Calling bot.brain.get("foo") fails if foo does not exist. It should return None instead.

Example error: https://app.ab.bot/skills/read-empty-key-repro/activity/2c453d65-7e1a-4601-868c-ef421f3069c2

Python Runner exception:

Runtime Error
Traceback (most recent call last):
File "/home/site/wwwroot/SkillRunner/bot/apiclient.py", line 91, in send
result.raise_for_status()
File "/home/site/wwwroot/requests/models.py", line 943, in raise_for_status
raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 404 Client Error: Not Found for url: https://app.ab.bot/api/skills/2384/brain?key=some_key

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/site/wwwroot/SkillRunner/bot/bot.py", line 166, in run_user_script
policy.exec(self.code, script_locals)
File "/home/site/wwwroot/SkillRunner/bot/policy.py", line 328, in exec
exec(byte_code, globals)
File "<skill code>", line 1, in <module>
File "/home/site/wwwroot/SkillRunner/bot/storage.py", line 27, in get
output = self._api_client.get(path)
File "/home/site/wwwroot/SkillRunner/bot/apiclient.py", line 56, in get
return self.send(path, 'GET')
File "/home/site/wwwroot/SkillRunner/bot/apiclient.py", line 96, in send
self.logger.exception(f"There was an error {method}ing to {url}", e)
File "/usr/local/lib/python3.9/logging/__init__.py", line 1481, in exception
self.error(msg, *args, exc_info=exc_info, **kwargs)
File "/usr/local/lib/python3.9/logging/__init__.py", line 1475, in error
self._log(ERROR, msg, args, **kwargs)
File "/usr/local/lib/python3.9/logging/__init__.py", line 1589, in _log
self.handle(record)
File "/usr/local/lib/python3.9/logging/__init__.py", line 1599, in handle
self.callHandlers(record)
File "/usr/local/lib/python3.9/logging/__init__.py", line 1661, in callHandlers
hdlr.handle(record)
File "/usr/local/lib/python3.9/logging/__init__.py", line 952, in handle
self.emit(record)
File "/azure-functions-host/workers/python/3.9/LINUX/X64/azure_functions_worker/dispatcher.py", line 713, in emit
msg = self.format(record)
File "/usr/local/lib/python3.9/logging/__init__.py", line 927, in format
return fmt.format(record)
File "/usr/local/lib/python3.9/logging/__init__.py", line 663, in format
record.message = record.getMessage()
File "/usr/local/lib/python3.9/logging/__init__.py", line 367, in getMessage
msg = msg % self.args
TypeError: not all arguments converted during string formatting

When loading the properties of bot, we do stuff like this:

self.from_user = skillInfo.get('From')

That works fine for simple properties, but in the case of bot.from_user it leads to surprising results. For example, to get the user name, you can't do the following:

1. bot.from_user.name NOPE!
2. bot.from_user.Name NOPE!

Instead you have to do this: bot.from_user.get('Name'). I don't think any python skill author would expect that. @pmn thinks there's a built-in that might help with this.

Reported by user netsplit in Discord:

When sending a request to a trigger, I'm getting HTTP 500 responses back with the following message. How would I setup my function to ensure HTTP 200 is returned?

[
  {
    "ErrorId": "AttributeError",
    "Description": "'TriggerResponse' object has no attribute '_raw_content'",
    "LineStart": 0,
    "LineEnd": 0,
    "SpanStart": 0,
    "SpanEnd": 0,
    "StackTrace": null
  }
]

See https://github.com/aseriousbiz/abbot/pull/3469 for details on the endpoint.

The tweet skill uses urllib to do its thing. Please include it into the runner so that we can tweet from chat.

When I run PIP I get a warning:

DEPRECATION: Python 2.7 reached the end of its life on January 1st, 2020. Please upgrade your Python as Python 2.7 is no longer maintained. pip 21.0 will drop support for Python 2.7 in January 2021. More details about Python 2 support in pip can be found at https://pip.pypa.io/en/latest/development/release-process/#python-2-support pip 21.0 will remove support for this functionality.

