Fix ICE on variable declarations in case branches

4 files changed, 115 insertions, 23 deletions

diff --git a/source/elna/lexer.elna b/source/elna/lexer.elna
index 64fcc22..4de178e 100644
--- a/source/elna/lexer.elna
+++ b/source/elna/lexer.elna
@@ -139,6 +139,26 @@ type
     trait,
     (* Token is complete, the last byte eaten. *)
     stop,
+    (* One-character token stop. *)
+    single,
+    (* Assignment operator stop. *)
+    assignment,
+    (* Arrow stop. *)
+    arrow,
+    (* Greater or equal stop. *)
+    greater_equal,
+    (* Less or equal stop. *)
+    less_equal,
+    (* Unequal stop. *)
+    not_equal,
+    (* Right shift stop. *)
+    shift_right,
+    (* Left shift stop. *)
+    shift_left,
+    (* Opening attribute bracket stop. *)
+    attribute_open,
+    (* Opening generic brace stop. *)
+    generic_open,
     (* Reject the token with an error. *)
     reject,
     (* Sentinel signaling end of input. *)
@@ -241,6 +261,9 @@ type
     kind: ElnaLexerTokenKind;
     position: ElnaSourceSpan
   end
+  ElnaLexerIdentifierToken = record(ElnaLexerToken)
+    identifier: StdString
+  end
 
 var
   (* Initializes the array with character classes. *)
@@ -554,17 +577,17 @@ begin
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.space, ElnaLexerState.skip);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.new_line, ElnaLexerState.skip);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.colon, ElnaLexerState.colon);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.equals, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.equals, ElnaLexerState.single);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.left_paren, ElnaLexerState.left_paren);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.right_paren, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.asterisk, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.single, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.right_paren, ElnaLexerState.single);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.asterisk, ElnaLexerState.single);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.single, ElnaLexerState.single);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.backquote, ElnaLexerState.wide);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.brace, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.square, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.brace, ElnaLexerState.single);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.square, ElnaLexerState.single);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.eof, ElnaLexerState.eof);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.dot, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.plus, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.dot, ElnaLexerState.single);
+  elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.plus, ElnaLexerState.single);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.minus, ElnaLexerState.minus);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.single_quote, ElnaLexerState.narrow);
   elna_lexer_set_transition(ElnaLexerState.start, ElnaLexerClass.double_quote, ElnaLexerState.string);
@@ -583,23 +606,23 @@ begin
   elna_lexer_set_transition(ElnaLexerState.skip, ElnaLexerClass.new_line, ElnaLexerState.skip);
 
   -- Colon state.
-  elna_lexer_set_transition(ElnaLexerState.colon, ElnaLexerClass.equals, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.colon, ElnaLexerClass.equals, ElnaLexerState.assignment);
 
   -- Left paren state.
   elna_lexer_set_transition(ElnaLexerState.left_paren, ElnaLexerClass.asterisk, ElnaLexerState.docblock);
 
   -- Minus state.
-  elna_lexer_set_transition(ElnaLexerState.minus, ElnaLexerClass.greater, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.minus, ElnaLexerClass.greater, ElnaLexerState.arrow);
   elna_lexer_set_transition(ElnaLexerState.minus, ElnaLexerClass.minus, ElnaLexerState.comment);
 
   -- Greater state.
-  elna_lexer_set_transition(ElnaLexerState.greater, ElnaLexerClass.greater, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.greater, ElnaLexerClass.equals, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.greater, ElnaLexerClass.greater, ElnaLexerState.shift_right);
+  elna_lexer_set_transition(ElnaLexerState.greater, ElnaLexerClass.equals, ElnaLexerState.greater_equal);
 
   -- Less state.
-  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.less, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.equals, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.greater, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.less, ElnaLexerState.shift_left);
+  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.equals, ElnaLexerState.less_equal);
+  elna_lexer_set_transition(ElnaLexerState.less, ElnaLexerClass.greater, ElnaLexerState.not_equal);
 
