Is SQLAlchemy still recommended if only used for raw sql query?

I use SQLAlchemy for direct queries all the time.

Primary advantage: it gives you the best protection against SQL injection attacks. SQLAlchemy does the Right Thing whatever parameters you throw at it.

I find it works wonders for adjusting the generated SQL based on conditions as well. Displaying a result set with multiple filter controls above it? Just build your query in a set of if/elif/else constructs and you know your SQL will be golden still.

Here is an excerpt from some live code (older SA version, so syntax could differ a little):

# Pull start and end dates from form# ...# Build a constraint if `start` and / or `end` have been set.created = Noneif start and end:    created = sa.sql.between(msg.c.create_time_stamp,         start.replace(hour=0, minute=0, second=0),        end.replace(hour=23, minute=59, second=59))elif start:    created = (msg.c.create_time_stamp >=                start.replace(hour=0, minute=0, second=0))elif end:    created = (msg.c.create_time_stamp <=                end.replace(hour=23, minute=59, second=59))# More complex `from_` object built here, elided for example# [...]# Final query buildquery = sa.select([unit.c.eli_uid], from_obj=[from_])query = query.column(count(msg.c.id).label('sent'))query = query.where(current_store)if created:    query = query.where(created)

The code where this comes from is a lot more complex, but I wanted to highlight the date range code here. If I had to build the SQL using string formatting, I'd probably have introduced a SQL injection hole somewhere as it is much easier to forget to quote values.

After I worked on a small project of mine, I decided to try to just use MySQLDB, without SQL Alchemy.

It works fine and it's quite easy to use, here's an example (I created a small class that handles all the work to the database)

import MySQLdbfrom MySQLdb.cursors import DictCursorclass DatabaseBridge():    def __init__(self, *args, **kwargs):        kwargs['cursorclass'] = DictCursor        self.cnx = MySQLdb.connect (**kwargs)        self.cnx.autocommit(True)        self.cursor = self.cnx.cursor()    def query_all(self, query, *args):        self.cursor.execute(query, *args)        return self.cursor.fetchall()    def find_unique(self, query, *args):        rows = self.query_all(query, *args);        if len(rows) == 1:            return rows[0]        return None    def execute(self, query, params):        self.cursor.execute(query, params)        return self.cursor.rowcount    def get_last_id(self):        return self.cnx.insert_id()    def close(self):        self.cursor.close()        self.cnx.close()database = DatabaseBridge(**{        'user': 'user',        'passwd': 'password',        'db': 'my_db'    })rows = database.query_all("SELECT id, name, email FROM users WHERE is_active = %s AND project = %s", (1, "My First Project"))

(It's a dumb example).

It works like a charm BUT you have to take these into consideration :

• Multithreading is not supported ! It's ok if you don't work with multiprocessing from Python.
• You won't have all the advantages of SQLAlchemy (Database to Class (model) wrapper, Query generation (select, where, order_by, etc)). This is the key point on how you want to work with your database.

But on the other hand, and like SQLAlchemy, there is protections agains't SQL injection attacks :

A basic query would be like this :

cursor.execute("SELECT * FROM users WHERE data = %s" % "Some value") # THIS IS DANGEROUS

But you should do :

cursor.execute("SELECT * FROM users WHERE data = %s", ("Some value",)) # This is secure!

Saw the difference ? Read again ;)

The difference is that I replaced %, by , : We pass the arguments as ... arguments to the execute, and these are escaped. When using %, arguments aren't escaped, enabling SQL Injection attacks!

The final word here is that it depends on your usage and what you plan to do with your project. For me, SQLAlchemy was on overkill (it's a basic shell script !), so MysqlDB was perfect.