What Python should we be on and is this something we can easily set up in bootstrap?

The production API endpoints are the fallback if the runner's environment does not specify an endpoint. This was done when we were running into issues maintaining env clearly. Since that problem has since been resolved, the fallback endpoints should point to https://localhost:4979 to make local development simpler (and to break correctly if the runner environment isn't set).

The variables that need to be updated are AbbotReplyApiUrl and SkillApiBaseUriFormatString.

The tweet skill stopped working sometime in the past month. It's reporting an exception when run:

name '_unpack_sequence_' is not defined

Noticed that someone was getting an error trying to run the sql skill:

libpq.so.5: cannot open shared object file: No such file or directory

This means that the psycopg2 binary was not loaded (the postgres drivers). Something is happening when we deploy abbot-py via Actions versus local deployment that is causing binaries to not load correctly.

The short term workaround is to deploy from local machine with the following, but the Actions deploy should always Do The Right Thing:tm:

Workaround command (run from the src directory):

func azure functionapp publish abbot-skills-python --verbose --python

Received a request via [email protected] for this.

Originally noticed when the yelling skill started raising exceptions.

It's unclear if this is a write issue or a read issue, but I'm guessing a read issue given the error message:

[NameError] name '_getitem_' is not defined

This skill has a repro, although I'm working on making the repro case smaller: https://app.ab.bot/skills/py-brain-test

Follow up to https://github.com/aseriousbiz/abbot/pull/2175

Looks like bot.brain.read has the correct behavior.

It's a lot easier to write Abbot Python skills if developers can see how their code is run. Plus, there are a lot of smart Python developers out there who might have great improvements to how we run their code.

Before we open source anything, we need to:

• Review the source code to ensure there are no secrets.
• Review the commit history to ensure there are no... inappropriate commit messages.
• Get a 👍 from @haacked, who is the permission-giver for these sorts of things.

The Python runner is currently not responding.

Originally reported by a user in Discord who was modifying their own version of a SQL command and noticed that even though the command completed successfully, Abbot was reporting an exception after some period of time.

On review of the audit logs, it looked like there was some kind of a timeout. My suspicion is that opening a connection pool with psycopg2 is keeping the runner working until it gets killed.

Here's the code they're using:

import psycopg2
import pandas


def run_query(query):
  connstr = bot.secrets.read("connstring")

  if not connstr:
    return "There's no connection string set up. Please add one before running this skill."
  
  with psycopg2.connect(connstr) as conn:
    conn.autocommit = True
    df = pandas.read_sql(query, conn)
    return f"```{df.to_markdown()}```"
   
QUERIES = {
    "test": """ redacted""",
    "astatus": """ redacted """,
}

# Check to see if the user has asked for a predefined query.
if bot.arguments in QUERIES:
  result = run_query(QUERIES[bot.arguments])
else:
  result = "I dont' recognize that query. Available queries are: "
  for k, _ in QUERIES.items():
    result += f"\n * {k}"

bot.reply(result)                    

When using a custom image, Abbot's console and abbot-cli return blank responses even if the skill ran correctly.

To replicate, set up a custom Python runner. The instructions for setting up a runner on Fly are good for this.

Once the custom runner is set up, create a new Python skill. The default skill code is sufficient for testing. Try running the skill in the console, and then run in the Slack workspace. Currently, the console will return a blank response and running in Slack will return correctly:

console:

Slack:

The audit log is reporting all runs successful:

Brand new MacBook M1. When I run script/server for this repository (I already ran script/bootstrap, I get an error.

ImportError: cannot import name 'cygrpc' from 'grpc._cython' (/usr/local/Cellar/azure-functions-core-tools@3/3.0.3904/workers/python/3.8/OSX/X64/grpc/_cython/__init__.py)

In JS-land, https://github.com/aseriousbiz/abbot-js/pull/94 populates process.argv with skill arguments. Python should do the same thing.