We Keep Coding

How to create entities dynamically with PonyORM?

I would like to create DB Entities in Pony ORM by a factory method so avoid code duplication for similar tables.
Here is my not fully working minimal example:
from pony.orm import *
def factory(db, tablename):
class TableTemplate(db.Entity):
_table_ = tablename
first_name = Required(str)
last_name = Required(str)
composite_index(first_name, last_name)
return TableTemplate
db = Database(provider='sqlite', filename=':memory:')
Table1 = factory(db, "TABLE_1")
# the following line produces the exception:
# pony.orm.core.ERDiagramError: Entity TableTemplate already exists
Table2 = factory(db, "TABLE_2")
db.generate_mapping(create_tables=True)
with db_session:
Table1(first_name="foo", last_name="bar")
The exception could be circumvented by creating the class with a dynamic name using type, but this does not work well with composite_index...
Is there a good way to have a table factory with Pony ORM?

Here's my take on your class factory:
def factory(db, tablename):
fields = {
'_table': tablename,
'first_name': Required(str)
# rest of the fields
}
table_template = type(tablename.capitalize(),(db.Entity,),fields)
return table_template
This will create a class by capitalizing the name in tablename and set the descriptors. I'm not sure about metaclasses though
UPDATE ON THE composite_index ISSUE
composite_index uses some pretty obscure features by calling this method:
def _define_index(func_name, attrs, is_unique=False):
if len(attrs) < 2: throw(TypeError,
'%s() must receive at least two attributes as arguments' % func_name)
cls_dict = sys._getframe(2).f_locals
indexes = cls_dict.setdefault('_indexes_', [])
indexes.append(Index(*attrs, is_pk=False, is_unique=is_unique))
A little experimentation tells me you mayb be able to perform the same by adding the field yourself. So that would make our factory fields variable looks like this:
fields = {
'_table': tablename,
'first_name': Required(str),
'_indexes_':[Index(('first_name','last_name'),is_pk=False,is_unique=False)]
# rest of the fields
}
Give it a try and let me know.
UPDATE ON OP EXPERIMENT
The final code would be something like this:
from pony.orm import *
from pony.orm.core import Index
def factory(db, tablename):
fields = {
'_table': tablename,
'first_name': Required(str)
# rest of the fields
}
fields['_indexes_'] = [Index(fields['first_name'],fields['last_name'],is_pk=False,is_unique=False)]
table_template = type(tablename.capitalize(),(db.Entity,),fields)
return table_template

