Maybe @amPerl could help with it

i'd have to write up an example

remind me to never google "pony model" again

;:D

Funny that @akozlovsky has no admin rights in this chat:)

Hello! is it possible to get objects in many-to-many relations in creation order?

class Structures(db.Entity): id = PrimaryKey(int, auto=True) data = Required(LongStr) reagents = Set('Reactions', reverse='reagents') products = Set('Reactions', reverse='products') class Reactions(db.Entity): id = PrimaryKey(int, auto=True) reagents = Set(Structures, reverse='reagents') products = Set(Structures, reverse='products')

r=Reactions() r.reagents.add(s1) r.reagents.add(s2) r.reagents.add(s3) list(r.reagents)

Currently Set collections are unordered. You can get ordered list of items if you performs a query: r.reagents.select().order_by(lambda r: r.id)[:]

By the way it is better to name all entities in a singlular form: Structure, Reagent, etc.

Also, it looks strange that both end of relation have the same name: Structure.reagents <-> Reaction.reagents Structure.products <-> Reaction.products Typically the name is different, like: Student.courses <-> Course.students

Also, it looks strange that both end of relation have the same name: Structure.reagents <-> Reaction.reagents Structure.products <-> Reaction.products Typically the name is different, like: Student.courses <-> Course.students

Oh great, one of them new urls

Thanks. But this recipe order objects by structures. This is chemical db. Any reaction have reagents and products. And any structures may be in reagent role or product

I will rewrite many to many to combination of 2 many to one

Ok, now I understand. I think you should not replace many-to-many with several many-to-one, it will be not correct. I think your current scheme is good

http://joxi.ru/KAxjMx7uMZOD0m?d=1 Thanks for magnificent ORM!

class Sticker(db.Entity): file_id = Required(str) # BQADBAAD6QIAAqQW9AX8GaFqUwWnrgI emoji = Required(unicode) # ❓ sticker_pack = Optional('StickerPack', column='sticker_pack_url') tags = Set('Tag', cascade_delete=True) sticker_messages = Set('StickerMessage') tl_id = Optional(int, size=8) # The id field of the telegram api schema the apps use. tl_access_hash = Optional(int, size=8) # The access_hash field of the telegram api schema the apps use. PrimaryKey(file_id, emoji) - class StickerPack(db.Entity): url = PrimaryKey(str) title = Required(str) owner = Required(User, column='owner_id') stickers = Set(Sticker) first_seen = Required(datetime, sql_default='NOW()')

For my own reference: sticker:BQADAgADRgADyIsGAAF-kAhqVu7-1QI

I think this is a bug, and I'll fix it. The reason for the error is: the attribute is specified as required, but in the database its value for some object is empty string. So when Pony attempts to load that object from the database it encounters validation error. I'll change that behavior so Pony will ignore the error for empty string values loaded from the database.

I think this is a bug, and I'll fix it. The reason for the error is: the attribute is specified as required, but in the database its value for some object is empty string. So when Pony attempts to load that object from the database it encounters validation error. I'll change that behavior so Pony will ignore the error for empty string values loaded from the database.

Easy:)

I think this is a bug, and I'll fix it. The reason for the error is: the attribute is specified as required, but in the database its value for some object is empty string. So when Pony attempts to load that object from the database it encounters validation error. I'll change that behavior so Pony will ignore the error for empty string values loaded from the database.

Yes, the string field is indeed empty.

Yeah, the value was set with a java client, where there is no such nice framework. (I was really sad)

Done: https://github.com/ponyorm/pony/commit/60ddecb060b8ea3f8637d87ec1bcbac51d8c4ddf

Have a good time of day, guys, I'm going to sleep

Thank you for the fix. Ping me when there is a new release, please

Until then pip install -e git://github.com/ponyorm/pony.git@dfedd28d02330fb351dc019f4dd03134299c23ab#egg=pony will work

---

TIL, https://telegram.me/addstickers/ is a stickerpack

https://telegram.me/StickerTagBot?start=sticker:BQADAgADRgADyIsGAAF-kAhqVu7-1QI

See the sticker pack URL

This just blow my database.

Welcome!

Thanks!

(How) can I order elements in a Set? pack = StickerPack.get(url=shortname) sticker_list = pack.stickers Now pack.stickers.order_by(Sticker.file_id)? Probably like this? stickers_list = orm.select(s for s in Sticker if s.sticker_pack and s.sticker_pack.url == shortname).order_by(Sticker.file_id)

Is that the way to do it?

Did it work for you?

