EDIT: Ehh, I missed that you also want to consider words separated by space, Foo Bar AND Baz. That won’t work here too I guess.
Just realized its similarity with Erlang expression, and I couldn’t resist abusing it.
expr = ~S|("Modesty Blase" AND "X") Or not "Willie Garvin and" and Foo|
out =>
"(\"Modesty Blase\" AND \"X\") Or not \"Willie Garvin and\" and Foo"
{:ok, tokens, _EndLine} = :erl_scan.string(to_charlist(expr))
tokens
out =>
[
{:"(", 1},
{:string, 1, ~c"Modesty Blase"},
{:var, 1, :AND},
{:string, 1, ~c"X"},
{:")", 1},
{:var, 1, :Or},
{:not, 1},
{:string, 1, ~c"Willie Garvin and"},
{:and, 1},
{:var, 1, :Foo}
]
tokens =
Enum.map(tokens, fn
{:var, anno, var} ->
term = String.downcase(to_string(var))
if term in ~w(and or not) do
{String.to_existing_atom(term), anno}
else
{:string, anno, to_string(var)}
end
{op, anno} when op in ~w|( )|a ->
{op, anno}
{type, anno, value} when type in ~w(string atom integer)a ->
{:string, anno, to_string(value)}
{boolean_op, _anno} = token when boolean_op in ~w(and or not)a ->
token
{op, anno} ->
{:string, anno, to_string(op)}
end)
out =>
[
{:"(", 1},
{:string, 1, "Modesty Blase"},
{:and, 1},
{:string, 1, "X"},
{:")", 1},
{:or, 1},
{:not, 1},
{:string, 1, "Willie Garvin and"},
{:and, 1},
{:string, 1, "Foo"}
]
IO.puts("Expression: " <> expr)
{:ok, abs_form} = :erl_parse.parse_exprs(tokens ++ [{:dot, 1}])
abs_form
out =>
Expression: ("Modesty Blase" AND "X") Or not "Willie Garvin and" and Foo
[
{:op, 1, :or, {:op, 1, :and, {:string, 1, "Modesty Blase"}, {:string, 1, "X"}},
{:op, 1, :and, {:op, 1, :not, {:string, 1, "Willie Garvin and"}}, {:string, 1, "Foo"}}}
]
You can then use the parsed abs_form term for further analysis.
Although the hack seems to work, but there is one major issue, it seems erl_scan dynamically creates atoms