I guess that would download the files before, I mean I need to get all the python dependencies too, so thats not that problematic to add a extra step there

@micaiahparker In the company I work at, we publish the project template we use internally for pretty much everything.

https://github.com/thorgate/django-project-template

Might be interesting

https://github.com/pythonql/pythonql

Pony-like thing

Looks interesting. But this is not a standard Python

Hello! Is it possible working with multiple db with cross-links?

e.g. user table in db1 and posts in db2?

Yes, but you need to process foreign keys manually: class User(db1.Entity): name = Required(str) class Post(db2.Entity): user_id = Required(int) title = Required(str) content = Required(str) ... with db_session: u1 = User.get(name='John') posts = select(p for p in Post if p.user_id == u1.id)

ok! Thank you.

q2. Can I use Sets bw 2 db?

No, but maybe it is not hard to do an equivalent operation without using Sets

https://github.com/stsouko/predictor

I try to implement project for online predictions of chemical properties and data store

this is my db model: https://github.com/stsouko/predictor/blob/master/MWUI/models.py

I want to combine 3 db

A way for replace set is to implement property with select query?

class Users(db_main.Entity): id = PrimaryKey(int, auto=True) active = Required(bool, default=True) email = Required(str, unique=True) password = Required(str) user_role = Required(int) tasks = Set('Tasks') token = Required(str) restore = Optional(str)

Why do you want to split data between several databases?

Probably because they already are?

db_data is db for storing train set data

db_pred is users saved predictions.

db_main store only articles for web-site

What database do you use?

postgres

data for train can be used separately

I think yes, you can define property with select inside to emulate Set across different database. Or just use select directly, like in the example I wrote with users and posts

Ok. Thank You! property is more clear.

I think

If I remember correctly, in PostgreSQL a server is called "database" and a database is called "schema". You can have three different schemata inside the same physial server. In that case, you can use normal Sets between schemata. In order to do this, you need to specify table names manually, as a list or a tuple of two components: schema name and table name. For example: db = Database() # common db object for all schemata class User(db.Entity): _table_ = ['db_main', 'users'] tasks = Set("Task") ... class Task(db.Entity): _table_ = ['db_pred', 'tasks'] user = Required("User") ... By the way, it is better to name entities in a singular form: User instead of Users, etc.

Hi. How i can create unique index for column by clause UPPER(column)?

In this recipe how to create tables in db? Manually or pony can?

In this recipe how to create tables in db? Manually or pony can?

If you create three empty schemata inside a single postgresql server, and specify two-element _table_ name for each entity, Pony should be able to create all tables automatically

Hi. How i can create unique index for column by clause UPPER(column)?

You can add this manually by executing corresponding CREATE INDEX command. When we release migrations it will be possible to put this code inside custom migration

Thank you very much!

Sure

Why not allow use raw_sql expressions for indexes? This way to got overloaded entity model in Python?

It's a good idea, we probably can implement it. You can open a corresponding issue. I think it may take some time to implement, because other tasks (like migrations or new JSON serialization API) right now have a bigger priority

Postgres has a plugin to have a select over multible databases (&hosts)

Hey a new stickerpack. The 2859th I came across.

That's cool

Flagging NSFW stickers is a lot of work :(

Anyone has experience with all that social stuff? Currently it is only crawling and displaying, but I want to add some user experience, like voting and stuff

For 'Recently Used' you could probably get away with a displaying them in a grid and then adding the emoji, open button, and flag button to the bottom of each sticker image

For 'Recently Used' you could probably get away with a displaying them in a grid and then adding the emoji, open button, and flag button to the bottom of each sticker image

oh, thats an awesome idea.

Also you might want to display page+1 so that users first page is page 1

Also you might want to display page+1 so that users first page is page 1

you are absolutely right!

The discovering process is funny, I need to use a regular Telegram client to look up the sticker pack, the Telegram Bot API doesn't feature that :S

Wow, you should request that haha

To summarize how it works:

what would an example query look like

Always fun when the API blocks me for 86000 seconds (~23:40 hours).

what would an example query look like

Regarding what?

Oh I guess I was trying to use it wrong

Like, if I was in a convo and I wanted to use @StickerTagBot to send a sticker to that person

Although you might want to add limits to prevent getting blocked like that haha

Like, if I was in a convo and I wanted to use @StickerTagBot to send a sticker to that person

Problem is currently, that the Database is too slow to answer telegram in times, so I basically get "Error: you are to slow" back when trying the inline query.

