Assigned Record Access

It is a tiny parse transform.

License: MIT

Author: Uvarov Michael ([email protected])

Allows to use a short access syntax (dot-syntax) for getting values from records.

Example 1

In this exampe, the call of access(#rec{}). returns 1.

Before:

-record(rec, {f1 = 1, f2 = 2}).

access(X = #rec{}) ->
    X.f1.

Other variant:

-record(rec, {f1 = 1, f2 = 2}).

access(#rec{f1 = F1}) ->
    F1.

After:

-compile({parse_transform, arak}).
-record(rec, {f1 = 1, f2 = 2}).

access(X = #rec{}) ->
    X.f1.

Illegal form:

access(X) ->
    X.f1.

Example 2

Not only match syntax can be used for getting the type of the record variable. This example shows, how a case syntax can be used instead. Binded variables can be used only inside clauses (they are not exported as in Erlang).

-compile({parse_transform, arak}).
-record(rec,  {f1, f2}).
-record(rec1, {f1, f2}).

bad_case(X) ->
    case X of
         #rec{}  -> X.f1;
         #rec1{} -> X.f1
    end.

Illegal form:

bad_case(X) ->
    case X of
        #rec{}  -> ok;
        #rec1{} -> ok
    end,
    X.f1.

Example 3

We can get mnemosyne-like fields combining this parse transform with QLC.

-compile({parse_transform, arak}).
-include_lib("stdlib/include/qlc.hrl").

revision_ids(RepId, undefined, Hashes) when RepId =/= undefined ->
    Q = qlc:q([R || R=#g_revision{} <- mnesia:table(g_revision),
                    B=#g_branch{}   <- mnesia:table(g_branch),
                    Hash <- Hashes,
                    B.id  =:= R.branch,
                    Hash  =:= R.commit_hash,
                    RepId =:= B.repository]),
    qlc:e(Q).