Handle varaint constants and types.

README.md

@@ -1,9 +1,92 @@
-aebytecode
+# aebytecode
 =====
 
-An OTP library
+An library and stand alone assembler for aeternity bytecode.
 
-Build
+This version supports Aevm bytecode and Fate bytecode.
+
+## Build
 -----
 
-    $ rebar3 compile
+    $ make
+
+## Fate Code
+---------
+
+Fate code exists in 3 formats:
+
+1. Fate byte code. This format is under consensus.
+2. Fate assembler. This is a text represenation of fate code.
+                     This is not under consensus and other
+                     implemenation and toolchains could have
+                     their own format.
+3. Internal. This is an Erlang representation of fate code
+               Used by this particular engin implementation.
+
+This library handles all tree representations.
+The byte code format is described in a separate document.
+The internal format is described in a separate document.
+The text representation is described below.
+
+### Fate Assembler Code
+-------------------
+
+Assembler code can be read from a file.
+The assembler has the following format:
+
+     Comments start with 2 semicolons and runs till end of line
+        `;; This is a comment`
+        Opcode mnemonics start with an upper case letter.
+       `DUP`
+     Identifiers start with a lower case letter
+        `an_identifier`
+     References to function arguments start with arg followed by an integer
+         `arg0`
+     References to variables/registers start with var followed by an integer
+         `var0`
+     References to stack postions is either a (for stack 0)
+      or start with stack followed by an integer
+         `stack1`
+         `a`
+
+     Immediate values can be of 9 types:
+     1a. Integers as decimals: {Digits} or -{Digits}
+         `42`
+         `-2374683271468723648732648736498712634876147`
+     1b. Integers as Hexadecimals::  0x{Hexdigits}
+         `0x0deadbeef0`
+      2. addresses, a base58 encoded string starting with #{base58char}
+         followed by up to 64 hex chars
+         `#nv5B93FPzRHrGNmMdTDfGdd5xGZvep3MVSpJqzcQmMp59bBCv`
+      3. Boolean  true or false
+         `true`
+         `false`
+      4. Strings  "{Characters}"
+         `"Hello"`
+      5. Map  { Key => Value }
+         `{}`
+         `{ 1 => { "foo" => true, "bar" => false}`
+      6. Lists [ Elements ]
+         `[]`
+         `[1, 2]`
+      7. Bit field < Bits > or !< Bits >
+         `<000>`
+         `<1010 1010>`
+         `<>`
+         `!<>`
+      8. Tuples ( Elements )
+         `()`
+         `(1, "foo")`
+      9. Varaiants: (| Size | Tag | ( Elements ) |)
+         `(| 42 | 12 | ( "foo", 12) |)`
+
+      Where Digits: [0123456789]
+            Hexdigits:  [0123456789abcdef]
+            base58char:  [123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz]
+            Characters any printable ascii character 0..255 (except " no quoting yet)
+            Key: any value except for a map
+            Bits: 01 or space
+            Elements: Nothing or Value , Elements
+            Size: Digits
+            Tag: Digits
+

src/aeb_fate_asm.erl

@@ -34,34 +34,47 @@
 %%%          stack1
 %%%          a
 %%%
-%%%      Immediates can be of 9 types:
+%%%      Immediate values can be of 9 types:
-%%%       1. Integers
+%%%      1a. Integers as decimals: {Digits} or -{Digits}
 %%%          42
 %%%          -2374683271468723648732648736498712634876147
-%%%       2. Hexadecimal integers starting with 0x
+%%%      1b. Integers as Hexadecimals::  0x{Hexdigits}
 %%%          0x0deadbeef0
-%%%       3. addresses, a 256-bit hash strings starting with #
+%%%       2. addresses, a base58 encoded string starting with #{base58char}
 %%%          followed by up to 64 hex chars
 %%%          #00000deadbeef
-%%%       4. Boolean
+%%%       3. Boolean  true or false
 %%%          true
 %%%          false
-%%%       5. Strings
+%%%       4. Strings  "{Characters}"
 %%%          "Hello"
-%%%       6. Map
+%%%       5. Map  { Key => Value }
 %%%          {}
 %%%          { 1 => { "foo" => true, "bar" => false}
-%%%       7. Lists
+%%%       6. Lists [ Elements ]
 %%%          []
 %%%          [1, 2]
-%%%       8. Bit field
+%%%       7. Bit field < Bits > or !< Bits >
 %%%          <000>
 %%%          <1010 1010>
 %%%          <>
 %%%          !<>
-%%%       9. Tuples
+%%%       8. Tuples ( Elements )
 %%%          ()
 %%%          (1, "foo")
+%%%       9. Varaiants: (| Size | Tag | ( Elements ) |)
+%%%          (| 42 | 12 | ( "foo", 12) |)
+%%%
+%%%       Where Digits: [0123456789]
+%%%             Hexdigits:  [0123456789abcdef]
+%%%             base58char:  [123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz]
+%%%             Characters any printable ascii character 0..255 (except " no quoting yet)
+%%%             Key: any value except for a map
+%%%             Bits: 01 or space
+%%%             Elements: Nothing or Value , Elements
+%%%             Size: Digits
+%%%             Tag: Digits
+%%%
 %%% @end
 %%% Created : 21 Dec 2017
 %%%-------------------------------------------------------------------
@@ -100,7 +113,7 @@ parse_function_call([{id,_,Name}, {'(',_}| Rest]) ->
 
