aioodbc support

[ColinBe95]

Is your feature request related to a problem? Please describe.

I am using aioodbc to connect to mssql asynchronously. However, eventhough aioodbc runs on top of pyodbc spans are not automatically captured.

Describe the solution you'd like

Native instrumentation for aioodbc similar to the existing support for aiopg and aiomysql.

Describe alternatives you've considered

/

Additional context

I am happy to create a PR if desired.

[xrmx]

It looks like aioodbc has not been touched since 3 years and this does not make it a good candidate to be added. Said that maybe there's something we can improve on the pyodbc instrumentation to make it work for aioodbc as well?

[ColinBe95]

I do believe that aioodbc is still widely used though. I guess there is simply not much development happening as it is simply a wrapper around pyodbc, running it in threads without blocking the event loop.

As far as I understand it the reason we are not seeing spans being created is because pyodbc.connect is called within a threadpool where the context is not propagated.

I was looking at the related PRs (aiomysql and aiopg) and noticed how they are both instrumenting their respective execute method and therefore was thinking this should be possible for aioodbc as well.

I was thinking about something like:

from elasticapm.instrumentation.packages.base import AsyncAbstractInstrumentedModule
from elasticapm.instrumentation.register import register
from elasticapm.traces import async_capture_span
from elasticapm.instrumentation.packages.dbapi2 import extract_signature
from elasticapm.utils import shorten


class AioodbcInstrumentation(AsyncAbstractInstrumentedModule):
    name = "aioodbc"
    
    instrument_list = [
        ("aioodbc.connection", "Connection.execute"),
    ]
    
    async def call(self, module, method, wrapped, instance, args, kwargs):
        sql = args[0] if len(args) else kwargs.get("sql", "")
        name = extract_signature(sql)
        query = shorten(sql, string_length=10000)
        
        context = {
            "db": {"type": "sql", "statement": query},
        }
        
        async with async_capture_span(
            name,
            leaf=True,
            span_type="db",
            span_subtype="odbc",
            span_action="query",
            extra=context,
        ):
            return await wrapped(*args, **kwargs)

WDYT?

[ColinBe95]

Hey @xrmx ,
do you see any chance in supporting aioodbc in the future? I am happy to assist/submit a PR if you think it's worth it. I personally dont think we can simply improve pyodbc support, but maybe I am wrong. I think this would be a great for everyone using Azure SQL