Dynamic Datasets and SQLAlchemy

I am refactoring some old SQLite3 SQL statements in Python into SQLAlchemy. In our framework, we have the following SQL statements that takes in a dict with certain known keys and potentially any number of unexpected keys and values (depending what information was provided).
import sqlite3
import sys
def dict_factory(cursor, row):
d = {}
for idx, col in enumerate(cursor.description):
d[col[0]] = row[idx]
return d
def Create_DB(db):
# Delete the database
from os import remove
remove(db)
# Recreate it and format it as needed
with sqlite3.connect(db) as conn:
conn.row_factory = dict_factory
conn.text_factory = str
cursor = conn.cursor()
cursor.execute("CREATE TABLE [Listings] ([ID] INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL UNIQUE, [timestamp] REAL NOT NULL DEFAULT(( datetime ( 'now' , 'localtime' ) )), [make] VARCHAR, [model] VARCHAR, [year] INTEGER);")
def Add_Record(db, data):
with sqlite3.connect(db) as conn:
conn.row_factory = dict_factory
conn.text_factory = str
cursor = conn.cursor()
#get column names already in table
cursor.execute("SELECT * FROM 'Listings'")
col_names = list(map(lambda x: x[0], cursor.description))
#check if column doesn't exist in table, then add it
for i in data.keys():
if i not in col_names:
cursor.execute("ALTER TABLE 'Listings' ADD COLUMN '{col}' {type}".format(col=i, type='INT' if type(data[i]) is int else 'VARCHAR'))
#Insert record into table
cursor.execute("INSERT INTO Listings({cols}) VALUES({vals});".format(cols = str(data.keys()).strip('[]'),
vals=str([data[i] for i in data]).strip('[]')
))
#Database filename
db = 'test.db'
Create_DB(db)
data = {'make': 'Chevy',
'model' : 'Corvette',
'year' : 1964,
'price' : 50000,
'color' : 'blue',
'doors' : 2}
Add_Record(db, data)
data = {'make': 'Chevy',
'model' : 'Camaro',
'year' : 1967,
'price' : 62500,
'condition' : 'excellent'}
Add_Record(db, data)
This level of dynamicism is necessary because there's no way we can know what additional information will be provided, but, regardless, it's important that we store all information provided to us. This has never been a problem because in our framework, as we've never expected an unwieldy number of columns in our tables.
While the above code works, it's obvious that it's not a clean implementation and thus why I'm trying to refactor it into SQLAlchemy's cleaner, more robust ORM paradigm. I started going through SQLAlchemy's official tutorials and various examples and have arrived at the following code:
from sqlalchemy import Column, String, Integer
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
Base = declarative_base()
class Listing(Base):
__tablename__ = 'Listings'
id = Column(Integer, primary_key=True)
make = Column(String)
model = Column(String)
year = Column(Integer)
engine = create_engine('sqlite:///')
session = sessionmaker()
session.configure(bind=engine)
Base.metadata.create_all(engine)
data = {'make':'Chevy',
'model' : 'Corvette',
'year' : 1964}
record = Listing(**data)
s = session()
s.add(record)
s.commit()
s.close()
and it works beautifully with that data dict. Now, when I add a new keyword, such as
data = {'make':'Chevy',
'model' : 'Corvette',
'year' : 1964,
'price' : 50000}
I get a TypeError: 'price' is an invalid keyword argument for Listing error. To try and solve the issue, I modified the class to be dynamic, too:
class Listing(Base):
__tablename__ = 'Listings'
id = Column(Integer, primary_key=True)
make = Column(String)
model = Column(String)
year = Column(Integer)
def __checker__(self, data):
for i in data.keys():
if i not in [a for a in dir(self) if not a.startswith('__')]:
if type(i) is int:
setattr(self, i, Column(Integer))
else:
setattr(self, i, Column(String))
else:
self[i] = data[i]
But I quickly realized this would not work at all for several reasons, e.g. the class was already initialized, the data dict cannot be fed into the class without reinitializing it, it's a hack more than anything, et al.). The more I think about it, the less obvious the solution using SQLAlchemy seems to me. So, my main question is, how do I implement this level of dynamicism using SQLAlchemy?
I've researched a bit to see if anyone has a similar issue. The closest I've found was Dynamic Class Creation in SQLAlchemy but it only talks about the constant attributes ("tablename" et al.). I believe the unanswered https://stackoverflow.com/questions/29105206/sqlalchemy-dynamic-attribute-change may be asking the same question. While Python is not my forte, I consider myself a highly skilled programmer (C++ and JavaScript are my strongest languages) in the context scientific/engineering applications, so I may not hitting the correct Python-specific keywords in my searches.
I welcome any and all help.

class Listing(Base):
__tablename__ = 'Listings'
id = Column(Integer, primary_key=True)
make = Column(String)
model = Column(String)
year = Column(Integer)
def __init__(self,**kwargs):
for k,v in kwargs.items():
if hasattr(self,k):
setattr(self,k,v)
else:
engine.execute("ALTER TABLE %s AD COLUMN %s"%(self.__tablename__,k)
setattr(self.__class__,Column(k, String))
setattr(self,k,v)
might work ... maybe ... I am not entirely sure I did not test it
a better solution would be to use a relational table
class Attribs(Base):
listing_id = Column(Integer,ForeignKey("Listing"))
name = Column(String)
val = Column(String)
class Listing(Base):
id = Column(Integer,primary_key = True)
attributes = relationship("Attribs",backref="listing")
def __init__(self,**kwargs):
for k,v in kwargs.items():
Attribs(listing_id=self.id,name=k,value=v)
def __str__(self):
return "\n".join(["A LISTING",] + ["%s:%s"%(a.name,a.val) for a in self.attribs])
another solution would be to store json
class Listing(Base):
__tablename__ = 'Listings'
id = Column(Integer, primary_key=True)
data = Column(String)
def __init__(self,**kwargs):
self.data = json.dumps(kwargs)
self.data_dict = kwargs
the best solution would be to use a no-sql key,value store (maybe even just a simple json file? or perhaps shelve? or even pickle I guess)

