You got like this

select is the function that creates SQL query. It has nothing to do with Json serialization.

with db_session: ps = select(p for p in PaymentSource)[:] print(ps)

i know

So you fetch your results. Then use to_dict() and json.dumps(...) with providing JSONEncoder which you can get from that stackoverflow link.

On Object?

Can i put it on list objects?

or i need get object then convert to_dict and then put new list in json_dump with function?

Yes it worked but when i need json dump list i need convert to_dict all object and put in together in new list then use json dump?

And how do it with child element in my query list?

with db_session: ps = select(p for p in PaymentSource)[:] result = [p.to_dict() for p in ps]

Try replace last line with result = json.dumps([p.to_dict() for p in ps], cls=UUIDEncoder)

something like that: def custom_encoder(value): if isinstance(value, UUID): return value.hex raise TypeError(value) with db_session: ps = select(p for p in PaymentSource)[:] raw_data = [p.to_dict() for p in ps] s = json.dumps(raw_data, default=custom_encoder)

Yep for this it worked. But if i add for example table via foreign key? It should work?

if UUID serialization was the only issue - it should work.

okay thanks

In practice, for serializing complex JSON data you may need to write something like this with db_session: # retrieve objects students = select(s for s in Student)[:] # prepare raw dicts: data = [ { 'id': s.id, 'uuid': s.uuid.hex, 'name': s.name, 'birth_data': s.birth_date.isoformat(), 'group': s.group.id, 'grades': [ { 'id:' grade.id, 'subject': grade.subject.name, 'value': grade.value, } for grade in s.grades ] } for s in students ] # make JSON from raw dicts json_data = json.dumps(data) I think it is not pretty, and after we finish work on migrations I want to do something with JSON

yes it really good thing

In practice, for serializing complex JSON data you may need to write something like this with db_session: # retrieve objects students = select(s for s in Student)[:] # prepare raw dicts: data = [ { 'id': s.id, 'uuid': s.uuid.hex, 'name': s.name, 'birth_data': s.birth_date.isoformat(), 'group': s.group.id, 'grades': [ { 'id:' grade.id, 'subject': grade.subject.name, 'value': grade.value, } for grade in s.grades ] } for s in students ] # make JSON from raw dicts json_data = json.dumps(data) I think it is not pretty, and after we finish work on migrations I want to do something with JSON

Also may add new integration layer with marshmallow? Like this https://github.com/marshmallow-code/marshmallow-sqlalchemy

Also may add new integration layer with marshmallow? Like this https://github.com/marshmallow-code/marshmallow-sqlalchemy

Permanent link to the marshmallow-code/marshmallow-sqlalchemy project you mentioned. (?)

I try use marshmallow with Pony Orm it work out of box

I try use marshmallow with Pony Orm it work out of box

is there anything you cannot do with Marshmallow, but would like to?

for sqlalchemy exist simple thin layer

for example

class Author(db.Model): id = db.Column(db.Integer, primary_key=True) name = db.Column(db.String(255))

then with layer

class AuthorSchema(ma.ModelSchema): class Meta: model = Author

layer just get fields from model

I mean do you need a more convenient or powerful tool for to/from JSON @norguhtar ?

yep

like java :D

not sure I understand

In java serializator really smart and powerfull. It just eating any object and return correct json

Ah, I see But it covers Java to Java serialization/deserislization And in our case we usually serialize at the backend and deserialize at the frontend

Python to JavaScript and back

Yep i need current this case

And now i think use marshmallow right way

Do you encounter any situation when marshmallow is not enough or it is not that convenient?

i think marshmallow enough for this case

I see

Thanks

Three cheers for activity in the github repo! Keep it going! 🥳

🔥

There is actually a lot of activity behind the scenes

That's great to hear! Really excited for what you have in the pipeline.

Hi Alexey , on Januar 18 there was a discussion about the online editor's Cascade delete setting still being broken, False becomes True in the python code export, or rather the value isn't stored in the online editor. Back then, you said it's fixed, but it seems to be broken again

Hi Alexey , on Januar 18 there was a discussion about the online editor's Cascade delete setting still being broken, False becomes True in the python code export, or rather the value isn't stored in the online editor. Back then, you said it's fixed, but it seems to be broken again

Thanks for reporting, will check and fix that

☺️

Alexey and while we're at it, back in 2016 there was a brief discussion on the topic of type annotation here in the Telegram group. I've been stumbling about that too. In my current project I'm (for the first time) type-annotating each and every function in order to avoid type errors from the outset. PyCharm has a great integration of type checking and reports missing annotations and wrong annotations. This is what I get for PonyORM entities (BMELKategorie in this case). I think it would be a nice improvement of PonyORM if it would deliver the entities with an inbuilt iterator class, e.g. `Iterable`

Alexey and while we're at it, back in 2016 there was a brief discussion on the topic of type annotation here in the Telegram group. I've been stumbling about that too. In my current project I'm (for the first time) type-annotating each and every function in order to avoid type errors from the outset. PyCharm has a great integration of type checking and reports missing annotations and wrong annotations. This is what I get for PonyORM entities (BMELKategorie in this case). I think it would be a nice improvement of PonyORM if it would deliver the entities with an inbuilt iterator class, e.g. `Iterable`

Right, thanks for the suggestion, Fritz! @metaprogrammer @aerok

Alexey and while we're at it, back in 2016 there was a brief discussion on the topic of type annotation here in the Telegram group. I've been stumbling about that too. In my current project I'm (for the first time) type-annotating each and every function in order to avoid type errors from the outset. PyCharm has a great integration of type checking and reports missing annotations and wrong annotations. This is what I get for PonyORM entities (BMELKategorie in this case). I think it would be a nice improvement of PonyORM if it would deliver the entities with an inbuilt iterator class, e.g. `Iterable`

Adding type annotations is a good idea, but I'm not sure we can did it since we still support Python2 as well which doesn't udenrstand type annotation syntax

Hi! Where can i read about pony user and role groups and access rule concept?

