I think I'm just going to change approach haha

The flask support for current_object with pony is a very nice improvement, thanks!

The flask support for current_object with pony is a very nice improvement, thanks!

Can you elaborate on this?

I meant current_user which is from flask-login

Flask-login plugin uses current_user global object https://flask-login.readthedocs.io/en/latest/#flask_login.current_user You can set it to Pony entity instance. The resulted object is not really Pony object, but Flask-login proxy. So, previously when you write something like select(m for m in Messages if m.author == current_user) it didn't work. because Pony didn't understand that current_user is equivalent to a entity object. The same problem was with statements like message.author = current_user this line resulted in TypeError Now Pony automatically dereferences current_user proxy to its underlying entity object in this and similar situations

I want that "Pony will create the database file if it doesn’t exist. If it already exists, Pony will use it." For sqlite there is create_db=True but I use postgresql. Is there anything similar for postgresql?

Another question; In psycopg2.connect I can use a url such as "postgres://sandikadmin:sandikadminpw@localhost:5432/sandikdb". Example: url = "postgres://sandikadmin:sandikadminpw@localhost:5432/sandikdb" dbapi2.connect(url) I tried this but not work. Can I use anything similar in Pony Orm.

For PostgreSQL you need to create database manually. For this you need admin rights, and for secrity reason application is typically running under restricted database user which doesn't have admin rights Database urls are not supported yet, you need to specify params separately: params = dict( provider='postgres', user='sandikadmin', password='sandikadminpw', host='localhost', port=5432, database='sandikdb' ) db.bind(**params)

update with ponyorm?

https://docs.ponyorm.org/working_with_entity_instances.html#updating-an-object

For PostgreSQL you need to create database manually. For this you need admin rights, and for secrity reason application is typically running under restricted database user which doesn't have admin rights Database urls are not supported yet, you need to specify params separately: params = dict( provider='postgres', user='sandikadmin', password='sandikadminpw', host='localhost', port=5432, database='sandikdb' ) db.bind(**params)

Thanks

"Starting with the version 3.9 SQLite provides the JSON1 extension module. This extension improves performance when working with JSON queries, although Pony can work with JSON in SQLite even without this module."

Does SQLite support all json query types with pony?

Yes, if JSON1 extension is not provided Pony uses fallback implementation written in Python

Great, thank you

sqlite is only for testing / local dev for me so that seems fine

Pony supports the same JSON features in PostgreSQL in SQLite, but in PostgreSQL they are implemented using native JSONB PostgreSQL type, while in SQLite Pony stores JSON values as strings and parse it using some Python functions, so the performance is not as great as in PostgreSQL

Understood :)

Does any one know of a tool that will generate pony code from a postgresql database ?

For simple schema you can try this:https://github.com/abetkin/pony-inspect Resulted entities may require manual editing. For example, this tool can't not recognize many-to-many relationship and produces additional intermediary entity with two one-to-many relationshups, you can manually replace it to more convenient many-to-many relationship

For simple schema you can try this:https://github.com/abetkin/pony-inspect Resulted entities may require manual editing. For example, this tool can't not recognize many-to-many relationship and produces additional intermediary entity with two one-to-many relationshups, you can manually replace it to more convenient many-to-many relationship

Permanent link to the abetkin/pony-inspect project you mentioned. (?)

Cool thanks I’ll check it out.

facebook_report = models.FacebookReport.get( lambda r: r.uuid == facebook_report_uuid and (r.facebook_account.user == current_user or r.facebook_ad_account.facebook_account.user == current_user or r.facebook_campaign.facebook_ad_account.facebook_account.user == current_user or r.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user == current_user or r.facebook_ad.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user == current_user) ) or abort(404)

Should this style of query work?

It runs, but doesn't find any results

Rather than chaining the ors like that, maybe you can stick them all in a any()?

Not sure if pony supports

but would be cleaner IMO if it does

also hi Matt

good idea

hi Jake

I hope you're well

I think it should work You can also try to rewrite it as facebook_report = models.FacebookReport.get( lambda r: r.uuid == facebook_report_uuid and current_user in (r.facebook_account.user, r.facebook_ad_account.facebook_account.user, r.facebook_campaign.facebook_ad_account.facebook_account.user, r.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user, r.facebook_ad.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user)) ) or abort(404) I cannot test it right now but I think that it should work too. But the result will be the same

thanks, good idea

I think I understand why this query can produce empty result

The query you gave seemed to produce the same SQL

I think I understand why this query can produce empty result

why do you think?

If you traverse attributes like r.foo.baz.bar this requires joining of all used tables. I think Pony will use inner join here. That means that if r does not have corresponding r.boo.bar.baz.user object, this specific`r` instance will be excluded from the result. And the same applies for all other r.xxx.yyy.user expressions as well. As any specific r instance will probably linked to only one user, all r will be excluded from the result, because at least one of r.foo.bar.baz.user path will be missing for each r Probably we can user left join for such queries, but I'm not sure it will be enough for all queries to give correct results. I think correct was to write such queries is to use multiple exists(...) subqueries combined with or

so exists() or exists() or exists() etc?

You can check do your original query uses INNER JOIN or LEFT JOIN

there is no mention of join, does that make it inner join?

I don't remember, maybe Pony already took this situaton into account and uses LEFT JOIN automatically

It is better to check resulted SQL to be sure

so exists() or exists() or exists() etc?