 to_args([{')', _}]) -> {[], []};
 to_args(Tokens) ->
-    case to_data(Tokens) of
+    case parse_value(Tokens) of
         {Arg, [{',', _} |  Rest]} ->
             {More, Rest2} = to_args(Rest),
             {[Arg|More], Rest2};
@@ -108,13 +121,6 @@ to_args(Tokens) ->
             {[Arg], Rest}
     end.
 
-to_data([{int,_line, Int}|Rest]) ->
-    {Int, Rest};
-to_data([{boolean,_line, Bool}|Rest]) ->
-    {Bool, Rest};
-to_data([{hash,_line, Hash}|Rest]) ->
-    {Hash, Rest}.
-
 pp(FateCode) ->
     Listing = to_asm(FateCode),
     io_lib:format("~ts~n",[Listing]).
@@ -572,7 +578,16 @@ deserialize_type(<<6, Rest/binary>>) ->
     {V, Rest3} = deserialize_type(Rest2),
     {{map, K, V}, Rest3};
 deserialize_type(<<7, Rest/binary>>) ->
-    {string, Rest}.
+    {string, Rest};
+deserialize_type(<<8, Size, Rest/binary>>) ->
+    {Variants, Rest2} = deserialize_variants(Size, Rest, []),
+    {{variant, Size, Variants}, Rest2}.
+
+deserialize_variants(0, Rest, Variants) ->
+    {lists:reverse(Variants), Rest};
+deserialize_variants(N, Rest, Variants) ->
+    {T, Rest2} = deserialize_type(Rest),
+    deserialize_variants(N-1, Rest2, [T|Variants]).
 
 
 
@@ -616,8 +631,14 @@ to_bytecode([{int,_line, Int}|Rest], Address, Env, Code, Opts) ->
     to_bytecode(Rest, Address, Env, [{immediate, Int}|Code], Opts);
 to_bytecode([{boolean,_line, Bool}|Rest], Address, Env, Code, Opts) ->
     to_bytecode(Rest, Address, Env, [{immediate, Bool}|Code], Opts);
-to_bytecode([{hash,_line, Hash}|Rest], Address, Env, Code, Opts) ->
+to_bytecode([{string,_line, String}|Rest], Address, Env, Code, Opts) ->
-    to_bytecode(Rest, Address, Env, [{immediate, Hash}|Code], Opts);
+    to_bytecode(Rest, Address, Env,
+                [{immediate, aeb_fate_data:make_string(String)}|Code],
+                Opts);
+to_bytecode([{address,_line, Value}|Rest], Address, Env, Code, Opts) ->
+    to_bytecode(Rest, Address, Env,
+                [{immediate, aeb_fate_data:make_address(Value)}|Code],
+                Opts);
 to_bytecode([{id,_line, ID}|Rest], Address, Env, Code, Opts) ->
     {Hash, Env2} = insert_symbol(ID, Env),
     to_bytecode(Rest, Address, Env2, [{immediate, Hash}|Code], Opts);
@@ -631,6 +652,10 @@ to_bytecode([{'(',_line}|Rest], Address, Env, Code, Opts) ->
     {Elements, Rest2} = parse_tuple(Rest),
     Tuple = aeb_fate_data:make_tuple(list_to_tuple(Elements)),
     to_bytecode(Rest2, Address, Env, [{immediate, Tuple}|Code], Opts);
+to_bytecode([{start_variant,_line}|_] = Tokens, Address, Env, Code, Opts) ->
+    {Size, Tag, Values, Rest} = parse_variant(Tokens),
+    Variant = aeb_fate_data:make_variant(Size, Tag, Values),
+    to_bytecode(Rest, Address, Env, [{immediate, Variant}|Code], Opts);
 to_bytecode([{bits,_line, Bits}|Rest], Address, Env, Code, Opts) ->
     to_bytecode(Rest, Address, Env,
                 [{immediate, aeb_fate_data:make_bits(Bits)}|Code], Opts);
@@ -689,13 +714,35 @@ parse_tuple(Tokens) ->
     end.
 
