Pretty printing, parsing structs

build.bat

@@ -2,5 +2,5 @@
 
 rem clang generate.c  -fdiagnostics-absolute-paths -std=c99 -g -o generate.exe -Wl,user32.lib
 rem generate.exe
-clang main.c -Wall -Wno-unused-function -fdiagnostics-absolute-paths -std=c99 -g -o main.exe -Wl,user32.lib
+clang main.c -Wall -Wno-unused-function -fdiagnostics-absolute-paths -std=c17 -g -o main.exe -Wl,user32.lib
 rem cl main.c -std:c17

main.c

@@ -13,6 +13,7 @@
 
 #include "new_lex.c"
 #include "new_ast.c"
+#include "new_print.c"
 #include "new_parse.c"
 
 int main(){

new_ast.c

@@ -164,6 +164,8 @@ typedef enum Decl_Kind{
   DECL_None,
   DECL_Struct,
   DECL_Union,
+  DECL_SubStruct,
+  DECL_SubUnion,
   DECL_Enum,
   DECL_Variable,
   DECL_Typedef,
@@ -497,6 +499,54 @@ function void
 typespec_function_push(Typespec *func, Typespec *arg){
   SLLQueuePush(func->func.first, func->func.last, arg);
 }
+//-----------------------------------------------------------------------------
+// Notes
+//-----------------------------------------------------------------------------
+function void
+decl_pass_notes(Decl *a, Note *b){
+  a->first_note = b->first;
+  a->last_note = b->last;
+}
+
+function Note *
+note_push_new(Arena *p, Note *parent, Token *pos, Intern_String name, Expr *expr){
+  Note *result = arena_push_struct(p, Note);
+  result->pos = pos;
+  result->name = name;
+  result->expr = expr;
+  SLLQueuePush(parent->first, parent->last, result);
+  return result;
+}
+
+function Note *
+find_note(Note *first, String string){
+  for(Note *n = first; n; n=n->next){
+    if(string_compare(string, n->name.s)){
+      return n;
+    }
+  }
+  return 0;
+}
+
+function String
+find_string_note(Note *first, String string, String default_string){
+  Note *note = find_note(first, string);
+  if(note){
+    return note->expr->token->intern_val.s;
+  }
+  return default_string;
+}
+
+function Note *
+decl_find_note(Decl *decl, String string){
+  return find_note(decl->first_note, string);
+}
+
+function String
+decl_find_string_note(Decl *decl, String string, String default_string){
+  return find_string_note(decl->first_note, string, default_string);
+}
+
 
 //-----------------------------------------------------------------------------
 // Declarations
@@ -512,7 +562,7 @@ decl_new(Arena *p, Decl_Kind kind, Token *pos, Intern_String name){
 
 function Decl *
 decl_struct(Arena *p, Decl_Kind kind, Token *pos, Intern_String name){
-  assert(kind == DECL_Struct || kind == DECL_Union);
+  assert(kind == DECL_Struct || kind == DECL_Union || kind == DECL_SubUnion || kind == DECL_SubStruct);
   Decl *result = decl_new(p, kind, pos, name);
   return result;
 }
@@ -570,7 +620,7 @@ decl_enum_push(Arena *p, Decl *parent, Token *pos, Intern_String name, Expr *exp
 
 function void
 decl_struct_push(Decl *parent, Decl *child){
-  assert(parent->kind == DECL_Struct || parent->kind == DECL_Union);
+  assert(parent->kind == DECL_Struct || parent->kind == DECL_Union || parent->kind == DECL_SubUnion || parent->kind == DECL_SubStruct);
   SLLQueuePush(parent->struct_decl.first, parent->struct_decl.last, child);
 }
 

new_parse.c

@@ -78,6 +78,15 @@ token_peek_is(Parser *p, Token_Kind kind, S64 i){
   return 0;
 }
 
