+1) = 2);''')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/koerner/miniconda3/envs/tests-sql/lib/python3.6/site-packages/format_sql/shortcuts.py", line 20, in format_sql
    parsed = list(parse(tokens))
  File "/home/koerner/miniconda3/envs/tests-sql/lib/python3.6/site-packages/format_sql/parser.py", line 762, in parse
    for statement, unused_count in _parse(toks):
  File "/home/koerner/miniconda3/envs/tests-sql/lib/python3.6/site-packages/format_sql/parser.py", line 756, in _parse
    statement, count = func(toks)
  File "/home/koerner/miniconda3/envs/tests-sql/lib/python3.6/site-packages/format_sql/parser.py", line 596, in _parse_where
    conditions, j = _parse_conditions(toks[1:])
  File "/home/koerner/miniconda3/envs/tests-sql/lib/python3.6/site-packages/format_sql/parser.py", line 701, in _parse_conditions
    raise InvalidCondition()
format_sql.parser.InvalidCondition

It seems that all (most?) queries with the following conditions <table>.<column> + 1 = <some other value> fail to parse:

select a.a from aaa a where a.b+1 = 2;

Lack of documentation on how to use from python

This package is part of the pip repository manager for python, fine, but how do you use it from a python script? There is only information on how to run it as a binary.

Parser broken for "INSERT INTO ... SELECT ... FROM ..." queries

Since this is standard (ANSI) SQL it should be supported :)

You can test with:

INSERT INTO spam
SELECT *
FROM eggs

Allow SQL file (i.e. SQL not embedded in Python)

Hi (thanks for the awesome software),

would the above be possible?