neotoma.erl

-module(neotoma).
-author("Sean Cribbs <seancribbs@gmail.com>").
-export([file/1, file/2, bootstrap/0]).

%% @doc Generates a parser from the specified file.
%% @equiv file(Filename, [])
%% @spec file(Filename::string()) -> ok
file(InputGrammar) ->
    file(InputGrammar, []).

%% @doc Generates a parser from the specified file.
%% <pre>    Options = [Option] <br />
%%   Option = {module, OutputModuleName::atom()} | <br/>             {output, OutputDirectory::string()} | <br />             {transform_module, TransformModuleName::atom()} </pre>
%% @spec file(Filename::string(), Options) -> ok
file(InputGrammar, Options) ->
    Basename = filename:basename(InputGrammar, ".peg"),
    InputDir = filename:dirname(InputGrammar),
    ModuleName = proplists:get_value(module, Options, list_to_atom(Basename)),
    OutputDir = proplists:get_value(output, Options, InputDir),
    OutputFilename = filename:join(OutputDir, atom_to_list(ModuleName) ++ ".erl"),
    TransformModule = proplists:get_value(transform_module, Options, false),
    validate_params(filename:absname(InputGrammar),
                    ModuleName,
                    TransformModule,
                    filename:absname(OutputFilename)),
    Parsed = parse_grammar(InputGrammar),
    Rules = proplists:get_value(rules, Parsed),
    Root = proplists:get_value(root, Parsed),
    GenTransform = proplists:get_value(transform, Parsed),
    ModuleAttrs = generate_module_attrs(ModuleName, Root),
    TransformFun = create_transform(TransformModule, OutputDir, GenTransform),
    {ok, PegIncludes} = file:read_file(code:priv_dir(neotoma) ++ "/peg_includes.erl"),
    file:write_file(OutputFilename, [ModuleAttrs, "\n", Rules, "\n", TransformFun, "\n", PegIncludes]).

validate_params(InputGrammar, _, _, OutputFile) when InputGrammar =:= OutputFile ->
    throw({badarg, "Input and output file are the same!"});
validate_params(_,ModName,_,_) when not is_atom(ModName) ->
    throw({badarg, "Output module name is not an atom!"});
validate_params(_,_, false, _) -> ok;
validate_params(_,_, TransformModule, _) when not is_atom(TransformModule) ->
    throw({badarg, "transform_module option must be an atom"});
validate_params(_,Basename, TransformModule, _) when Basename =:= TransformModule ->
    throw({badarg, "Transform module named same as parser module!"});
validate_params(_,_, TransformModule, OutputFile) ->
    OutMod = list_to_atom(filename:basename(OutputFile, ".erl")),
    case OutMod of
        TransformModule -> throw({badarg, "Transform module file same as parser output file!"});
        _ -> ok
    end.

generate_module_attrs(ModName, Root) ->
    {RootRule,_} = Root,
    ["-module(",atom_to_list(ModName),").\n",
     "-export([parse/1,file/1]).\n",
     % This option could be problematic if your grammar is broken in
     % some way, but hides warnings about unused parser combinators
     % and unused Node/Idx variables in your transform functions.
     % In a future version we should just emit the used combinators,
     % excluding the rest.
     "-compile(nowarn_unused_vars).\n",
     "-compile({nowarn_unused_function,[p/4, p/5, p_eof/0, p_optional/1, p_not/1, p_assert/1, p_seq/1, p_and/1, p_choose/1, p_zero_or_more/1, p_one_or_more/1, p_label/2, p_string/1, p_anything/0, p_charclass/1]}).\n\n",
     "file(Filename) -> {ok, Bin} = file:read_file(Filename), parse(binary_to_list(Bin)).\n\n",
     "parse(Input) ->\n",
     "  setup_memo(),\n",
     "  Result = case '",RootRule,"'(Input,{{line,1},{column,1}}) of\n",
     "             {AST, [], _Index} -> AST;\n",
     "             Any -> Any\n"
     "           end,\n",
     "  release_memo(), Result.\n"].

parse_grammar(InputFile) ->
    case neotoma_parse:file(InputFile) of
        {fail, Index} ->
            throw({grammar_error, {fail, Index}});
        {Parsed, Remainder, Index} ->
            io:format("WARNING: Grammar parse ended unexpectedly at ~p, generated parser may be incorrect.~nRemainder:~n~p",
                      [Index, Remainder]),
            Parsed;
        L when is_list(L) -> L;
        _ -> throw({error, {unknown, grammar, InputFile}})
    end.
create_transform(_,_,[]) -> [];
create_transform(false,_,_) ->
    "transform(_,Node,_Index) -> Node.";
create_transform(ModName,Dir,_) when is_atom(ModName) ->
    XfFile = filename:join(Dir, atom_to_list(ModName) ++ ".erl"),
    case filelib:is_regular(XfFile) of
        true -> io:format("'~s' already exists, skipping generation.~n", [XfFile]);
        false -> generate_transform_stub(XfFile, ModName)
    end,
    ["transform(Symbol,Node,Index) -> ",atom_to_list(ModName),":transform(Symbol, Node, Index)."].

generate_transform_stub(XfFile,ModName) ->
    Data = ["-module(",atom_to_list(ModName),").\n",
            "-export([transform/3]).\n\n",
            "%% Add clauses to this function to transform syntax nodes\n",
            "%% from the parser into semantic output.\n",
            "transform(Symbol, Node, _Index) when is_atom(Symbol) ->\n  Node."],
    file:write_file(XfFile, Data).

%% @doc Bootstraps the neotoma metagrammar.  Intended only for internal development!
%% @equiv file("src/neotoma_parse.peg")
bootstrap() ->
    file("src/neotoma_parse.peg").