Hi! Role groups and access rules were an experiment which never been released officially. This concept turned out to be too complex to use

@metaprogrammer ok, thank you

Is it possible to return value from async def function/coroutine decorated by orm.db_session? I try but get None. This is my test code models.py from pony import orm db = orm.Database() class Data(db.Entity): id = orm.PrimaryKey(int, auto=True, sql_type='serial') def init(create_tables=False): db.bind(...) db.generate_mapping(create_tables=create_tables) app.py import logging from fastapi import FastAPI from starlette.responses import Response from pony import orm import models app = FastAPI() log = logging.getLogger() log.setLevel('DEBUG') models.init(create_tables=True) models.orm.set_sql_debug(debug=True, show_values=True) @orm.db_session async def get_data(): models.Data() models.db.flush() data = models.Data.select()[:] log.info('get_data: %s', data) return data @app.get('/test') async def test(response: Response): data = await get_data() log.info('test: %s', data) return data console output INFO: Uvicorn running on http://0.0.0.0:80 (Press CTRL+C to quit) INFO: Started reloader process [1] could not connect to server: Connection refused Is the server running on host "db" (192.168.224.2) and accepting TCP/IP connections on port 5432? INFO: Started server process [7] INFO: Waiting for application startup. INFO: ('10.0.2.2', 58674) - "GET /docs HTTP/1.1" 200 INFO: ('10.0.2.2', 58674) - "GET /openapi.json HTTP/1.1" 200 INFO: GET NEW CONNECTION INFO: INSERT INTO "data" DEFAULT VALUES RETURNING "id" INFO: SELECT "d"."id" FROM "data" "d" INFO: get_data: [Data[1]] INFO: COMMIT INFO: RELEASE CONNECTION INFO: test: None INFO: ('10.0.2.2', 58674) - "GET /test HTTP/1.1" 200

contextmanager "with orm.db_session" works but documentation says do not use it with async coro

hi, is there limitation of using nested db_session in entity instantiation.. i have about 3 nested object instantiation but always getting this error pony.orm.core.ERDiagramError: Mapping is not generated for entity 'OAuth2AuthorizationCode' any clue? thank in advance

btw I tried to instantiate that object in python shell it run okay..

Hello. I believe I found a bug: >>> from pony.orm import * >>> db = Database() >>> class Test(db.Entity): ... pass ... >>> db.bind(provider='sqlite', filename=':memory:') >>> db.generate_mapping(create_tables=True) >>> test_entity = Test() >>> Test.select()[:] [Test[1]] >>> Test.select().delete(bulk=True) 1 >>> Test.select()[:] [Test[1]]

The second select()[:] should return empty list, since the only entity has been deleted, but instead it returns a result from the cache, which appears to not have been updated.

Thanks

Should I create an issue on GitHub or is it enough here for such a minor thing?

I think it is not necessary to create an issue, we should fix it today

Great, thanks!

Just out of curiosity: do you have any corporate sponsors of Pony right now? Because I don't see any on the front page.

Well, I don't even see any backers on GitHub repo, which is sad. Did you forget to put them or will I be the first?

that'd be so nice of you

Well, I don't even see any backers on GitHub repo, which is sad. Did you forget to put them or will I be the first?

Hi Krysztof, Here is the page with the Pony ORM respectful backers: https://github.com/ponyorm/pony/blob/orm/BACKERS.md

Hi Krysztof, Here is the page with the Pony ORM respectful backers: https://github.com/ponyorm/pony/blob/orm/BACKERS.md

Permanent link to the file BACKERS.md mentioned. (?)

Also, after the migrations release we are going to add Pony ORM Priority Support which provides 1 and 2 business days email support in case of any unforeseen issues and Custom Development for Python/PonyORM/AWS/TypeScript/React stack

> after the migrations release I hope at least my son will meet that release :D

+

You also will

Is there a way to avoid the expensive iterrows() in writing to DB record by record from a pandas DataFrame like that: for index, row in df.iterrows(): i = EntityX(attribute1=row.attr1col, attribute2=row.attr2col)

probably you can use itertuples: for val1, val2, val2 in df.itertuples(): obj = EntityX(attr1=val1, attr2=val2) You can also avoid entity instance creation altohether by inserting row directly into database, it should be much faster: table_name = EntityX._table_ for val1, val2, val2 in df.itertuples(): db.insert(table_name, column1=val1, column2=val2)

Thank you for your hints @metaprogrammer ! Inserting rows directly into the database works fine, yet got it only working with iterrows: for index, row in df.iterrows(): db.insert(EntityX, column1=row.val1, column2=row.val2) On the itertuples idea, I thought this should do it, but it doesn't: for row in df.itertuples(index=False, name='Betriebszweig'): i = row The form of row is exactly what one would want, i.e. EntityX(col1=val1, col2=val2) Yet to combine itertuples and db.insert, the form of row should be (EntityX/table_name, col1=val1, col2=val2), or could they also be handed over as positional arguments?

db.insert doesn't know about table structure, so you need to pass column names to is as a keyword arguments. Or you can create a dict using zip function: columns = ["col1", "col2", "col3"] for row in df.itertuples(): db.insert(table_name, dict(zip(columns, row)))

When i try select data

pony.orm.sqltranslation.IncomparableTypesError: Incomparable types 'datetime' and 'NoneType' in expression: p.tsFrom <= tsTo

i can't use None in compare?