1. SQLAlchemy Using python to interact with SQL databases
Eowyn Baughman, Oct

2. PyData & SQL Python is popular in data science
SQL is widely used to interact with databases– e.g. PostGRES, Oracle
Query (SELECT)
Conditionals (WHERE)
Sorting (GROUP BY)
Combining data (JOIN)
Update the database (INSERT)
SQLAlchemy is a Python package to natively interact with databases within Python

3. SQLAlchemy Tutorial Step-by-step tutorial With Examples:
“You almost never need to write SQL by hand.”
Goes much more into detail, explores more advanced topics
Full Documentation:

4. SQLAlchemy: Establish dataset
db = create_engine(‘adatabase.db’)
Open connection to database
The resulting object has methods specific to the type of database
metadata = BoundMetaData(db)
Create an object that knows what the columns of the data are, what types of data they contain, etc.
Used to re-create analogous data structures in Python
users = Table('users', metadata, autoload=True)
Figure out automagically what the table in the database looks like and load it for Python to work with

5. SQLAlchemy: Building SQL Statements
Main idea: Create “SQL statement objects” instead of writing SQL
Then call its “execute” method that belongs to the object
Table objects have methods like:
users.select() to pull all or some of the data to your local table
users.insert() to update data in your local table
Both of which can accept conditional statements
Can also uses join(table1,table2).select() or outerjoin(table1,table2).select() to work on multiple tables

6. Mapping objects to the database
Saving your data back to the original database
Create empty classes to serve as data classes
“map” your data tables onto the empty data classes, which then take on the correct column names and types
Create a “session” object to keep track of data classes
Select (load) data from database into data classes
Modify them however you like
“Flush” the session to push changes to the database

7. Example of Mapping from sqlalchemy import *
db = create_engine('sqlite:///joindemo.db’)
metadata = BoundMetaData(db)
users = Table('users', metadata, autoload=True)
s = Table(' s', metadata, autoload=True)
# These are the empty classes that will become our data classes
class User(object):
pass
class (object):
usermapper = mapper(User, users)
mapper = mapper( , s)
session = create_session()
mary = session.query(User).selectfirst(users.c.name=='Mary’)
#you are forced to be explicit about which table, and #what kind of object, you are selecting data from. “In #the face of ambiguity, refuse the temptation to #guess”
mary.age += 1
session.flush()