sqlalchemy: union query few columns from multiple tables with condition

I'm trying to adapt some part of a MySQLdb application to sqlalchemy in declarative base. I'm only beginning with sqlalchemy.
The legacy tables are defined something like:
student: id_number*, semester*, stateid, condition, ...
choice: id_number*, semester*, choice_id, school, program, ...
We have 3 tables for each of them (student_tmp, student_year, student_summer, choice_tmp, choice_year, choice_summer), so each pair (_tmp, _year, _summer) contains information for a specific moment.
select *
from `student_tmp`
inner join `choice_tmp` using (`id_number`, `semester`)
My problem is the information that is important to me is to get the equivalent of the following select:
SELECT t.*
FROM (
(
SELECT st.*, ct.*
FROM `student_tmp` AS st
INNER JOIN `choice_tmp` as ct USING (`id_number`, `semester`)
WHERE (ct.`choice_id` = IF(right(ct.`semester`, 1)='1', '3', '4'))
AND (st.`condition` = 'A')
) UNION (
SELECT sy.*, cy.*
FROM `student_year` AS sy
INNER JOIN `choice_year` as cy USING (`id_number`, `semester`)
WHERE (cy.`choice_id` = 4)
AND (sy.`condition` = 'A')
) UNION (
SELECT ss.*, cs.*
FROM `student_summer` AS ss
INNER JOIN `choice_summer` as cs USING (`id_number`, `semester`)
WHERE (cs.`choice_id` = 3)
AND (ss.`condition` = 'A')
)
) as t
* used for shorten the select, but I'm actually only querying for about 7 columns out of the 50 availables.
This information is used in many flavors... "Do I have new students? Do I still have all students from a given date? Which students are subscribed after the given date? etc..." The result of this select statement is to be saved in another database.
Would it be possible for me to achieve this with a single view-like class? The information is read-only so I don't need to be able to modify/create/delte. Or do I have to declare a class for each table (ending up with 6 classes) and every time I need to query, I have to remember to filter?
Thanks for pointers.
EDIT: I don't have modification access to the database (I cannot create a view). Both databases may not be on the same server (so I cannot create a view on my second DB).
My concern is to avoid the full table scan before filtering on condition and choice_id.
EDIT 2: I've set up declarative classes like this:
class BaseStudent(object):
id_number = sqlalchemy.Column(sqlalchemy.String(7), primary_key=True)
semester = sqlalchemy.Column(sqlalchemy.String(5), primary_key=True)
unique_id_number = sqlalchemy.Column(sqlalchemy.String(7))
stateid = sqlalchemy.Column(sqlalchemy.String(12))
condition = sqlalchemy.Column(sqlalchemy.String(3))
class Student(BaseStudent, Base):
__tablename__ = 'student'
choices = orm.relationship('Choice', backref='student')
#class StudentYear(BaseStudent, Base):...
#class StudentSummer(BaseStudent, Base):...
class BaseChoice(object):
id_number = sqlalchemy.Column(sqlalchemy.String(7), primary_key=True)
semester = sqlalchemy.Column(sqlalchemy.String(5), primary_key=True)
choice_id = sqlalchemy.Column(sqlalchemy.String(1))
school = sqlalchemy.Column(sqlalchemy.String(2))
program = sqlalchemy.Column(sqlalchemy.String(5))
class Choice(BaseChoice, Base):
__tablename__ = 'choice'
__table_args__ = (
sqlalchemy.ForeignKeyConstraint(['id_number', 'semester',],
[Student.id_number, Student.semester,]),
)
#class ChoiceYear(BaseChoice, Base): ...
#class ChoiceSummer(BaseChoice, Base): ...
Now, the query that gives me correct SQL for one set of table is:
q = session.query(StudentYear, ChoiceYear) \
.select_from(StudentYear) \
.join(ChoiceYear) \
.filter(StudentYear.condition=='A') \
.filter(ChoiceYear.choice_id=='4')
but it throws an exception...
"Could not locate column in row for column '%s'" % key)
sqlalchemy.exc.NoSuchColumnError: "Could not locate column in row for column '*'"
How do I use that query to create myself a class I can use?

If you can create this view on the database, then you simply map the view as if it was a table. See Reflecting Views.
# DB VIEW
CREATE VIEW my_view AS -- #todo: your select statements here
# SA
my_view = Table('my_view', metadata, autoload=True)
# define view object
class ViewObject(object):
def __repr__(self):
return "ViewObject %s" % str((self.id_number, self.semester,))
# map the view to the object
view_mapper = mapper(ViewObject, my_view)
# query the view
q = session.query(ViewObject)
for _ in q:
print _
If you cannot create a VIEW on the database level, you could create a selectable and map the ViewObject to it. The code below should give you the idea:
student_tmp = Table('student_tmp', metadata, autoload=True)
choice_tmp = Table('choice_tmp', metadata, autoload=True)
# your SELECT part with the columns you need
qry = select([student_tmp.c.id_number, student_tmp.c.semester, student_tmp.stateid, choice_tmp.school])
# your INNER JOIN condition
qry = qry.where(student_tmp.c.id_number == choice_tmp.c.id_number).where(student_tmp.c.semester == choice_tmp.c.semester)
# other WHERE clauses
qry = qry.where(student_tmp.c.condition == 'A')
You can create 3 queries like this, then combine them with union_all and use the resulting query in the mapper:
view_mapper = mapper(ViewObject, my_combined_qry)
In both cases you have to ensure though that a PrimaryKey is properly defined on the view, and you might need to override the autoloaded view, and specify the primary key explicitely (see the link above). Otherwise you will either receive an error, or might not get proper results from the query.
Answer to EDIT-2:
qry = (session.query(StudentYear, ChoiceYear).
select_from(StudentYear).
join(ChoiceYear).
filter(StudentYear.condition == 'A').
filter(ChoiceYear.choice_id == '4')
)
The result will be tuple pairs: (Student, Choice).
But if you want to create a new mapped class for the query, then you can create a selectable as the sample above:
student_tmp = StudentTmp.__table__
choice_tmp = ChoiceTmp.__table__
.... (see sample code above)

