Parsing signed-float error

[GiggleLiu]

HELP NEEDED.

This PR contains

• Fixed 3 bugs.
  1. can not mutate options when using keyword arguments priority="invert".
  2. fix #27
  3. fix a priority comparing issue, non-boolean used in comparing.
• Adding a tree printing functionality to help debugging, which I find very helpful. (NOTE: AbstractTrees is a very light weighted dependency, the increased loading time is negligible)

  julia> Lerche.print_tree(types; maxdepth=10)
  object
  └─ genericobj3
  ├─ list
  │  ├─ object
  │  │  └─ genericobj3
  │  │     ├─ list
  │  │     │  ├─ object
  │  │     │  │  └─ list
  │  │     │  │     └─ object
  │  │     │  │        └─ 0
  │  │     │  └─ object
  │  │     │     └─ list
  │  │     └─ "Core.Array{Core.String, 1}"
  │  └─ object
  │     └─ genericobj3
  │        ├─ list
  │        │  ├─ object
  │        │  │  └─ list
  │        │  └─ object
  │        │     └─ list
  │        └─ "Base.Dict{Core.String, Core.Tuple{Core.Array{Core.String, 1}, Core.Array{Core.String, 1}}}"
  └─ "JugsawIR.TypeTable"

The following is a minimum one

julia> g = Lark("""start: a
                       ?a: b
                       b: "x"
                    """,parser="lalr",lexer="standard",debug=true);

julia> r = Lerche.parse(g,"x")
Tree(start, Any[Tree(b, Union{Token, Tree}[])
])

julia> print_tree(r)
start
└─ b

One issue not yet resolved

An unfixed one is kind of related to #29 . The parser does not know how to parse "SIGNED_FLOAT" and "INT" properly.

Setting priority does not work. Why it always parse INT first? Even if I define integer and float manually and set their priority manually, the parser still parse integer first.

[GiggleLiu]

@jamesrhester Sorry for mixing multiple features in one PR, this is just convenient for my local development. Really need your help in parsing floating point numbers and integers properly.

[jamesrhester]

I would like to keep the AbstractTrees changes in a separate PR, and I will merge the two obvious bugfixes in fe65bf8 immediately.

[GiggleLiu]

Sounds good.