pr eso use el Float

previous value of _ingresos should be Decimal too

it is not possible to mix floats and decimals

previous value of _ingresos should be Decimal too

how?

I don't know how you initially construct it

should be _ingresos = Decimal('0.0')

0.00 is float, not decimal

wauuuuu!! thanks!! @metaprogrammer

Sure

problem solved, thanks for your help

generated without errors

Hi folks. I'm looking for a way to build the json path used in the select statement for JSON structure queries (as shown here https://docs.ponyorm.org/json.html#querying-json-structures) in a dynamic way. I tried passing a custom subgenerator to the select function but this seems not to work because the select statement doesn't allow a lambda function with any more then a single parameter

Is there any other way how I could build the path used for quering dynamicly? (nested JSON structures)

Hi, what do you mean by building path in a dynamic way? Can you give an example?

quick example of what I'm trying to do. Let's take the example from the link I mentioned: Product.select(lambda p: p.info['display']['size'] > 5) —> I'd like to have "display" and "size" and possibly more to be variables

I'm aware that I can use variables for the slection but what I want is to pass an array, containing multiple strings (variables of arbitrary length) and then each of them is used as key for the select

An array of arbitrary length?

let me write a quick pseudo function

Also, what database do you use?

sqlite so variables should work in the select

let me write a quick pseudo function

Yeah, I think it would be nice to see an example

def get_custom_data(path: List[str], value:str): return Product.select(lambda p: p.info[path[0]][path[1]]...[path[n]] is value) get_custom_data(path=["my", "custom", "json", "path", "to", "select"], value="something")

if the path is of fixed length then this is easy. but I struggle with variable length...

I tried using a subgenerator that would build the path dynamicly and passing it into the lambda function of the select(), like this:

def deref(data, keys): yield from deref(data[keys[0]], keys[1:]) if keys else (yield data) Model.select(lambda x, y, z: next(deref(data=x, keys=y)) == z)

but this doesn't work. getting the error: Lambda query requires exactly one parameter name, like Model.select(lambda x: ...). Got: 3 parameters.

the reason for this seems to be here: https://github.com/ponyorm/pony/blob/b4d3da69d2fbd18c870db204d7de12a06e0b2328/pony/orm/core.py#L4353

Not sure if it's clear what I'm trying to do here... let me know if I should try to explain it in a different way.

Pony json expressions does not support pathes with a variable length. So you need to use a bit of raw sql here:

def find_product(path, value): sql_expr = '''CAST(py_json_extract("p"."info", '%s') as text)''' % path.replace('$', '$$') return Product.select(lambda p: raw_sql(sql_expr) == value) path = "$.key1.key2[10][20]" result = find_product(path, 'some_value') You need to construct path argument dynamically as a string

yes that's what I saw is used internally here https://github.com/ponyorm/pony/blob/a8b0a5ca5cd8fab48b27c97a65aee211952ae2ac/pony/orm/dbproviders/sqlite.py#L546

or, a bit simpler: def find_product(path, value): sql_expr = '''CAST(py_json_extract("p"."info", '$$.%s') as text)''' % path return Product.select(lambda p: raw_sql(sql_expr) == value) path = "key1.key2[10][20]" result = find_product(path, 'some_value')

I'll try that, thanks!

Works perfectly! Thanks alot!

Sure

Is there a way to make sure some values in a JSON attribute are unique? Currently when creating the entity I'm getting the relevant values from the JSON data and am saving them as a seperate key with unique set to true.

In some databases it is possible to construct indexes (including unique indexes) based on expressions (including JSON path expressions) At this moment Pony does not support support such indexes I think it is easier in your case just to define separate unique column. If you have such a column, maybe it is not necessary to have the same data in json field?

This is how you can define expression-based index in SQLite https://www.sqlite.org/expridx.html Ptobably at this moment you can't use py_json_extract in such expression

true but in my case I just want to dump the json data if the query does match so it makes sense to have the data duplicated since that field should be included as well

cool, thanks for the info. totatly works for me to have that column duplicated. I was just wondering if it's possible. haven't seen anything about it while browing the docs. would have been a "nice to have" but as I said: duplicated column is fine.

Is there any recommanded way to pass the pony.orm.Database()? I'm doing the database init in my main.py but then other modules/classes need to access it. Currently I just save it in an module level constant _DB and import it where needed (e.g. Model classes). A more elegant way might be to package it inside a singleton but maybe there is another best practise? Also I'm not sure what the implications of having a single _DB constant are considering it will be accessed by multiple workers of my WSGI application. Is there anything to look out for? Do I need seperate connections for every worker / thread or is that handled within the Database() instance?

There are several ways, but all of them consists of creating db = Database() on module level and then import it like from app import db or from models import db.

I see. And then connections are handled inside Database() in a thread-safe manner I guess?

So there's really nothing I need to handle? I remember in pewee there have been WSGI server impl. specific session handling decorators and stuff

I guess this answers my question: https://docs.ponyorm.org/api_reference.html?highlight=thread#database-class

"The Database object manages database connections using a connection pool. It is thread safe and can be shared between all threads in your application. "

Awesome. So it's all handled when using the db_session decorator or context manager

Finally a Python ORM that doesn't suck.

.😁

Is there a way to just return all entities of a model? Can't figure out what the lambda equivalent to "Select * from Product" would be...

orm.select(p for p in Product)[:] works thought. Still dunno how the lambda would look like

you can also do Product.select()[:]

I thought I tried that and it failed with some exception. But now that you mentioned it I tried it again and it worked O_o. I guess I'm getting tired.

hi, good morning, hellp with its error: pony.orm.sqltranslation.IncomparableTypesError: Incomparable types 'date' and 'str' in expression

you try put string in datetime

replace data['fetcha'] to datetime.strptime(data['fetcha'],'%Y-%m-%d')

thanks!!

Hi guys!! I cannot figure out how can stablish a 1-many relatión with a composite key. I tried with reverse/columns but... With no success

class Parent(db.Entity):     pk_1 = Required(int, size=64)     pk_2 = Required(int, size=64)     config = Optional(Json)     PrimaryKey(pk_1, pk_2)     watermarks = Set('Child') class Child(db.Entity):     pk_1 = Required(Parent, reverse='watermarks')     pk_2 = Required(Parent, reverse='watermarks')     pk_3 = Required(int, size=64)     config = Optional(Json)     PrimaryKey(pk_1, pk_2, pk_3)

class Parent(db.Entity): pk_1 = Required(int, size=64) pk_2 = Required(int, size=64) config = Optional(Json) PrimaryKey(pk_1, pk_2) watermarks = Set('Child') class Child(db.Entity): parent = Required(Parent, reverse='watermarks') pk_3 = Required(int, size=64) config = Optional(Json) PrimaryKey(parent, pk_3) The physical primary key of Child table will consist of 3 columns (parent_pk_1, parent_pk_2 and pk_3)

Hi Any news about migrations?

It’s almost done

😳 great)

We need to implement downgrade, test it and write documentation

It's 90% ready. Writing tests, finding some bugs, fixing it

:)

Thx Alexander

We need to implement downgrade, test it and write documentation

Ha?? Can't belive that. Although being an atheist I will light a candle for you after release.

Hehehe.

It’s not a main goal now but in future we want to support it. Main problem is transactions are kinda anti-async

Any quick way to convert TrackedDict returned by the JSON Data Type into a normal dict? I see that json.dumps works on TrackedDict but I don't want to convert it into JSON just to convert it back into a editable dict...