This is to show what I ended up doing, any comment welcomed.
class JoinedYear(Base):
__table__ = sqlalchemy.select(
[
StudentYear.id_number,
StudentYear.semester,
StudentYear.stateid,
ChoiceYear.school,
ChoiceYear.program,
],
from_obj=StudentYear.__table__.join(ChoiceYear.__table__),
) \
.where(StudentYear.condition == 'A') \
.where(ChoiceYear.choice_id == '4') \
.alias('YearView')
and I will elaborate from there...
Thanks #van

SqlAlchemy: array of Postgresql custom types

So in my postgres DB I have the following custom type:
create type my_pg_type as (
sting_id varchar(32),
time_diff interval,
multiplier integer
);
To further complicate things, this is being used as an array:
alter table my_table add column my_keys my_pg_type [];
I'd like to map this with SQLAlchemy (0.6.4) !!
(apologies for elixir)
from sqlalchemy.dialects.postgresql import ARRAY
from sqlalchemy.types import Enum
from elixir import Entity, Field
class MyTable(Entity):
# -- snip --
my_keys = Field(ARRAY(Enum))
I know 'Enum' is incorrect in the above.
For an example of a value coming back from the database for that array column, I've shown below the value in ARRAY.result_processor(self, dialect, coltype):
class ARRAY(sqltypes.MutableType, sqltypes.Concatenable, sqltypes.TypeEngine):
# -- snip --
def result_processor(self, dialect, coltype):
item_proc = self.item_type.result_processor(dialect, coltype)
if item_proc:
def convert_item(item):
if isinstance(item, list):
return [convert_item(child) for child in item]
else:
return item_proc(item)
else:
def convert_item(item):
if isinstance(item, list):
return [convert_item(child) for child in item]
else:
return item
def process(value):
if value is None:
return value
"""
# sample value:
>>> value
'{"(key_1,07:23:00,0)","(key_2,01:00:00,20)"}'
"""
return [convert_item(item) for item in value]
return process
So the above process function incorrectly splits the string, assuming it's already a list.
So far, I've successfully subclassed ARRAY to properly split the string, and instead of Enum, I've tried to write my own type (implementing Unicode) to recreate the (string, timedelta, integer) tuple, but have run into a lot of difficulties, specifically the proper conversion of the interval to the Python timedelta.
I'm posting here in case I'm missing an obvious precedent way of doing this?

UPDATE See the recipe at the bottom for a workaround
I worked up some example code to see what psycopg2 is doing here, and this is well within their realm - psycopg2 is not interpreting the value as an array at all. psycopg2 needs to be able to parse out the ARRAY when it comes back as SQLA's ARRAY type assumes at least that much has been done. You can of course hack around SQLAlchemy's ARRAY, which here would mean basically not using it at all in favor of something that parses out this particular string value psycopg2 is giving us back.
But what's also happening here is that we aren't even getting at psycopg2's mechanics for converting timedeltas either, something SQLAlchemy normally doesn't have to worry about. In this case I feel like the facilities of the DBAPI are being under-utilized and psycopg2 is a very capable DBAPI.
So I'd advise you work with psycopg2's custom type mechanics over at http://initd.org/psycopg/docs/extensions.html#database-types-casting-functions.
If you want to mail their mailing list, here's a test case:
import psycopg2
conn = psycopg2.connect(host="localhost", database="test", user="scott", password="tiger")
cursor = conn.cursor()
cursor.execute("""
create type my_pg_type as (
string_id varchar(32),
time_diff interval,
multiplier integer
)
""")
cursor.execute("""
CREATE TABLE my_table (
data my_pg_type[]
)
""")
cursor.execute("insert into my_table (data) "
"values (CAST(%(data)s AS my_pg_type[]))",
{'data':[("xyz", "'1 day 01:00:00'", 5), ("pqr", "'1 day 01:00:00'", 5)]})
cursor.execute("SELECT * from my_table")
row = cursor.fetchone()
assert isinstance(row[0], (tuple, list)), repr(row[0])
PG's type registration supports global registration. You can also register the types on a per-connection basis within SQLAlchemy using the pool listener in 0.6 or connect event in 0.7 and further.
UPDATE - due to https://bitbucket.org/zzzeek/sqlalchemy/issue/3467/array-of-enums-does-not-allow-assigning I'm probably going to recommend people use this workaround type for now, until psycopg2 adds more built-in support for this:
class ArrayOfEnum(ARRAY):
def bind_expression(self, bindvalue):
return sa.cast(bindvalue, self)
def result_processor(self, dialect, coltype):
super_rp = super(ArrayOfEnum, self).result_processor(dialect, coltype)
def handle_raw_string(value):
inner = re.match(r"^{(.*)}$", value).group(1)
return inner.split(",")
def process(value):
return super_rp(handle_raw_string(value))
return process