   -- Numbers prefixed by 0 followed by a character (0b, 0x).
   elna_lexer_set_transition(ElnaLexerState.leading_zero, ElnaLexerClass.zero, ElnaLexerState.integer);
@@ -666,8 +689,8 @@ begin
   -- Attribute, generic parameter or trait marker ('#').
   elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.exponent, ElnaLexerState.trait);
   elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.alpha, ElnaLexerState.trait);
-  elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.square, ElnaLexerState.stop);
-  elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.brace, ElnaLexerState.stop);
+  elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.square, ElnaLexerState.generic_open);
+  elna_lexer_set_transition(ElnaLexerState.number_sign, ElnaLexerClass.brace, ElnaLexerState.attribute_open);
 
   -- Trait identifier.
   elna_lexer_set_transition(ElnaLexerState.trait, ElnaLexerClass.exponent, ElnaLexerState.trait);
@@ -751,16 +774,32 @@ proc elna_lexer_skip_trivia(cursor: ^ElnaLexerCursor)
   until cursor^.state <> ElnaLexerState.skip
 return
 
+(*
+  Creates a token of the given kind. The created token is a simple token without
+  any payload.
+*)
 proc elna_lexer_token_create(kind: ElnaLexerTokenKind) -> ^ElnaLexerToken
 var
   result: ^ElnaLexerToken
 begin
   result := malloc#[ElnaLexerToken](#size(ElnaLexerToken));
-  memset(result, 0i8, #size(ElnaLexerToken));
   result^.kind := kind
 return result
 
--- TODO: Identifier (else) should save a payload with the identifier.
+(*
+  Frees the memory allocated for the token itself and for its internal storage.
+*)
+proc elna_lexer_token_destroy*(token: ^ElnaLexerToken)
+  if token^.kind = ElnaLexerTokenKind.identifier then
+  var
+    identifier_token := cast(token: ^ElnaLexerIdentifierToken)
+  begin
+    free(identifier_token^.identifier.ptr)
+  end;
+
+  free(token)
+return
+
 proc elna_lexer_identify_keyword(start, cursor: ^const Word8) -> ^ElnaLexerToken
 var
   result: ^ElnaLexerToken
@@ -804,7 +843,43 @@ begin
     "program": result := elna_lexer_token_create(ElnaLexerTokenKind.program_) |
     "module": result := elna_lexer_token_create(ElnaLexerTokenKind.module_)
   else
-    result := elna_lexer_token_create(ElnaLexerTokenKind.identifier)
+  var
+    identifier_token := malloc#[ElnaLexerIdentifierToken](#size(ElnaLexerToken))
+    identifier_pointer := malloc#[const Word8](identifier.length)
+  begin
+    memcpy(identifier_pointer, identifier.ptr, identifier.length);
+
+    identifier_token^.kind := ElnaLexerTokenKind.identifier;
+    identifier_token^.identifier := identifier_pointer[1 to identifier.length];;
+    result := identifier_token
+  end
+return result
+
+proc elna_lexer_stop_single(start: ^const Word8) -> ^ElnaLexerToken
+var
+  result: ^ElnaLexerToken
+begin
+  case start^ of
+    '&': result := elna_lexer_token_create(ElnaLexerTokenKind.and) |
+    '|': result := elna_lexer_token_create(ElnaLexerTokenKind.pipe) |
+    '~': result := elna_lexer_token_create(ElnaLexerTokenKind.not) |
+    '/': result := elna_lexer_token_create(ElnaLexerTokenKind.division) |
+    ')': result := elna_lexer_token_create(ElnaLexerTokenKind.right_paren) |
+    '[': result := elna_lexer_token_create(ElnaLexerTokenKind.left_square) |
+    ']': result := elna_lexer_token_create(ElnaLexerTokenKind.right_square) |
+    '{': result := elna_lexer_token_create(ElnaLexerTokenKind.left_brace) |
+    '}': result := elna_lexer_token_create(ElnaLexerTokenKind.right_brace) |
+    '=': result := elna_lexer_token_create(ElnaLexerTokenKind.equals) |
+    ';': result := elna_lexer_token_create(ElnaLexerTokenKind.semicolon) |
+    '.': result := elna_lexer_token_create(ElnaLexerTokenKind.dot) |
+    ',': result := elna_lexer_token_create(ElnaLexerTokenKind.comma) |
+    '+': result := elna_lexer_token_create(ElnaLexerTokenKind.plus) |
+    '*': result := elna_lexer_token_create(ElnaLexerTokenKind.multiplication) |
+    '%': result := elna_lexer_token_create(ElnaLexerTokenKind.remainder) |
+    '^': result := elna_lexer_token_create(ElnaLexerTokenKind.hat) |
+    '@': result := elna_lexer_token_create(ElnaLexerTokenKind.at)
+  else
+    assert(false)
   end
 return result
 