 
+parse_variant([{start_variant,_line}
+              , {int,_line, Size}
+              , {'|',_}
+              , {int,_line, Tag}
+              , {'|',_}
+              , {'(',_}
+              | Rest]) when (Size > 0), (Tag < Size) ->
+    {Elements , [{end_variant, _} | Rest2]} = parse_tuple(Rest),
+    {Size, Tag, list_to_tuple(Elements), Rest2}.
+
+
 parse_value([{int,_line, Int} | Rest]) -> {Int, Rest};
 parse_value([{boolean,_line, Bool} | Rest]) -> {Bool, Rest};
 parse_value([{hash,_line, Hash} | Rest]) -> {Hash, Rest};
 parse_value([{'{',_line} | Rest]) -> parse_map(Rest);
 parse_value([{'[',_line} | Rest]) -> parse_list(Rest);
-parse_value([{'(',_line} | Rest]) -> parse_tuple(Rest).
+parse_value([{'(',_line} | Rest]) ->
-
+    {T, Rest2} = parse_tuple(Rest),
+    {aeb_fate_data:make_tuple(list_to_tuple(T)), Rest2};
+parse_value([{bits,_line, Bits} | Rest]) ->
+    {aeb_fate_data:make_bits(Bits), Rest};
+parse_value([{start_variant,_line}|_] = Tokens) ->
+    {Size, Tag, Values, Rest} = parse_variant(Tokens),
+    Variant = aeb_fate_data:make_variant(Size, Tag, Values),
+    {Variant, Rest};
+parse_value([{string,_line, String} | Rest]) ->
+    {aeb_fate_data:make_string(String), Rest};
+parse_value([{address,_line, Address} | Rest]) ->
+    {aeb_fate_data:make_address(Address), Rest}.
 
 to_fun_def([{id, _, Name}, {'(', _} | Rest]) ->
     {ArgsType, [{'to', _} | Rest2]} = to_arg_types(Rest),
@@ -732,7 +779,22 @@ to_type([{'{', _}, {id, _, "map"}, {',', _} | Rest]) ->
     %% TODO: Error handling
     {KeyType, [{',', _}| Rest2]} = to_type(Rest),
     {ValueType, [{'}', _}| Rest3]} = to_type(Rest2),
-    {{map, KeyType, ValueType}, Rest3}.
+    {{map, KeyType, ValueType}, Rest3};
+to_type([{'{', _}
+        , {id, _, "variant"}
+        , {',', _}
+        , {int, _, Size}
+        , {',', _}
+        , {'[', _}
+         | Rest]) when Size > 0
+                       , Size < 256 ->
+    {ElementTypes, [{'}', _}| Rest2]} = to_list_of_types(Rest),
+    %% TODO: Error handling
+    if Size =:= length(ElementTypes) ->
+            {{variant, Size, ElementTypes}, Rest2}
+    end.
+
+
 
 to_list_of_types([{']', _} | Rest]) -> {[], Rest};
 to_list_of_types(Tokens) ->
@@ -756,8 +818,12 @@ serialize_type({tuple, Ts}) ->
 serialize_type(address) -> [4];
 serialize_type(bits) -> [5];
 serialize_type({map, K, V}) -> [6 | serialize_type(K) ++ serialize_type(V)];
-serialize_type(string) -> [7].
+serialize_type(string) -> [7];
-
+serialize_type({variant, Size, ListOfVariants})
+  when Size > 0
+       , Size < 256
+       , Size =:= length(ListOfVariants) ->
+    [8, Size | [serialize_type(T) || T <- ListOfVariants]].
 
 
 %% -------------------------------------------------------------------

src/aeb_fate_asm_scan.template

@@ -12,9 +12,10 @@ DIGIT    = [0-9]
 HEXDIGIT = [0-9a-fA-F]
 LOWER    = [a-z_]
 UPPER    = [A-Z]
+BASE58   = [123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz]
 INT      = {DIGIT}+
 HEX      = 0x{HEXDIGIT}+
-HASH     = #{HEXDIGIT}+
+HASH     = #{BASE58}+
 WS       = [\000-\s]
 ID       = {LOWER}[a-zA-Z0-9_]*
 STRING   = "[^"]*"
@@ -42,9 +43,9 @@ FUNCTION       : {token, {function, TokenLine, 'FUNCTION' }}.
 -{INT} :
  {token, {int, TokenLine, parse_int(TokenChars)}}.
 {HASH} :
- {token, {hash, TokenLine, parse_hash(TokenChars)}}.
+ {token, {address, TokenLine, parse_hash(TokenChars)}}.
 {STRING} :
- {token, {hash, TokenLine, list_to_binary(TokenChars)}}.
+ {token, {string, TokenLine, list_to_binary(TokenChars)}}.
 {BITS} :
  {token, {bits, TokenLine, bits(TokenChars)}}.
 
