Improve exception management when Blob does not exist

When a function is triggered through HTTP with a named parameter (e.g., userparam) and we have a blob binding with such parameter in the name and it does not exist, the runtime throws:

System.Private.CoreLib: Exception while executing function: Functions.test_missing_blob. System.Private.CoreLib: Result: Failure
Exception: AttributeError: 'NoneType' object has no attribute 'type'
Stack:   File "...\azure-functions-core-tools\bin\workers\python\3.7\WINDOWS\X64\azure_functions_worker\dispatcher.py", line 306, in _handle__invocation_request
    pytype=pb_type_info.pytype)
  File "...\azure_functions_worker\bindings\meta.py", line 63, in from_incoming_proto
    return binding.decode(datum, trigger_metadata=metadata)
  File "...\azure-functions-core-tools\bin\workers\python\3.7\windows\x64\azure\functions\blob.py", line 78, in decode
    data_type = data.type

The definition of the function can be something like:

{
  "scriptFile": "__init__.py",
  "bindings": [
    {
      "authLevel": "anonymous",
      "type": "httpTrigger",
      "direction": "in",
      "name": "req",
      "methods": [
        "get"
      ]
    },
    {
        "name": "notebooksindexin",
        "type": "blob",
        "direction": "in",
        "path": "file-{userparam}.json",
        "connection":"AzureWebJobsStorage"
    },
    {
      "type": "http",
      "direction": "out",
      "name": "$return"
    }
  ]
}

The key here is to pass “userparam” such that “file-{userparam}.json” does not exist.

This eventually fails and returns a 500. However, I don’t think this is the most graceful behavior. My proposal would be to:

1. Instead of waiting all the way to the dispatcher to get an empty “data” field, identify this early and log it in a friendly way.
2. Surface it to the client with a full HTTP response that would say somewhere (maybe the body) that the blob is missing.

I am running the azure functions runtime locally and executing queries but I think that’s not relevant other than I should be able to test a possible solution fast.

[anthonychu]

Yes, I think both cases should pass None. We want to let user code decide what to do because an error isn’t always the desired result. Here’s an example of another language (C# script) where it receives null if the file doesn’t exist and the function handles it by creating the file with a blob output: https://github.com/anthonychu/slack-user-change-alerts/blob/master/src/SlackUserChangeAlerts.Function/CheckUserListChanges/run.csx#L13

[reubano]

Just realized something, how will we know the name of the blob we attempted to fetch if all we get passed is None?