Fix #96 Add EBNF grammar for ROS2 names in self-testing python script

[tkruse]

trafficstars

Prototype for #96. Since last time I extended a bit. Because of the fully-qualified vs. non-fully-qualified, the grammar has grown a bit.

The grammar still does not prevent double underscores, the python script around it can handle that much easier I think.

The script itself does not yet handle maximum name length constraints.

[tkruse]

Some things to clarify (consider clarifying in the article):

• is '{foo}' only allowed? -> current script assumes yes
• is /foo/123/bar allowed? -> current script assumes yes
• "balanced" curly braces may imply nesting? -> current script assumes no
• ~ only in starting position? -> current script assumes yes, only in starting position
• url scheme only 'rostopic' and 'rosservice' allowed? -> current script allows any alphabetic name

[wjwwood]

Thanks for putting the current state into a pr, that's easier to track.

is {foo} only allowed? -> current script assumes yes

I'd say yes, so long as the contents of foo are valid on their own, i.e. not empty.

is /foo/123/bar allowed? -> current script assumes yes

I would say no, the section for name tokes says:

https://github.com/ros2/design/blob/gh-pages/articles/115_topic_and_service_name_mapping.md#name-tokens

Topic and service name tokens:

...

• must not start with numeric characters ([0-9])

Since 123 in your example is a token, I think it must start with an alphabetical character.

"balanced" curly braces may imply nesting? -> current script assumes no

I suppose that it could, but the document does not appear to be clear on that point right now. I vaguely remember discussing it, but I don't see it in the issue history at the moment, so I'll open an issue to address that concern: https://github.com/ros2/design/issues/98

~ only in starting position? -> current script assumes yes, only in starting position

Yes, that is described in the document I as:

https://github.com/ros2/design/blob/gh-pages/articles/115_topic_and_service_name_mapping.md#ros-2-topic-and-service-name-constraints

For convenience here is a summary of all rules for topic and service names in ROS 2: ...

• may contain alphanumeric characters ([0-9|a-z|A-Z]), underscores (_), or forward slashes (/)
• ...
• may start with a tilde (~), the private namespace substitution character

It is also mentioned later in the document in more explicit terms.

url scheme only 'rostopic' and 'rosservice' allowed? -> current script allows any alphabetic name

I think that is find for the script, for now. I imagine we'll want to limit what goes there, but I also don't think rostopic and rosservice will be exhaustive in the end. We can leave it open ended for now (like with the substitution names).

[wjwwood]

I just wanted to clarify my response to:

is /foo/123/bar allowed? -> current script assumes yes

The document says, more specifically:

https://github.com/ros2/design/blob/gh-pages/articles/115_topic_and_service_name_mapping.md#name-tokens

Topic and service name tokens:

• ...
• may use alphanumeric characters ([0-9|a-z|A-Z]), single underscores (_), and/or balanced curly braces ({})
• must not start with numeric characters ([0-9])

So, it cannot start with a numeric (so 123 is not a valid token name), but it can start with more than just an alpha, i.e. _ underscore or { curly brace.

Thanks @dhood for pointing this out.

[tkruse]

Ah, somehow I missed the "token" bit. That makes the EBNF simpler to write. Updated the PR, added more positive and negative examples to tests.