+function Token *
+token_peek_is_keyword(Parser *p, Intern_String keyword, S64 i){
+  assert(lex_is_keyword(keyword));
+  Token *result = token_peek(p, i);
+  if(result->kind == TK_Keyword && intern_compare(result->intern_val, keyword))
+    return result;
+  return 0;
+}
+
 function Token *
 token_is(Parser *p, Token_Kind kind){
   Token *result = token_get(p);
@@ -536,35 +545,155 @@ name::enum{}
 name::typedef = name2;
 name::const   = 4254;
 
+// Thing::struct{
+//   var1, var2: U32;
+//   union {
+//     thing: *S64;
+  //   }
+//   named: union {
+//     data: [24]U32;
+//   }
+// }
+
 */
-function Decl *
+function void 
-parse_enum(Parser *p, Token *name){
+parse_note_list(Parser *p, Note *parent) {
-  Typespec *typespec = 0;
+  if(token_match(p, TK_OpenParen)) {
+    if(token_match(p, TK_CloseParen)){
+      return;
+    }
+    do {
+      Token *name = token_expect(p, TK_Identifier);
+      Note *current = note_push_new(p->arena, parent, name, name->intern_val, 0);
+      parse_note_list(p, current);
+      if(token_match(p, TK_Assign)) {
+        current->expr = parse_expr(p);
+      }
+    } while(token_match(p, TK_Comma));
+    token_expect(p, TK_CloseParen);
+  }
+}
+
+function void 
+parse__notes(Parser *p, Note *result) {
+  while(token_match(p, TK_At)) {
+    Token *name = token_expect(p, TK_Identifier);
+    Note *current = note_push_new(p->arena, result, name, name->intern_val, 0);
+    parse_note_list(p, current);
+    if(token_match(p, TK_Assign)) {
+      current->expr = parse_expr(p);
+    }
+    token_match(p, TK_Semicolon);
+  }
+}
+
+function Note
+parse_notes(Parser *p){
+  Note result = {0};
+  parse__notes(p, &result);
+  return result;
+}
+
+function Expr *
+parse_assign_expr(Parser *p){
+  if(token_match(p, TK_Assign))
+    return parse_expr(p);
+  return 0;
+}
+
+function Typespec *
+parse_optional_type(Parser *p){
+  Typespec *result = 0;
   if(token_match(p, TK_Colon)){
-    typespec = parse_typespec(p);
+    result = parse_typespec(p);
   }
   else{
-    typespec = typespec_name(p->arena, token_get(p), intern_int);
+    result = typespec_name(p->arena, token_get(p), intern_int);
   }
-  
+  return result;
+}
+
+function Decl *
+parse_enum(Parser *p, Token *name){
+  Typespec *typespec = parse_optional_type(p);
   Decl *result = decl_enum(p->arena, name, name->intern_val, typespec);
   token_expect(p, TK_OpenBrace);
   do{
+    Note notes = parse_notes(p);
     Token *val = token_expect(p, TK_Identifier);
-    Expr *expr = 0;
+    Expr *expr = parse_assign_expr(p);
-    if(token_match(p, TK_Assign)){
+    decl_enum_push(p->arena, result, val, val->intern_val, expr, &notes);
-      expr = parse_expr(p);
+  } while(token_match(p, TK_Comma));
-    }
-    decl_enum_push(p->arena, result, val, val->intern_val, expr, 0);
-  } while(token_is(p, TK_Comma));
   token_expect(p, TK_CloseBrace);
   return result;
 }
 
