Thomas, maybe it would be easier to not have many tables and work out how to take the results from pandas and use pony to insert them into the database?

@matthewrobertbell there is always more than one path to solve this. I'll think about it.

Another idea would be to calculate the rankings on the fly by request, but this may cause heavy load, whenever too much requests are coming in.

I think you can likely have only one set of tables, without any splitting by league etc. You can still precalculate your records using one set of tables

@metaprogrammer good morning/evening

I spent some time working with the semi-broken clipboard API to copy postgres and python code out of editor using selenium (you may have noticed all my logins 😜

In any case to make basic export possible within the confines of selelnium headless given editors current configuration I think it's necessary to ask for one very small changer

when "Copy All" is pressed - could you save the output to a hidden node's data attribute? <div id="python_code_node" bytes="import ponyorm..." />

this way the code is accessible to selenium - otherwise I believe the clipboard API will cause much frustration across headless browsers

you can read a bit about it here - navigator.clipboard.readText().then(clipText => document.getElementById("outbox").innerText = clipText);

worthy to mention that I tried this solution and without success

https://medium.com/@sahajamit/getting-clipboard-content-from-remote-selenium-chrome-nodes-67a4c4d862bd

Hi @minskmaz , Is there any issue with copying of visible block of code?

yes - it's all nodes that I would have to parse through to make sense of - it's dom

all those nodes are for highlighting. If you remove html and concatenate the content you should get plain code. I wondering if huge hidden text can affect page performance.

In puppeteer you can listen to raw response (check get-code XHR) https://stackoverflow.com/questions/45822058/puppeteer-how-to-listen-to-a-specific-response

hm - well I could perhaps give a specific flag which if not present the write to the nodes wouldn't occur

checking the link you sent now

It's only possible with puppeteer (maybe also with phantomjs) but not with selenium

yes I could give an unsupprted (by ther server) flag in the url and ediutor js code could do the writes for the one instance

https://editor.ponyorm.com/user/someuser/test/designer?client=selenium

in the meantime I will look at the nodes and explior concat

if you know how to do the concat would you mind giving an example? It looks like spaghetti to me

sure, https://stackoverflow.com/questions/822452/strip-html-from-text-javascript

thanks I'll patch this in and see if it work Roman

Good luck :)

Roman I got it! Thank you for walking me through it buddy. I'll let you guys know when I have something to show.

How can I make Set optional?

Set can be empty

They are since they can be empty

They are since they can be empty

Thanks, they are

hey there is https://editor.ponyorm.com/ site down?

Hi, it works again

Awesome, thanks!

Hello. I have a problem with my code. database.py: from pony import orm db = orm.Database() class User(db.Entity): f_id = orm.Required(int) db.bind(provider='sqlite', filename=':memory:') db.generate_mapping(create_tables=True) bot.py: from threading import Thread class BotPollingThread(Thread): def __init__(self, bot_handler) -> None: super(BotPollingThread, self).__init__() self.setDaemon(True) def run(self) -> None: with orm.db_session: user = database.User.get(telegram_id=message.chat.id) main.py (startpoint): from bot import BotPollingThread def main(): BotPollingThread.start() if __name__ == '__main__': main()

When thread starting, i get error: pony.orm.dbapiprovider.OperationalError: no such table: User

But i binded database and made generate_mapping. What is my probem?

In order to use database from multiple threads you need to store it in a file

Replace filename=':memory:' to filename='db.sqlite'

Ok. I'll try. Thanks

Replace filename=':memory:' to filename='db.sqlite'

It works!

@metaprogrammer yes you have a HTTPS issue

We'll be able to fix it a bit later

hi all

just getting started with pony ORM

I may have a quaetion..

how do you initialize db_session before running your unit tests with nose2 ?

I used to do this with orm.db_session: unittest.TextTestRunner(verbosity=2).run(suite)

however with nose2 i can only specify an entry point for my commands

I would prefere use nose2 than unittest directly

Simplest way is to wrap your tests with db_session decorator

The other way is def setUp(self): rollback() db_session.__enter__() def tearDown(self): rollback() db_session.__exit__()

Or you can wrap test with a separate @db_session decorator

Ok thanks. I might go for the setUp solution as I don't want to wrp every single method

is rollback compulsory ?

do I need to import it to declare it inside my TestClass ?

Thanks!

Rollback goal is to return database after test is finished to previous state, so all tests see the same database state

rollback seems to be called within the exit function already

so it might be redundant to call it before

It called if there was some exception during the execution of db_session and we want to rollback unfinished transaction But in case of tests you need to roll back all db_sessions, even successful ones

cheers guys anyway

you rock

Or you need to recreate database state before each test

ok i'll give it a try:)

Hey I have a question about delete hooks. Assume I have a many to many relationship between a and b and I want a cascading deletion between them in the sense that if after a deletion of a any b that that has an empty a set should also be deleted. I wondered how to do that and stumbled upon deletion hooks. but when I try to delete the "to be orphaned" bs in a's before delete hook I get the exception that a is marked for deletion. I worked around this by manually deleting b afterwards but i would prefer something automatic. What's the best way to solve this in pony

I was doing: def before_delete(self): # a's hook delete(b for b in B if not b.a_set)

Is there a way to disable cascade delete knowing it might possibly invalidate linkages?

What do you mean?

Say I have a table of enums

And I drop one

But that is referenced by another table

Deleting an enum will kill all linked refs

Which is probably good

But I'm fighting this behavior and would prefer to orphan a reference

I think you if declare entities as class Enum(db.Entity): objects = Set("Object") class Object(db.Entity): enum = Optional("Enum", cascade_delete=False) it should works. Object.enum value will be set to None for deleted enums

Ahh, let me try that! This seems like what I was missing

So in this case, the use of Object.enum.Required will cause an exception, so use Optional even if required?

How it required if you want to orphan a reference

If I expect object to always have an Enum, but for some reason that Enum was deprecated..