Sangria: generalized record access

It is a tiny parse transform.

License: MIT

Author: Uvarov Michael ([email protected])

Installation

Add this repository as dependence of your project:

{deps, [
   {sangria,  ".*", {git, "git://github.com/mad-cocktail/sangria.git", "HEAD"}}
   ]}.

Include this command into each file, where you want to use this parse transform:

-compile({parse_transform, sangria}).

Idea

If you need to set same field for different records, you can write something like this:

set_label(L, X=#node{}) -> X#node{label = L};
set_label(L, X=#edge{}) -> X#edge{label = L}.

get_label(#node{label = L}) -> L;
get_label(#edge{label = L}) -> L.

The idea of this parse transform is to let the compiper to do this work for us:

set_label(L, X) -> with_any([node, edge], X#any{label = L}).

get_label(X) -> with_any([node, edge], X#any.label).

The first argument is a list of the record names and the second one is an generalized action with X#any, where X is one of the records.

If nothing matched, then the badmatch error will occured.

If you want to handle unmathed values, then use the 3rd argument:

set_label_or_ignore(L, X) ->
    with_any([node, edge], X#any{label = L}, X).

set_label_or_error(L, X) ->
    with_any([node, edge], X#any{label = L}, error(bad_record)).

set_label_or_undefined(L, X) ->
    with_any([node, edge], X#any{label = L}, undefined).

The third argument will be evaluated, if nothing matched. It is disallowed to use #any record inside this argument.

Mix Rum and Sangria

If you want to use Rum, than this parse transform must be executed before Rum:

-compile({parse_transform, sangria}).
-compile({parse_transform, rum}).