+function void
+parse_var_decl(Parser *p, Decl *parent, Token *name){
+  Token *name_stack[64];
+  S64 name_stack_len = 0;
+  name_stack[name_stack_len++] = name;
+  while(token_match(p, TK_Comma))
+    name_stack[name_stack_len++] = token_expect(p, TK_Identifier);
+  token_expect(p, TK_Colon);
+  Typespec *typespec = parse_typespec(p);
+  token_expect(p, TK_Semicolon);
+  for(S64 i = 0; i < name_stack_len; i++){
+    Decl *decl = decl_variable(p->arena, name_stack[i], name_stack[i]->intern_val, typespec, 0);
+    decl_struct_push(parent, decl);
+  }
+}
+
+function Decl *
+parse_struct(Parser *p, Token *name, Decl_Kind struct_kind){
+  Intern_String intern_name = name ? name->intern_val : (Intern_String){0};
+  Token *token = name ? name : token_get(p);
+  Decl *result = decl_struct(p->arena, struct_kind, token, intern_name);
+  token_expect(p, TK_OpenBrace);
+  do {
+    Token *token = token_get(p);
+    if(token_match_keyword(p, keyword_union)){
+      Decl *decl = parse_struct(p, 0, DECL_SubUnion);
+      decl_struct_push(result, decl);
+    }
+    else if(token_match_keyword(p, keyword_struct)){
+      Decl *decl = parse_struct(p, 0, DECL_SubStruct);
+      decl_struct_push(result, decl);
+    }
+    else if(token_match(p, TK_Identifier)){
+      if(token_is(p, TK_Colon)){
+        if(token_peek_is_keyword(p, keyword_union, 1)){
+          token_next(p);
+          token_next(p);
+          Decl *decl = parse_struct(p, token, DECL_SubUnion);
+          decl_struct_push(result, decl);
+        }
+        else if(token_peek_is_keyword(p, keyword_struct, 1)){
+          token_next(p);
+          token_next(p);
+          Decl *decl = parse_struct(p, token, DECL_SubStruct);
+          decl_struct_push(result, decl);
+        }
+        else{
+          parse_var_decl(p, result, token);
+        }
+      }
+      else{
+        parse_var_decl(p, result, token);
+      }
+    }
+    else{
+      parser_push_error(p, token, "Failed to parse struct, unexpected token of kind '%s'", token_kind_string[token->kind]);
+    }
+  } while(!token_match(p, TK_CloseBrace));
+  return result;
+}
+
 function Decl *
 parse_decl(Parser *p){
   Decl *result = 0;
   Token *name = 0;
+  Note note = parse_notes(p);
   if((name = token_match(p, TK_Identifier))){
     if(token_match(p, TK_DoubleColon)){
       Token *token = 0;
@@ -572,10 +701,10 @@ parse_decl(Parser *p){
         result = parse_enum(p, name);
       }
       else if((token = token_match_keyword(p, keyword_union))){
-        // Union
+        result = parse_struct(p, name, DECL_Union);
       }
       else if((token = token_match_keyword(p, keyword_struct))){
-        // Struct
+        result = parse_struct(p, name, DECL_Struct);
       }
       else if((token = token_match_keyword(p, keyword_const))){
         // Const value
@@ -591,183 +720,16 @@ parse_decl(Parser *p){
       parser_push_error(p, token_get(p), "Expected token of kind '::'");
     }
   }
+  if(result){
+    decl_pass_notes(result, &note);
+  }
+  
   return result;
 }
 