Yes, this is what I mean. But it is hard to suggest the query itself, maybe it is easier to write it using reverse attributes like user.something

maybe :

models.exists(r2 for r2 in models.FacebookReport if r2 == r and r2 .facebook_account.user == current_user)

?

Yes, I think it should work

You can also try to rewrite original query using left join: facebook_report = left_join( r for in models.FacebookReport if r.uuid == facebook_report_uuid and (r.facebook_account.user == current_user or r.facebook_ad_account.facebook_account.user == current_user or r.facebook_campaign.facebook_ad_account.facebook_account.user == current_user or r.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user == current_user or r.facebook_ad.facebook_ad_set.facebook_campaign.facebook_ad_account.facebook_account.user == current_user) ) or abort(404)

left join worked :D

thank you

Hi all! Does it make sense to ask now about orm-migration issues? :0)

Hi, I think yes

why view exist a register?

how to see if there is a record

?

https://docs.ponyorm.org/api_reference.html?highlight=exists#Entity.exists

Hi! Let's say I create some bindings like this: class Exam(entity): student=orm.Optional('Student', reverse='exam') grade = orm.Optional(int, default=0) class Student(entity): exam = orm.Optional('Exam', reverse='student') And then I want to sort byt Students based on grade. So I do sort them by something like Student.select().sort_by("lambda g: g.exam.grade") But, say, there is a bug in the system, and some students don't have a corresponding Exam.

Will the query return the correct ones sorted, or will it fail?

How does Pony handles mismatches if _some_ of the related objects cannot be found?

If some grades are missing the corresponding students will be exluded from the result list To prevent this, you can create original query using left_join: query = left_join(s for s in Student) query = query.filter(lambda s: s.exam.grade)

Then group will be joined using LEFT JOIN, and student without group will not be excluded from the list Maybe Pony should use LEFT JOIN even with usual select

Thanks!

Hi, is it possible to perform Entity inheritance, with some classes sharing the same attributes? example adding interest to student and professor: class Person(db.Entity): name = Required(str) disc = Discriminator(str) class Personal(Person): _discriminator_ = “personal” role = Optional(str) class Student(Person): _discriminator_ = “student” gpa = Optional(Decimal) mentor = Optional("Professor") interest = Optional(‘str’) class Professor(Person): _discriminator_ = “professor” degree = Required(str) students = Set("Student") interest = Optional(‘str’)

use Mixin ?

Hi, currently you need to move interest attribute to a common base class of Student and Professor (if it make sense) or you need to name attributes differently class Person(db.Entity): name = Required(str) disc = Discriminator(str) class Personal(Person): _discriminator_ = “personal” role = Optional(str) class PersonWithInterest(Person): interest = Optional(‘str’) class Student(PersonWithInterest): _discriminator_ = “student” gpa = Optional(Decimal) mentor = Optional("Professor") class Professor(PersonWithInterest): _discriminator_ = “professor” degree = Required(str) students = Set("Student") or class Person(db.Entity): name = Required(str) disc = Discriminator(str) class Personal(Person): _discriminator_ = “personal” role = Optional(str) class Student(Person): _discriminator_ = “student” gpa = Optional(Decimal) mentor = Optional("Professor") student_interest = Optional(‘str’) class Professor(Person): _discriminator_ = “professor” degree = Required(str) students = Set("Student") professor_interest = Optional(‘str’) Currently entity mixins can't add attributes, only methods

I see, thanks Alexander

Is there any way to drop the objects cache, or tell Pony not to cache some objects when creating them? E.g., I have some University object that has add_student method, and I want to add a bunch of students in a cycle, but I do not want to cache them.

Currently you can't clear cache partially. You can use low-level db.insert method to insert rows without creating correspondng objects https://docs.ponyorm.org/api_reference.html?highlight=insert#Database.insert

I'm trying to use Pony as storage backend in some project. The user can choose between a basic in memory dict, mongodb and hopefully sql databases with ponyorm: I basically need to store (and retrieve by chat_id, user_id) the following data:state = { 'chat_id': 1234, 'user_id': 1234, 'state': 'SOMETHING_CAPSLOCK', 'data': {'any': 'valid', 'json': True}, } The problem is, I can't create a database table as I don't have access to the db.Entity object I'd need to subclass, as the user has to create that at a later time.

Sound basically like another use case for the later.Entity, as discussed above:

Digging through it the problem so far is that there is no easy to use database.register(Entity) but all the checking and code is in the EntityMeta making that hard extend, use and so to postpone it until we have the database ready

^ this

Hey @metaprogrammer, did you have the time to look into registering an Entity to a Database at a later time? I just got another use case, that I have a Users table I want to use in different projects. from somewhere import User class Post(later.Entity): poster = Required(User) text = Required(str) # end class db = Database('postgres://...') User.register_db(db) db.register(Post) In that way I could init just that table if I don't already have it, - without creating all the unneeded tables from the other project - without having to duplicate the code and keeping it up to date through several (7+) projects

^ and this

operations matematicas in pony?

+, -, * y/

'Primary key attribue %s was not found in query result set' % attr)

helpme, please!

Show us your code

Instead of Pedido.select_by_sql("select sum(...) ...") use db.select("select sum(...) ...") because this query does not return Pedido instance