Checkout the sqlalchemy_utils documentation:
CompositeType provides means to interact with
`PostgreSQL composite types`_. Currently this type features:
* Easy attribute access to composite type fields
* Supports SQLAlchemy TypeDecorator types
* Ability to include composite types as part of PostgreSQL arrays
* Type creation and dropping
Usage:
from collections import OrderedDict
import sqlalchemy as sa
from sqlalchemy_utils import Composite, CurrencyType
class Account(Base):
__tablename__ = 'account'
id = sa.Column(sa.Integer, primary_key=True)
balance = sa.Column(
CompositeType(
'money_type',
[
sa.Column('currency', CurrencyType),
sa.Column('amount', sa.Integer)
]
)
)
Array Of Composites:
from sqlalchemy_utils import CompositeArray
class Account(Base):
__tablename__ = 'account'
id = sa.Column(sa.Integer, primary_key=True)
balances = sa.Column(
CompositeArray(
CompositeType(
'money_type',
[
sa.Column('currency', CurrencyType),
sa.Column('amount', sa.Integer)
]
)
)
)

How to do an upsert with SqlAlchemy?

I have a record that I want to exist in the database if it is not there, and if it is there already (primary key exists) I want the fields to be updated to the current state. This is often called an upsert.
The following incomplete code snippet demonstrates what will work, but it seems excessively clunky (especially if there were a lot more columns). What is the better/best way?
Base = declarative_base()
class Template(Base):
__tablename__ = 'templates'
id = Column(Integer, primary_key = True)
name = Column(String(80), unique = True, index = True)
template = Column(String(80), unique = True)
description = Column(String(200))
def __init__(self, Name, Template, Desc):
self.name = Name
self.template = Template
self.description = Desc
def UpsertDefaultTemplate():
sess = Session()
desired_default = Template("default", "AABBCC", "This is the default template")
try:
q = sess.query(Template).filter_by(name = desiredDefault.name)
existing_default = q.one()
except sqlalchemy.orm.exc.NoResultFound:
#default does not exist yet, so add it...
sess.add(desired_default)
else:
#default already exists. Make sure the values are what we want...
assert isinstance(existing_default, Template)
existing_default.name = desired_default.name
existing_default.template = desired_default.template
existing_default.description = desired_default.description
sess.flush()
Is there a better or less verbose way of doing this? Something like this would be great:
sess.upsert_this(desired_default, unique_key = "name")
although the unique_key kwarg is obviously unnecessary (the ORM should be able to easily figure this out) I added it just because SQLAlchemy tends to only work with the primary key. eg: I've been looking at whether Session.merge would be applicable, but this works only on primary key, which in this case is an autoincrementing id which is not terribly useful for this purpose.
A sample use case for this is simply when starting up a server application that may have upgraded its default expected data. ie: no concurrency concerns for this upsert.