 //-----------------------------------------------------------------------------
 // Test code
 //-----------------------------------------------------------------------------
-function void expr_print(Expr *expr);
-function B32 typespec_print(Typespec *spec);
-function void
-token_print(Token *token){
-  printf("%.*s", (S32)token->len, token->str);
-}
-
-function void
-expr_compound_print(Expr_Compound_Field *field){
-  switch(field->kind){
-    case COMPOUND_Default: {
-      expr_print(field->init);
-    }break;
-    case COMPOUND_Named: {
-      printf("[%s] = ", field->name.s.str);
-      expr_print(field->init);
-    }break;
-    case COMPOUND_Index: {
-      printf("[");
-      expr_print(field->index);
-      printf("] = ");
-      expr_print(field->init);
-    }break;
-    default: invalid_codepath;
-  }
-}
-
-function void
-expr_print(Expr *expr){
-  switch(expr->kind) {
-    case EK_Int:case EK_String:case EK_Identifier: {
-      token_print(expr->token);
-    } break;
-    
-    case EK_SizeExpr:{
-      printf("size_expr(");
-      expr_print(expr->size_expr.expr);
-      printf(")");
-    }break;
-    
-    case EK_Compound:{
-      if(expr->compound.typespec){
-        printf("(");
-        typespec_print(expr->compound.typespec);
-        printf(")");
-      }
-      printf("{");
-      for(Expr_Compound_Field *n = expr->compound.first; n; n=n->next){
-        expr_compound_print(n);
-        if(n!=expr->compound.last) printf(",");
-      }
-      printf("}");
-    } break;
-    
-    case EK_SizeType:{
-      printf("size_type(");
-      printf(")");
-    }break;
-    
-    case EK_Paren:{
-      printf("(");
-      expr_print(expr->paren.expr);
-      printf(")");
-    } break;
-    
-    case EK_Field:{
-      expr_print(expr->field.expr);
-      printf(".%s", expr->field.name.s.str);
-    } break;
-    
-    case EK_Binary:{
-      printf("(");
-      expr_print(expr->binary.left);
-      token_print(expr->token);
-      expr_print(expr->binary.right);
-      printf(")");
-    } break;
-    
-    case EK_PostfixUnary:{
-      printf("(");
-      expr_print(expr->unary.expr);
-      token_print(expr->token);
-      printf(")");
-    } break;
-    
-    case EK_Unary:{
-      printf("(");
-      token_print(expr->token);
-      expr_print(expr->unary.expr);
-      printf(")");
-    } break;
-    
-    case EK_Ternary:{
-      printf("(");
-      expr_print(expr->ternary.cond);
-      printf("?");
-      expr_print(expr->ternary.on_true);
-      printf(":");
-      expr_print(expr->ternary.on_false);
-      printf(")");
-    } break;
-    
-    case EK_Cast:{
-      printf("(");
-      printf("(");
-      typespec_print(expr->cast.typespec);
-      printf(")");
-      expr_print(expr->cast.expr);
-      printf(")");
-    } break;
-    
-    case EK_Index:{
-      expr_print(expr->index.atom);
-      printf("[");
-      expr_print(expr->index.index);
-      printf("]");
-    }break;
-    
-    case EK_Call:{
-      expr_print(expr->call.atom);
-      printf("(");
-      for(Expr_Compound_Field *n = expr->call.first; n; n=n->next){
-        expr_compound_print(n);
-        if(n!=expr->call.last) printf(",");
-      }
-      printf(")");
-    }break;
-    default: {invalid_codepath;} break;
-  }
-}
-
-function B32
-typespec_print(Typespec *spec){
-  switch(spec->kind) {
-    case TS_Name: {
-      printf("%s", spec->name.s.str);
-    } break;
-    
-    case TS_NamedArgument: {
-      printf("%s: ", spec->named.name.s.str);
-      typespec_print(spec->named.base);
-    }break;
-    
-    case TS_Pointer: {
-      typespec_print(spec->base);
-      printf("*");
-    } break;
-    
-    case TS_Array: {
-      typespec_print(spec->arr.base);
-      printf("[");
-      expr_print(spec->arr.size);
-      printf("]");
-    } break;
-    
-    case TS_Function: {
-      printf("(");
-      for(Typespec *n = spec->func.first; n; n=n->next){
-        typespec_print(n);
-        if(n!=spec->func.last) 
-          printf(", ");
-      }
-      
-      printf(")");
-      typespec_print(spec->func.ret);
-    } break;
-    default: {invalid_codepath;} break;
-  }
-  return true;
-}
-
 