Oh

Would it be faster if you swapped out Flask for something faster? I hear Sanic (however stupid the name) is freaaaky fast

And its async which sounds good for something like this

Oh, never mind, I killed the query somehow. TypeError: Function 'startswith' cannot be used this way: query_text_lower.startswith(p_t.string.lower())

Hmm

Shouldn't it be the other way around? p_t.string.lower().startswith(query_text_lower)?

Shouldn't it be the other way around? p_t.string.lower().startswith(query_text_lower)?

Probably

for text search http://blog.scoutapp.com/articles/2016/07/12/how-to-make-text-searches-in-postgresql-faster-with-trigram-similarity

I am trying a trigram index, see if that speeds up str in clolumn, and I can use the in parameter again

By the way, do you currently have index on lower(p_t.string)?

I was thinking about deploying a elasticsearch instance to make that string based querying faster, and also get a relevance factor but keeping the data in sync is a problem of it's own.

By the way, do you currently have index on lower(p_t.string)?

Following the article linked above, I uses CREATE INDEX customer_names_on_last_name_idx ON customer_names USING GIN(last_name gin_trgm_ops);

To be specific: -- http://blog.scoutapp.com/articles/2016/07/12/how-to-make-text-searches-in-postgresql-faster-with-trigram-similarity CREATE EXTENSION IF NOT EXISTS pg_trgm; -- index on tag.string CREATE INDEX IF NOT EXISTS tag_string_trigram_idx ON tag USING GIN(string gin_trgm_ops);

Article says, it is case insensitive anyway, so that's not a problem

Indeed, it was not timeouting, but the query was wrong.

Thats better

I thought I had the PonyORM sticker in my database as well

This one

You should) But may be you can add ponyorm.com under it?

What do you think?

How can I use ILIKE?

You can write raw sql fragment

something like and raw_sql('table_name.column_name ILIKE $(python_expr)')

There is nothing built-in?

Python does not have equivalent operation, I think it should be easy to write raw sql fragment inside a generator query

My current line with that is orm.exists(p_t for p_t in st.sticker.sticker_pack.tags if query_text_lower in p_t.string) I was thinking there could be a modifyer like orm.desc(...) Like orm.ignorecase(query_text_lower in p_t.string) or query_text_lower in orm.ignorecase(p_t.string)

something like and raw_sql('table_name.column_name ILIKE $(python_expr)')

Do I have to make the '%' escaping myself there?

My current line with that is orm.exists(p_t for p_t in st.sticker.sticker_pack.tags if query_text_lower in p_t.string) I was thinking there could be a modifyer like orm.desc(...) Like orm.ignorecase(query_text_lower in p_t.string) or query_text_lower in orm.ignorecase(p_t.string)

Or maybe even a function orm.ilike(query_text, p_t.string)

Curretly we don't support such modifiers. You can process escapes manually outside of a query: like_template = '%' + query_string.replace('!', '!!').replace('%', '!%').replace('_', '!_') + '%' select(p_t for p_t in st.sticker.sticker_pack.tags if raw_sql("p_t.string ILIKE $like_template ESCAPE '!'"))

Maybe I can modify my index to have LOWER(string) instead of string: CREATE INDEX IF NOT EXISTS tag_string_trigram_idx ON tag USING GIN(LOWER(string) gin_trgm_ops); And just query query_string.lower() in p_t.string.lower() instead of query_string in p_t.string

Yes, it is possible as well

I'll try that, not sure if the index part will work

Curretly we don't support such modifiers. You can process escapes manually outside of a query: like_template = '%' + query_string.replace('!', '!!').replace('%', '!%').replace('_', '!_') + '%' select(p_t for p_t in st.sticker.sticker_pack.tags if raw_sql("p_t.string ILIKE $like_template ESCAPE '!'"))

With the raw_sql I always am missing the options to include the parameters I want as actual arguments, like "{} {}".format(arg1, arg2) which feels safer to sql injection and also IDE's won't mark variables as unused.

I'll try that, not sure if the index part will work

Index creation seems to work

When you use $expr inside raw_sql it should convert it to actual parameter. About IDEs, maybe we need to allow to explicitly pass variables to raw_sql, like in python_expr = 'hello' raw_sql("column ILIKE $expr", expr=python_expr)