SQLAlchemy supports ON CONFLICT with two methods on_conflict_do_update() and on_conflict_do_nothing().
Copying from the documentation:
from sqlalchemy.dialects.postgresql import insert
stmt = insert(my_table).values(user_email='a#b.com', data='inserted data')
stmt = stmt.on_conflict_do_update(
index_elements=[my_table.c.user_email],
index_where=my_table.c.user_email.like('%#gmail.com'),
set_=dict(data=stmt.excluded.data)
)
conn.execute(stmt)

SQLAlchemy does have a "save-or-update" behavior, which in recent versions has been built into session.add, but previously was the separate session.saveorupdate call. This is not an "upsert" but it may be good enough for your needs.
It is good that you are asking about a class with multiple unique keys; I believe this is precisely the reason there is no single correct way to do this. The primary key is also a unique key. If there were no unique constraints, only the primary key, it would be a simple enough problem: if nothing with the given ID exists, or if ID is None, create a new record; else update all other fields in the existing record with that primary key.
However, when there are additional unique constraints, there are logical issues with that simple approach. If you want to "upsert" an object, and the primary key of your object matches an existing record, but another unique column matches a different record, then what do you do? Similarly, if the primary key matches no existing record, but another unique column does match an existing record, then what? There may be a correct answer for your particular situation, but in general I would argue there is no single correct answer.
That would be the reason there is no built in "upsert" operation. The application must define what this means in each particular case.

Nowadays, SQLAlchemy provides two helpful functions on_conflict_do_nothing and on_conflict_do_update. Those functions are useful but require you to swich from the ORM interface to the lower-level one - SQLAlchemy Core.
Although those two functions make upserting using SQLAlchemy's syntax not that difficult, these functions are far from providing a complete out-of-the-box solution to upserting.
My common use case is to upsert a big chunk of rows in a single SQL query/session execution. I usually encounter two problems with upserting:
For example, higher level ORM functionalities we've gotten used to are missing. You cannot use ORM objects but instead have to provide ForeignKeys at the time of insertion.
I'm using this following function I wrote to handle both of those issues:
def upsert(session, model, rows):
table = model.__table__
stmt = postgresql.insert(table)
primary_keys = [key.name for key in inspect(table).primary_key]
update_dict = {c.name: c for c in stmt.excluded if not c.primary_key}
if not update_dict:
raise ValueError("insert_or_update resulted in an empty update_dict")
stmt = stmt.on_conflict_do_update(index_elements=primary_keys,
set_=update_dict)
seen = set()
foreign_keys = {col.name: list(col.foreign_keys)[0].column for col in table.columns if col.foreign_keys}
unique_constraints = [c for c in table.constraints if isinstance(c, UniqueConstraint)]
def handle_foreignkeys_constraints(row):
for c_name, c_value in foreign_keys.items():
foreign_obj = row.pop(c_value.table.name, None)
row[c_name] = getattr(foreign_obj, c_value.name) if foreign_obj else None
for const in unique_constraints:
unique = tuple([const,] + [row[col.name] for col in const.columns])
if unique in seen:
return None
seen.add(unique)
return row
rows = list(filter(None, (handle_foreignkeys_constraints(row) for row in rows)))
session.execute(stmt, rows)

I use a "look before you leap" approach:
# first get the object from the database if it exists
# we're guaranteed to only get one or zero results
# because we're filtering by primary key
switch_command = session.query(Switch_Command).\
filter(Switch_Command.switch_id == switch.id).\
filter(Switch_Command.command_id == command.id).first()
# If we didn't get anything, make one
if not switch_command:
switch_command = Switch_Command(switch_id=switch.id, command_id=command.id)
# update the stuff we care about
switch_command.output = 'Hooray!'
switch_command.lastseen = datetime.datetime.utcnow()
session.add(switch_command)
# This will generate either an INSERT or UPDATE
# depending on whether we have a new object or not
session.commit()
The advantage is that this is db-neutral and I think it's clear to read. The disadvantage is that there's a potential race condition in a scenario like the following:
we query the db for a switch_command and don't find one
we create a switch_command
another process or thread creates a switch_command with the same primary key as ours
we try to commit our switch_command