 function S64
 eval_expr(Expr *expr){
@@ -904,6 +866,19 @@ parse_test_expr(){
 
 function void
 parse_test_decls(){
+  Arena *scratch = arena_begin_scratch();
+  Parser p = parser_make(scratch);
+  
+  String decls[] = {
+    lit("Thing::enum: U64{ Thing_1 = 1<<1, Thing_2 = 2 }"),
+    lit("Struct::struct{ thing, thing2: *S64; struct { thing:U32; thing2: [64]Vec2; }  inner: struct {i:U32;} }"),
+  };
+  for(SizeU i = 0; i < buff_cap(decls); i++){
+    parser_restream(&p, decls[i], lit("Decl_Test"));
+    Decl *decl = parse_decl(&p);
+    assert(decl);
+    decl_print(decl);
+  }
   
 }
 

new_print.c

@@ -0,0 +1,261 @@
+function void expr_print(Expr *expr);
+function B32 typespec_print(Typespec *spec);
+
+global S64 indent;
+
+#define println(...) do{ printf("\n"); print_indent(); printf(__VA_ARGS__); }while(0)
+
+function void
+print_indent(){
+  for(S64 i = 0; i < indent*2; i++)
+    printf(" ");
+}
+
+function void
+token_print(Token *token){
+  printf("%.*s", (S32)token->len, token->str);
+}
+
+function void
+expr_compound_print(Expr_Compound_Field *field){
+  switch(field->kind){
+    case COMPOUND_Default: {
+      expr_print(field->init);
+    }break;
+    case COMPOUND_Named: {
+      printf("[%s] = ", field->name.s.str);
+      expr_print(field->init);
+    }break;
+    case COMPOUND_Index: {
+      printf("[");
+      expr_print(field->index);
+      printf("] = ");
+      expr_print(field->init);
+    }break;
+    default: invalid_codepath;
+  }
+}
+
+function void
+expr_print(Expr *expr){
+  switch(expr->kind) {
+    case EK_Int:case EK_String:case EK_Identifier: {
+      token_print(expr->token);
+    } break;
+    
+    case EK_SizeExpr:{
+      printf("size_expr(");
+      expr_print(expr->size_expr.expr);
+      printf(")");
+    }break;
+    
+    case EK_Compound:{
+      if(expr->compound.typespec){
+        printf("(");
+        typespec_print(expr->compound.typespec);
+        printf(")");
+      }
+      printf("{");
+      for(Expr_Compound_Field *n = expr->compound.first; n; n=n->next){
+        expr_compound_print(n);
+        if(n!=expr->compound.last) printf(",");
+      }
+      printf("}");
+    } break;
+    
+    case EK_SizeType:{
+      printf("size_type(");
+      printf(")");
+    }break;
+    
+    case EK_Paren:{
+      printf("(");
+      expr_print(expr->paren.expr);
+      printf(")");
+    } break;
+    
+    case EK_Field:{
+      expr_print(expr->field.expr);
+      printf(".%s", expr->field.name.s.str);
+    } break;
+    
+    case EK_Binary:{
+      printf("(");
+      expr_print(expr->binary.left);
+      token_print(expr->token);
+      expr_print(expr->binary.right);
+      printf(")");
+    } break;
+    
+    case EK_PostfixUnary:{
+      printf("(");
+      expr_print(expr->unary.expr);
+      token_print(expr->token);
+      printf(")");
+    } break;
+    
+    case EK_Unary:{
+      printf("(");
+      token_print(expr->token);
+      expr_print(expr->unary.expr);
+      printf(")");
+    } break;
+    
+    case EK_Ternary:{
+      printf("(");
+      expr_print(expr->ternary.cond);
+      printf("?");
+      expr_print(expr->ternary.on_true);
+      printf(":");
+      expr_print(expr->ternary.on_false);
+      printf(")");
+    } break;
+    
+    case EK_Cast:{
+      printf("(");
+      printf("(");
+      typespec_print(expr->cast.typespec);
+      printf(")");
+      expr_print(expr->cast.expr);
