All of the tables are collected in the tables attribute of the SQLAlchemy MetaData object. To get a list of the names of those tables:

>>> metadata.tables.keys()
['posts', 'comments', 'users']

If you’re using the declarative extension, then you probably aren’t managing the metadata yourself. Fortunately, the metadata is still present on the baseclass,

>>> Base = sqlalchemy.ext.declarative.declarative_base()
>>> Base.metadata
MetaData(None)

If you are trying to figure out what tables are present in your database, even among the ones you haven’t even told SQLAlchemy about yet, then you can use table reflection. SQLAlchemy will then inspect the database and update the metadata with all of the missing tables.

>>> metadata.reflect(engine)

For Postgres, if you have multiple schemas, you’ll need to loop thru all the schemas in the engine:

from sqlalchemy import inspect
inspector = inspect(engine)
schemas = inspector.get_schema_names()

for schema in schemas:
    print("schema: %s" % schema)
    for table_name in inspector.get_table_names(schema=schema):
        for column in inspector.get_columns(table_name, schema=schema):
            print("Column: %s" % column)