There are multiple answers and here comes yet another answer (YAA). Other answers are not that readable due to the metaprogramming involved. Here is an example that
Uses SQLAlchemy ORM
Shows how to create a row if there are zero rows using on_conflict_do_nothing
Shows how to update the existing row (if any) without creating a new row using on_conflict_do_update
Uses the table primary key as the constraint
A longer example in the original question what this code is related to.
import sqlalchemy as sa
import sqlalchemy.orm as orm
from sqlalchemy import text
from sqlalchemy.dialects.postgresql import insert
from sqlalchemy.orm import Session
class PairState(Base):
__tablename__ = "pair_state"
# This table has 1-to-1 relationship with Pair
pair_id = sa.Column(sa.ForeignKey("pair.id"), nullable=False, primary_key=True, unique=True)
pair = orm.relationship(Pair,
backref=orm.backref("pair_state",
lazy="dynamic",
cascade="all, delete-orphan",
single_parent=True, ), )
# First raw event in data stream
first_event_at = sa.Column(sa.TIMESTAMP(timezone=True), nullable=False, server_default=text("TO_TIMESTAMP(0)"))
# Last raw event in data stream
last_event_at = sa.Column(sa.TIMESTAMP(timezone=True), nullable=False, server_default=text("TO_TIMESTAMP(0)"))
# The last hypertable entry added
last_interval_at = sa.Column(sa.TIMESTAMP(timezone=True), nullable=False, server_default=text("TO_TIMESTAMP(0)"))
#staticmethod
def create_first_event_if_not_exist(dbsession: Session, pair_id: int, ts: datetime.datetime):
"""Sets the first event value if not exist yet."""
dbsession.execute(
insert(PairState).
values(pair_id=pair_id, first_event_at=ts).
on_conflict_do_nothing()
)
#staticmethod
def update_last_event(dbsession: Session, pair_id: int, ts: datetime.datetime):
"""Replaces the the column last_event_at for a named pair."""
# Based on the original example of https://stackoverflow.com/a/49917004/315168
dbsession.execute(
insert(PairState).
values(pair_id=pair_id, last_event_at=ts).
on_conflict_do_update(constraint=PairState.__table__.primary_key, set_={"last_event_at": ts})
)
#staticmethod
def update_last_interval(dbsession: Session, pair_id: int, ts: datetime.datetime):
"""Replaces the the column last_interval_at for a named pair."""
dbsession.execute(
insert(PairState).
values(pair_id=pair_id, last_interval_at=ts).
on_conflict_do_update(constraint=PairState.__table__.primary_key, set_={"last_interval_at": ts})
)

The below works fine for me with redshift database and will also work for combined primary key constraint.
SOURCE : this
Just few modifications required for creating SQLAlchemy engine in the function
def start_engine()
from sqlalchemy import Column, Integer, Date ,Metadata
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy.dialects.postgresql import insert
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
from sqlalchemy.dialects import postgresql
Base = declarative_base()
def start_engine():
engine = create_engine(os.getenv('SQLALCHEMY_URI',
'postgresql://localhost:5432/upsert'))
connect = engine.connect()
meta = MetaData(bind=engine)
meta.reflect(bind=engine)
return engine
class DigitalSpend(Base):
__tablename__ = 'digital_spend'
report_date = Column(Date, nullable=False)
day = Column(Date, nullable=False, primary_key=True)
impressions = Column(Integer)
conversions = Column(Integer)
def __repr__(self):
return str([getattr(self, c.name, None) for c in self.__table__.c])
def compile_query(query):
compiler = query.compile if not hasattr(query, 'statement') else
query.statement.compile
return compiler(dialect=postgresql.dialect())
def upsert(session, model, rows, as_of_date_col='report_date', no_update_cols=[]):
table = model.__table__
stmt = insert(table).values(rows)
update_cols = [c.name for c in table.c
if c not in list(table.primary_key.columns)
and c.name not in no_update_cols]
on_conflict_stmt = stmt.on_conflict_do_update(
index_elements=table.primary_key.columns,
set_={k: getattr(stmt.excluded, k) for k in update_cols},
index_where=(getattr(model, as_of_date_col) < getattr(stmt.excluded, as_of_date_col))
)
print(compile_query(on_conflict_stmt))
session.execute(on_conflict_stmt)
session = start_engine()
upsert(session, DigitalSpend, initial_rows, no_update_cols=['conversions'])

This allows access to the underlying models based on string names
def get_class_by_tablename(tablename):
"""Return class reference mapped to table.
https://stackoverflow.com/questions/11668355/sqlalchemy-get-model-from-table-name-this-may-imply-appending-some-function-to
:param tablename: String with name of table.
:return: Class reference or None.
"""
for c in Base._decl_class_registry.values():
if hasattr(c, '__tablename__') and c.__tablename__ == tablename:
return c
sqla_tbl = get_class_by_tablename(table_name)
def handle_upsert(record_dict, table):
"""
handles updates when there are primary key conflicts
"""
try:
self.active_session().add(table(**record_dict))
except:
# Here we'll assume the error is caused by an integrity error
# We do this because the error classes are passed from the
# underlying package (pyodbc / sqllite) SQLAlchemy doesn't mask
# them with it's own code - this should be updated to have
# explicit error handling for each new db engine
# <update>add explicit error handling for each db engine</update>
active_session.rollback()
# Query for conflic class, use update method to change values based on dict
c_tbl_primary_keys = [i.name for i in table.__table__.primary_key] # List of primary key col names
c_tbl_cols = dict(sqla_tbl.__table__.columns) # String:Col Object crosswalk
c_query_dict = {k:record_dict[k] for k in c_tbl_primary_keys if k in record_dict} # sub-dict from data of primary key:values
c_oo_query_dict = {c_tbl_cols[k]:v for (k,v) in c_query_dict.items()} # col-object:query value for primary key cols
c_target_record = session.query(sqla_tbl).filter(*[k==v for (k,v) in oo_query_dict.items()]).first()
# apply new data values to the existing record
for k, v in record_dict.items()
setattr(c_target_record, k, v)

