I was working on a preceding system for AST unparser and @pablogsal said there is an issue with docstrings (they printed in the same line even it was like 1000 chars long). So determining something is a triple quoted or not is simple with preceding system but I think it would be better if we can do it simpler and public. 

What I am thinking is using col_offset and end_col_offset information to determine if a string is a triple quoted or not. With a function called `ast.is_triple_quoted` or `ast.get_quote_count` etc. 

I can submit a patch if this wanted.I dont think this is big enough to discuss it on python-ideas but if it is, then I'm happy to port discussion over there.

I'm not sure the usage should be covered by "Abstract" syntax tree.

Isn't it a "Concrete" syntax tree?

It is impossible because of using newline escaping and string literal concatenation. In the following examples 
lineno != end_lineno, but strings use single quotes:

    print('Hello '
          'world!')

    print('Hello\
 world!')

Triple quotes can be also used for strings which occupy a single line of code and do not contain \n:

    print("""exec("print('Hi!')")""")

We cannot also distinguish 1234 from 0x4_d2 and 1234.0 from 1.234e3 at the AST level.

@inada.naoki I thought that would be a nice to have feature in certain cases AST used but as @serhiy.storchaka said there is an issue with implicit string concatenation which makes this impossible to expose. Thank you both.