+      printf(")");
+    } break;
+    
+    case EK_Index:{
+      expr_print(expr->index.atom);
+      printf("[");
+      expr_print(expr->index.index);
+      printf("]");
+    }break;
+    
+    case EK_Call:{
+      expr_print(expr->call.atom);
+      printf("(");
+      for(Expr_Compound_Field *n = expr->call.first; n; n=n->next){
+        expr_compound_print(n);
+        if(n!=expr->call.last) printf(",");
+      }
+      printf(")");
+    }break;
+    default: {invalid_codepath;} break;
+  }
+}
+
+
+function B32
+typespec_print(Typespec *spec){
+  switch(spec->kind) {
+    case TS_Name: {
+      printf("%s", spec->name.s.str);
+    } break;
+    
+    case TS_NamedArgument: {
+      printf("%s: ", spec->named.name.s.str);
+      typespec_print(spec->named.base);
+    }break;
+    
+    case TS_Pointer: {
+      typespec_print(spec->base);
+      printf("*");
+    } break;
+    
+    case TS_Array: {
+      typespec_print(spec->arr.base);
+      printf("[");
+      expr_print(spec->arr.size);
+      printf("]");
+    } break;
+    
+    case TS_Function: {
+      printf("(");
+      for(Typespec *n = spec->func.first; n; n=n->next){
+        typespec_print(n);
+        if(n!=spec->func.last) 
+          printf(", ");
+      }
+      
+      printf(")");
+      typespec_print(spec->func.ret);
+    } break;
+    default: {invalid_codepath;} break;
+  }
+  return true;
+}
+
+function void
+print_assign_expr(Expr *expr){
+  if(expr){
+    printf(" = ");
+    expr_print(expr);
+  }
+}
+
+function void
+decl_print(Decl *node){
+  //print_notes(p, node->first_note);
+  
+  switch(node->kind) {
+    
+    case DECL_List: {
+      for(Decl *n = node->list_decl.first; n; n=n->next){
+        decl_print(n);
+        printf("\n");
+      }
+    } break;
+    
+    case DECL_Variable:{
+      println("%s: ", node->name.s.str);
+      B32 r = typespec_print(node->var_decl.type);
+      print_assign_expr(node->var_decl.expr);
+      if(r) printf(";");
+    } break;
+    
+    case DECL_Typedef:{
+      println("typedef %s ", node->name.s.str);
+      typespec_print(node->typedef_decl.type);
+      printf(";");
+    } break;
+    
+    
+    case DECL_Struct:
+    case DECL_Union :{
+      const char *struct_name = node->kind==DECL_Struct ? "struct" : "union";
+      println("%s %s{", struct_name, node->name.s.str?(char*)node->name.s.str:"");
+      indent++;
+      for(Decl *n = node->struct_decl.first; n; n=n->next){
+        decl_print(n);
+      }
+      indent--;
+      println("};");
+    } break;
+    
+    case DECL_SubStruct:
+    case DECL_SubUnion :{
+      const char *struct_name = node->kind==DECL_SubStruct ? "struct" : "union";
+      println("%s %s{", struct_name, node->name.s.str?(char*)node->name.s.str:"");
+      indent++;
+      for(Decl *n = node->struct_decl.first; n; n=n->next){
+        decl_print(n);
+      }
+      indent--;
+      println("};");
+    } break;
+    
+    
+    case DECL_Enum:{
+      println("enum %s : ", node->name.s.str);
+      typespec_print(node->enum_decl.typespec);
+      printf("{");
+      indent++;
+      for(Decl_Enum_Child *n = node->enum_decl.first; n; n=n->next){
+        //print_notes(p, n->first_note);
+        println("%s", n->name.s.str);
+        print_assign_expr(n->expr);
+        printf(",");
+      }
+      indent--;
+      println("};");
+    } break;
+    default: {invalid_codepath;} break;
+  }
+}
+