You can have even more complex ordering condition using lambdas https://docs.ponyorm.com/api_reference.html#Set.order_by

Yes, It works.

Welcome!

@Mikkass Thanks.

Can I apply queries when getting elements from a Set attribute?

Summary: isinstance(pack, StickerPack) and orm.select(s for s in pack.stickers) results in TypeError: Cannot iterate over non-entity object

pack.stickers is of type orm.Set

I think I can do that on python side, with: data["stickers"] = [_sticker_to_json(s, add_pack_info=False) for s in [s_ for s_ in pack.stickers if not "nsfw" in [t.string.lower() for t in s_.tags]][:include_stickers]]

But that kinda sucks.

What is the best way to do it? Doing it in-memory on python side is not what I really want to do...

In any case, here my relevant data stucture, but it didn't change much. class Sticker(db.Entity): file_id = Required(str) # BQADBAAD6QIAAqQW9AX8GaFqUwWnrgI emoji = Required(unicode) # ❓ sticker_pack = Optional('StickerPack', column='sticker_pack_url') tags = Set('Tag', cascade_delete=True) sticker_messages = Set('StickerMessage') tl_id = Optional(int, size=8) # The id field of the telegram api schema the apps use. tl_access_hash = Optional(int, size=8) # The access_hash field of the telegram api schema the apps use. PrimaryKey(file_id, emoji) class Tag(db.Entity): id = PrimaryKey(int, auto=True) user = Required(User, column='user_id') message_id = Optional(int) # Of the message with the text, the tag. None means it got added via web gui. sticker = Required(Sticker) string = Required(unicode) use_global = Required(bool, default=False) composite_key(user, sticker, string) class StickerPack(db.Entity): url = PrimaryKey(str) title = Required(str) owner = Required(User, column='owner_id') stickers = Set(Sticker) first_seen = Required(datetime, sql_default='NOW()')

At this moment there is a technical limitation, that only entity class can be used as a source of generator. We can overcome it, but other tasks were more important. If you want to select from a collection you need to use lambda syntax: result = pack.stickers.select( lambda s: ...)

I'll try that, and hope this will be added eventually. But indeed this is not the most urgent feature to have :D

Not quite sure how to write that query tho

lambda should return true for objects you want to select. like: student1.courses.select(lambda c: c.credits > 20 and "philosophy" in c.name.lower())

Oh. I see.

I removed the if ... else ..., so It will return a boolean.

But it seems lambda s: "nsfw" not in [t.string.lower() for t in s.tags] is not possible? Still failing with TypeError: LIST_APPEND() takes 1 positional argument but 2 were given

Try to use generator instead of list comprehension: (...) instead of [...]

Wow, it works

This concept of using (...) instead of [...] always confuses me

Apparently, we need to fix bytecode decompiling for [...] in last Python. They both should work. The fix should be pretty easy

I am using ... hold on

python3.5

To be exact: Docker, tiangolo/uwsgi-nginx-flask:flask-python3.5

Hi, I'm using pony and I want to return list of results from a function with the @db_session decorator. the results are the relations from one of my classes. however when I try to access the properties of those results outside the function I get pony.orm.core.DatabaseSessionIsOver: Cannot load attribute Ramal[2].ramal: the database session is over . How can I query all properties from those objects and return them to use them later?

Hi Federico Why don't you process the objects inside the db_session?

I was trying to modularize and save the results obtained in the search to be used in other functions.

what if you could convert the result to JSON?

How would I do that?

this is something we are working on now

glad to hear that, would it be a good idea to also add a way to copy the result to a python dictionary to be used outside the db_session?

thanks for the help, I'm really liking the orm!

glad to hear that, would it be a good idea to also add a way to copy the result to a python dictionary to be used outside the db_session?

Currently I do that myself

Currently you can use to_dict method of entity instance. You can get a query result and convert it to list of dicts using a list comprehension.

https://docs.ponyorm.com/api_reference.html?highlight=to_dict#Entity.to_dict

Currently you can use to_dict method of entity instance. You can get a query result and convert it to list of dicts using a list comprehension.

huh.

The limitation of to_dict is may be hard to convert deep nested structures

The limitation of to_dict is may be hard to convert deep nested structures

How does it handle timestamps?

I understiood not at all?

It should return attribute values as a dict values. It does not convert them to JSON-compatible values automatically. But when you call json.dumps you can pass default argument which is a function. You can pass a function which convert a timestamp to suitable representation

Right. Awesome once again

BTW, how is that migration mode going on?

awesome, thanks!

It is in process, we have some tests which a fail currently and want to fix them before release