@@ -54,6 +55,8 @@ FUNCTION       : {token, {function, TokenLine, 'FUNCTION' }}.
 \:    : {token, {to, TokenLine}}.
 
 \=\>  : {token, {arrow, TokenLine}}.
+\(\|  : {token, {start_varaint, TokenLine}}.
+\|\)  : {token, {end_varaint, TokenLine}}.
 
 ,   : {token, {',', TokenLine}}.
 \(  : {token, {'(', TokenLine}}.
@@ -62,6 +65,7 @@ FUNCTION       : {token, {function, TokenLine, 'FUNCTION' }}.
 \]  : {token, {']', TokenLine}}.
 \{  : {token, {'{', TokenLine}}.
 \}  : {token, {'}', TokenLine}}.
+\|  : {token, {'|', TokenLine}}.
 
 ;;.* :
  {token, {comment, TokenLine, drop_prefix($;, TokenChars)}}.
@@ -98,8 +102,7 @@ parse_acc("a" ++ N) -> list_to_integer(N).
 
 
 parse_hash("#" ++ Chars) ->
-    N = list_to_integer(Chars, 16),
+    base58_to_address(Chars).
-    <<N:256>>.
 
 scan(S) ->
     string(S).
@@ -117,3 +120,23 @@ bits([$> |_Rest], Acc) -> Acc;
 bits([$0 | Rest], Acc) -> bits(Rest, Acc bsl 1);
 bits([$1 | Rest], Acc) -> bits(Rest, (Acc bsl 1) bor 1);
 bits([$  | Rest], Acc) -> bits(Rest, Acc).
+
+char_to_base58(C) ->
+    binary:at(<<0,1,2,3,4,5,6,7,8,0,0,0,0,0,0,0,9,10,11,12,13,14,15,16,0,17,
+                18,19,20,21,0,22,23,24,25,26,27,28,29,30,31,32,0,0,0,0,0,0,
+                33,34,35,36,37,38,39,40,41,42,43,0,44,45,46,47,48,49,50,51,
+                52,53,54,55,56,57>>,  C-$1).
+
+base58_to_integer(C, []) -> C;
+base58_to_integer(C, [X | Xs]) ->
+	base58_to_integer(C * 58 + char_to_base58(X), Xs).
+
+base58_to_integer([]) -> error;
+base58_to_integer([Char]) -> char_to_base58(Char);
+base58_to_integer([Char | Str]) ->
+	base58_to_integer(char_to_base58(Char), Str).
+
+base58_to_address(Base58) ->
+    I = base58_to_integer(Base58),
+    Bin = <<I:256>>,
+    Bin.

src/aeb_fate_data.erl

@@ -144,7 +144,7 @@ format(?FATE_BITS(B)) when B < 0 ->
 format(?FATE_VARIANT(Size, Tag, T)) ->
     ["(| ",
       lists:join("| ", [integer_to_list(Size), integer_to_list(Tag) |
-                        [format(E) || E <- erlang:tuple_to_list(T)]]),
+                        [format(make_tuple(T))]]),
      " |)"];
 format(M) when ?IS_FATE_MAP(M) ->
     ["{ ", format_kvs(maps:to_list(?FATE_MAP_VALUE(M))), " }"];
@@ -153,18 +153,13 @@ format(V) -> exit({not_a_fate_type, V}).
 
 format_bits(0, Acc) -> Acc;
 format_bits(N, Acc) ->
-    case N band 1 of
+    Bit = $0 + (N band 1),
-        1 -> format_bits(N bsr 1, [$1|Acc]);
+    format_bits(N bsr 1, [Bit|Acc]).
-        0 -> format_bits(N bsr 1, [$0|Acc])
-    end.
 
 format_nbits(0, Acc) -> Acc;
 format_nbits(N, Acc) ->
-    case N band 1 of
+    Bit = $1 - (N band 1),
-        1 -> format_nbits(N bsr 1, [$0|Acc]);
+    format_nbits(N bsr 1, [Bit|Acc]).
-        0 -> format_nbits(N bsr 1, [$1|Acc])
-    end.
-
 
 format_list(List) ->
     ["[ ", lists:join(", ", [format(E) || E <- List]), " ]"].

test/asm_code/immediates.fate

@@ -65,4 +65,9 @@ FUNCTION tuple() : {tuple, [integer, boolean, string, {tuple, [integer, integer]
          RETURNR (42, true, "FooBar", (1, 2))
 
 
+FUNCTION address() : address
+         RETURNR #deadbeef
 
+;; Option(integer) = NONE | SOME(integer)
+FUNCTION varaint() : {variant, 2, [{tuple, []}, {tuple, [integer]}]}
+         RETURNR #deadbeef