@@ -816,12 +891,21 @@ begin
   elna_lexer_skip_trivia(cursor);
 
   case cursor^.state of
-    ElnaLexerState.identifier: result := elna_lexer_identify_keyword(cursor^.start, cursor^.cursor)
+    ElnaLexerState.identifier: result := elna_lexer_identify_keyword(cursor^.start, cursor^.cursor) |
+    ElnaLexerState.single: result := elna_lexer_stop_single(cursor^.start) |
+    ElnaLexerState.attribute_open: result := elna_lexer_token_create(ElnaLexerTokenKind.attribute_open) |
+    ElnaLexerState.generic_open: result := elna_lexer_token_create(ElnaLexerTokenKind.generic_open) |
+    ElnaLexerState.shift_left: result := elna_lexer_token_create(ElnaLexerTokenKind.shift_left) |
+    ElnaLexerState.shift_right: result := elna_lexer_token_create(ElnaLexerTokenKind.shift_right) |
+    ElnaLexerState.greater_equal: result := elna_lexer_token_create(ElnaLexerTokenKind.greater_equal) |
+    ElnaLexerState.less_equal: result := elna_lexer_token_create(ElnaLexerTokenKind.less_equal) |
+    ElnaLexerState.not_equal: result := elna_lexer_token_create(ElnaLexerTokenKind.not_equal) |
+    ElnaLexerState.assignment: result := elna_lexer_token_create(ElnaLexerTokenKind.assignment) |
+    ElnaLexerState.arrow: result := elna_lexer_token_create(ElnaLexerTokenKind.arrow)
   else
     result := malloc#[ElnaLexerToken](#size(ElnaLexerToken));
     memset(result, 0i8, #size(ElnaLexerToken))
   end
-
 return result
 
 end.
diff --git a/source/main.elna b/source/main.elna
index ee4079e..fb58766 100644
--- a/source/main.elna
+++ b/source/main.elna
@@ -23,6 +23,8 @@ begin
     printf("Token: %.*s, length: %lu, final state: %i, position: %lu, kind: %i\n\0".ptr,
         global_cursor.cursor - global_cursor.start, global_cursor.start,
         global_cursor.cursor - global_cursor.start, global_cursor.state,
-        global_cursor.position, token^.kind)
+        global_cursor.position, token^.kind);
+
+    elna_lexer_token_destroy(token)
   until global_cursor.state = ElnaLexerState.eof
 end.
diff --git a/source/sys/posix/stdlib.elna b/source/sys/posix/stdlib.elna
index 0016d9c..66b383a 100644
--- a/source/sys/posix/stdlib.elna
+++ b/source/sys/posix/stdlib.elna
@@ -6,4 +6,7 @@ module
 proc malloc#[T]*(size: Word) -> ^T
 extern
 
+proc free#[T]*(ptr: ^T)
+extern
+
 end.
diff --git a/source/sys/posix/string.elna b/source/sys/posix/string.elna
index 9994679..b699606 100644
--- a/source/sys/posix/string.elna
+++ b/source/sys/posix/string.elna
@@ -6,4 +6,7 @@ module
 proc memset#[T]*(b: ^T; c: Int8; len: Word) -> ^T
 extern
 
+proc memcpy#[T]*(dst: ^T; src: ^const T; n: Word) -> ^T
+extern
+
 end.