BLOB type is always decoded to string

[robinvandernoord]

I'm having an issue with blobs in the database. According to the documentation, this field type is mapped to Python str. When looking at the code, this is done with .decode('utf-8'). However, in our database we have some raw images, which of course can't be decoded as utf-8. When looking through the adapter code (emmett/orm/adapters.py 229: def parse), I saw an option blob_decode. However, I can't seem to find where to set that option. Is this an argument to app.config.db or adapter_args? I have monkey patched the emmett.orm.adapters.parse method to forcefully set blob_decode to False, but that's very hacky and I've already seen some selects where my patch is not applied.

Summary: how do I get raw bytes from my bytea/blob database fields?

[gi0baro]

BLOB encoding/decoding is a pyDAL feature kept in place for retro-compatibility reasons.

In order to use raw BLOB values with postgres, you should define your own parser/representer, eg:

from emmett.orm.engines import adapters
from emmett.orm.engines.postgres import (
    PostgresAdapter, 
    JSONBPostgreParser,
    JSONBPostgreRepresenter,
    parse_type,
    repr_type
)

class PlainBlobParser(JSONBPostgreParser):
    @parse_type('blob')
    def _blob(self, value):
        return value

class PlainBlobRepresenter(JSONBPostgreRepresenter):
    @repr_type('blob')
    def _blob(self, value):
        return value

@adapters.register('postgres')
class PlainBlobPostgresAdapter(PostgresAdapter):
    def _load_dependencies(self):
        super()._load_dependencies()
        self.parser = PlainBlobParser(self)
        self.representer = PlainBlobRepresenter(self)

I know it's not super fancy, eventually for 2.6 I came up with a setting in Database config to do this.

[robinvandernoord]

Thanks for the solution, I'll try this soon!