initial prototype of decorator is in progress

[lukesmurray]

These are the initial semantics.

Getter Format

get {ClassName}.{PropertyName}\t{value}

Setter Format

set {ClassName}.{PropertyName}\t{oldValue} -> {newValue}

We do not log access for any properties whose name starts with underscore since these are "hidden".

We do not log anything which is touched while accessing the initial property. A simple example of when this occurs is Message.sender. Our code for Message.sender returns Message.from_. Without hiding nested logs we would log a getter for Message.sender and a getter for Message.from_. With the hiding we only log a getter for Message.sender.

Here is some code and an example of the output which looks good.

def on_message(my_message):
    my_message.subject
    my_message.sender
    my_message.deadline = datetime.now()

output

get Message.subject Testing Email 😎Yeet
get Message.sender [email protected]
set Message.deadline None -> 2019-05-25 17:08:53.215517

Now consider the following example. We have one line but we get two lines of logged output. This is because we are accessing two properties in a row. First we get the sender then we get the name. @soyapark maybe you have some insight about what we want to do here.

def on_message(my_message):
    my_message.sender.name

output

get Message.sender [email protected]
get Contact.name luke murray

Right now these features are only implemented in the simulation code not in the regular interpret code.

The current state of the code for this feature is very poor but the semantics are interesting.

[lukesmurray]

fixes #122

[lukesmurray]

Also added the ability to get the name of attachment files without downloading any data. message.attachments return a list of all the attached files.

fixes #125

[soyapark]

Now consider the following example. We have one line but we get two lines of logged output. This is because we are accessing two properties in a row. First we get the sender then we get the name. @soyapark maybe you have some insight about what we want to do here.

def on_message(my_message):
my_message.sender.name

hmm interesting! I think it would be nice to print out only the last results, but show the rest of traces if users want to see them i.e. we can fold the rest and users can expand it.

[soyapark]

Looks great overall! Didn't imagine this could be this simple & clean. Just left a few comments & suggestion

[soyapark]

Is this a comment?

[soyapark]

Can we move this if statement right after property_name = _get_method_name(self.fget) so we can do early-exit?

[soyapark]

repeating: Can we move this if statement right after property_name = _get_method_name(self.fget) so we can do early-exit?

[soyapark]

So we are just appending logs in one string object. Can we make it more parse-able? Can we include additional information which line of user's code this property log generated from? For example,

def on_message(my_message):
    if '😎' in my_message.subject and 'Luke Murray' == my_message.sender:
        my_message.deadline = datetime.now()

this will be

{ 2: [{'type': 'get', 'log': 'Message.subject	Testing Email 😎Yee'}, {'type': 'get', 'log': 'Message.sender [email protected]'}]
  3: [{'type': 'set', 'log': 'Message.deadline	None -> 2019-05-25 17:08:53.215517'}]