This works for me with sqlite3 and postgres. Albeit it might fail with combined primary key constraints and will most likely fail with additional unique constraints.
try:
t = self._meta.tables[data['table']]
except KeyError:
self._log.error('table "%s" unknown', data['table'])
return
try:
q = insert(t, values=data['values'])
self._log.debug(q)
self._db.execute(q)
except IntegrityError:
self._log.warning('integrity error')
where_clause = [c.__eq__(data['values'][c.name]) for c in t.c if c.primary_key]
update_dict = {c.name: data['values'][c.name] for c in t.c if not c.primary_key}
q = update(t, values=update_dict).where(*where_clause)
self._log.debug(q)
self._db.execute(q)
except Exception as e:
self._log.error('%s: %s', t.name, e)

As we had problems with generated default-ids and references which lead to ForeignKeyViolation-Errors like
update or delete on table "..." violates foreign key constraint
Key (id)=(...) is still referenced from table "...".
we had to exclude the id for the update dict, as otherwise the it will be always generated as new default value.
In addition the method is returning the created/updated entity.
from sqlalchemy.dialects.postgresql import insert # Important to use the postgresql insert
def upsert(session, data, key_columns, model):
stmt = insert(model).values(data)
# Important to exclude the ID for update!
exclude_for_update = [model.id.name, *key_columns]
update_dict = {c.name: c for c in stmt.excluded if c.name not in exclude_for_update}
stmt = stmt.on_conflict_do_update(
index_elements=key_columns,
set_=update_dict
).returning(model)
orm_stmt = (
select(model)
.from_statement(stmt)
.execution_options(populate_existing=True)
)
return session.execute(orm_stmt).scalar()
Example:
class UpsertUser(Base):
__tablename__ = 'upsert_user'
id = Column(Id, primary_key=True, default=uuid.uuid4)
name: str = Column(sa.String, nullable=False)
user_sid: str = Column(sa.String, nullable=False, unique=True)
house_admin = relationship('UpsertHouse', back_populates='admin', uselist=False)
class UpsertHouse(Base):
__tablename__ = 'upsert_house'
id = Column(Id, primary_key=True, default=uuid.uuid4)
admin_id: Id = Column(Id, ForeignKey('upsert_user.id'), nullable=False)
admin: UpsertUser = relationship('UpsertUser', back_populates='house_admin', uselist=False)
# Usage
upserted_user = upsert(session, updated_user, [UpsertUser.user_sid.name], UpsertUser)
Note: Only tested on postgresql but could work also for other DBs which support ON DUPLICATE KEY UPDATE e.g. MySQL

In case of sqlite, the sqlite_on_conflict='REPLACE' option can be used when defining a UniqueConstraint, and sqlite_on_conflict_unique for unique constraint on a single column. Then session.add will work in a way just like upsert. See the official documentation.

I use this code for upsert
Before using this code, you should add primary keys to table in database.
from sqlalchemy import create_engine
from sqlalchemy import MetaData, Table
from sqlalchemy.inspection import inspect
from sqlalchemy.engine.reflection import Inspector
from sqlalchemy.dialects.postgresql import insert
def upsert(df, engine, table_name, schema=None, chunk_size = 1000):
metadata = MetaData(schema=schema)
metadata.bind = engine
table = Table(table_name, metadata, schema=schema, autoload=True)
# olny use common columns between df and table.
table_columns = {column.name for column in table.columns}
df_columns = set(df.columns)
intersection_columns = table_columns.intersection(df_columns)
df1 = df[intersection_columns]
records = df1.to_dict('records')
# get list of fields making up primary key
primary_keys = [key.name for key in inspect(table).primary_key]
with engine.connect() as conn:
chunks = [records[i:i + chunk_size] for i in range(0, len(records), chunk_size)]
for chunk in chunks:
stmt = insert(table).values(chunk)
update_dict = {c.name: c for c in stmt.excluded if not c.primary_key}
s = stmt.on_conflict_do_update(
index_elements= primary_keys,
set_=update_dict)
conn.execute(s)