I have a python list, say l
l = [1,5,8]
I want to write a sql query to get the data for all the elements of the list, say
select name from students where id = |IN THE LIST l|
How do I accomplish this?
Answers so far have been templating the values into a plain SQL string. That's absolutely fine for integers, but if we wanted to do it for strings we get the escaping issue.
Here's a variant using a parameterised query that would work for both:
placeholder= '?' # For SQLite. See DBAPI paramstyle.
placeholders= ', '.join(placeholder for unused in l)
query= 'SELECT name FROM students WHERE id IN (%s)' % placeholders
cursor.execute(query, l)
The SQL you want is
select name from studens where id in (1, 5, 8)
If you want to construct this from the python you could use
l = [1, 5, 8]
sql_query = 'select name from studens where id in (' + ','.join(map(str, l)) + ')'
The map function will transform the list into a list of strings that can be glued together by commas using the str.join method.
Alternatively:
l = [1, 5, 8]
sql_query = 'select name from studens where id in (' + ','.join((str(n) for n in l)) + ')'
if you prefer generator expressions to the map function.
UPDATE: S. Lott mentions in the comments that the Python SQLite bindings don't support sequences. In that case, you might want
select name from studens where id = 1 or id = 5 or id = 8
Generated by
sql_query = 'select name from studens where ' + ' or '.join(('id = ' + str(n) for n in l))
string.join the list values separated by commas, and use the format operator to form a query string.
myquery = "select name from studens where id in (%s)" % ",".join(map(str,mylist))
(Thanks, blair-conrad)
I like bobince's answer:
placeholder= '?' # For SQLite. See DBAPI paramstyle.
placeholders= ', '.join(placeholder for unused in l)
query= 'SELECT name FROM students WHERE id IN (%s)' % placeholders
cursor.execute(query, l)
But I noticed this:
placeholders= ', '.join(placeholder for unused in l)
Can be replaced with:
placeholders= ', '.join(placeholder*len(l))
I find this more direct if less clever and less general. Here l is required to have a length (i.e. refer to an object that defines a __len__ method), which shouldn't be a problem. But placeholder must also be a single character. To support a multi-character placeholder use:
placeholders= ', '.join([placeholder]*len(l))
If you're using PostgreSQL with the Psycopg2 library you can let its tuple adaption do all the escaping and string interpolation for you, e.g:
ids = [1,2,3]
cur.execute(
"SELECT * FROM foo WHERE id IN %s",
[tuple(ids)])
i.e. just make sure that you're passing the IN parameter as a tuple. if it's a list you can use the = ANY array syntax:
cur.execute(
"SELECT * FROM foo WHERE id = ANY (%s)",
[list(ids)])
note that these both will get turned into the same query plan so you should just use whichever is easier. e.g. if your list comes in a tuple use the former, if they're stored in a list use the latter.
Solution for @umounted answer, because that broke with a one-element tuple, since (1,) is not valid SQL.:
>>> random_ids = [1234,123,54,56,57,58,78,91]
>>> cursor.execute("create table test (id)")
>>> for item in random_ids:
cursor.execute("insert into test values (%d)" % item)
>>> sublist = [56,57,58]
>>> cursor.execute("select id from test where id in %s" % str(tuple(sublist)).replace(',)',')'))
>>> a = cursor.fetchall()
>>> a
[(56,), (57,), (58,)]
Other solution for sql string:
cursor.execute("select id from test where id in (%s)" % ('"'+'", "'.join(l)+'"'))
To run a select from where field is in list of strings (instead of int), as per this question use repr(tuple(map(str, l))). Full example:
l = ['a','b','c']
sql = f'''
select name
from students
where id in {repr(tuple(map(str, l)))}
'''
print(sql)
Returns:
select name from students where id in ('a', 'b', 'c')
For a list of dates in Oracle, this worked
dates_str = ','.join([f'DATE {repr(s)}' for s in ['2020-11-24', '2020-12-28']])
dates_str = f'({dates_str})'
sql_cmd = f'''
select *
from students
where
and date in {dates_str}
'''
Returns:
select * from students where and date in (DATE '2020-11-24',DATE '2020-12-28')
This uses parameter substitution and takes care of the single value list case:
l = [1,5,8]
get_operator = lambda x: '=' if len(x) == 1 else 'IN'
get_value = lambda x: int(x[0]) if len(x) == 1 else x
query = 'SELECT * FROM table where id ' + get_operator(l) + ' %s'
cursor.execute(query, (get_value(l),))