Tried to create an AST

7 files changed, 138 insertions, 18 deletions

diff --git a/Makefile b/Makefile
index 76b7081..8f87383 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,13 @@
 build: grammar.y lex.l
-	bison -o build/grammar.tab.c -d grammar.y
-	flex -o build/lex.yy.c lex.l
-	gcc -o build/henshin build/grammar.tab.c build/lex.yy.c -lfl -ly
+	bison \
+		-o build/grammar.tab.c \
+		-d grammar.y \
+		#-Wcounterexamples
+	flex \
+		-o build/lex.yy.c \
+		lex.l
+	gcc \
+		-o build/henshin \
+		build/grammar.tab.c build/lex.yy.c \
+		-I ./ \
+		-lfl -ly
diff --git a/Readme.txt b/Readme.txt
index 7ccd854..09f9e49 100644
--- a/Readme.txt
+++ b/Readme.txt
@@ -5,3 +5,4 @@
 - https://www.ecosia.org/search?q=transpiler%20with%20bison%20and%20lex
 - https://github.com/labis7/Flex_Bison_Transpiler
 - https://gnuu.org/2009/09/18/writing-your-own-toy-compiler/
+- https://lloydrochester.com/post/flex-bison/json-parse-ast/
diff --git a/ast.c b/ast.c
new file mode 100644
index 0000000..d2b8782
--- /dev/null
+++ b/ast.c
@@ -0,0 +1,18 @@
+#include "ast.h"
+
+
+ast_node* create_node() {
+  ast_node* node = malloc(sizeof(ast_node));
+
+  return node;
+}
+
+ast_node* create_program_node(ast_node* previous_node, ast_node* current_node) {
+  ast_node* node = malloc(sizeof(ast_node));
+
+  node->type = PROGRAM_NODE;
+  node->left = previous_node;
+  node->right = current_node;
+
+  return node;
+}
diff --git a/ast.h b/ast.h
new file mode 100644
index 0000000..3b59d2b
--- /dev/null
+++ b/ast.h
@@ -0,0 +1,17 @@
+#ifndef AST_H
+#define AST_H
+
+#include <stdlib.h>
+
+#define PROGRAM_NODE 0
+
+typedef struct ast_node {
+  int type;
+  void* value;
+  struct ast_node* left;
+  struct ast_node* right;
+} ast_node;
+
+ast_node* create_program_node(ast_node* previous_node, ast_node* current_node);
+
+#endif // AST_H
diff --git a/grammar.y b/grammar.y
index 446b651..c85da97 100644
--- a/grammar.y
+++ b/grammar.y
@@ -1,13 +1,24 @@
 %{
 #include <stdio.h>
 
+#include "ast.c"
+
 extern FILE *yyin;
+
 extern int yylex();
 extern void yyerror();
+
+ast_node* yyast = NULL;
 %}
 
 %define parse.error verbose
 
+%union {
+  char* string;
+  int number;
+  ast_node* node;
+}
+
 %token OPERATOR_PLUS
 %token OPERATOR_MINUS
 %token ASSIGN
@@ -19,6 +30,10 @@ extern void yyerror();
 %token CONST
 %token VAR
 
+%token TYPE_INTEGER
+%token TYPE_STRING
+%token TYPE_VOID
+
 %token PARENTHESIS_LEFT
 %token PARENTHESIS_RIGHT
 %token BRACE_LEFT
@@ -27,6 +42,7 @@ extern void yyerror();
 %token BRACKET_RIGHT
 
 %token COMMA;
+%token COLON;
 
 %token NUMBER;
 %token IDENTIFIER
@@ -34,20 +50,65 @@ extern void yyerror();
 
 %token END_OF_LINE;
 
+%type <node> program
+%type <node> function
+%type <node> arguments
+%type <node> argument
+%type <node> return_type
+%type <node> variable_type
+%type <node> statements
+%type <node> statement
+%type <node> expression
+
+%type <node> IDENTIFIER
+%type <node> RETURN
+
+%start program
+
 %%
-program:      
-            | program statement END_OF_LINE
-            ;
-
-statement:    expression
-            | RETURN expression
-            | CONST IDENTIFIER ASSIGN expression
-            | VAR IDENTIFIER ASSIGN expression
-            ;
-
-expression:   NUMBER
-            | IDENTIFIER
-            ;
+program: { $$ = NULL; }
+  | program statement END_OF_LINE { $$ = create_program_node($1, $2); }
+
+
+// function henshin(): void {}
+// function henshin(hen: integer, shin: integer): void {}
+function:
+  | FUNCTION IDENTIFIER PARENTHESIS_LEFT PARENTHESIS_RIGHT COLON return_type BRACE_LEFT statements BRACE_RIGHT
+  | FUNCTION IDENTIFIER PARENTHESIS_LEFT arguments PARENTHESIS_RIGHT COLON return_type BRACE_LEFT statements BRACE_RIGHT { $$ = create_node(); $$->value = $2; }
+
+arguments:
+  | argument
+  | arguments COMMA argument
+
+argument:
+  | IDENTIFIER COLON variable_type
+
+
+return_type:
+  | TYPE_INTEGER
+  | TYPE_STRING
+  | TYPE_VOID
+
+variable_type:
+  | TYPE_INTEGER
+  | TYPE_STRING
+
+
+statements:
+  | statement
+  | statements END_OF_LINE statement
+
+statement:
+  // const henshin: integer = 2
+  | CONST IDENTIFIER COLON variable_type ASSIGN expression
+  | RETURN expression { $$ = create_node(); $$->left = $1; $$->right = $2; }
+  | IDENTIFIER PARENTHESIS_LEFT PARENTHESIS_RIGHT
+  | function
+
+
+expression:
+  | NUMBER
+  | IDENTIFIER
 %%
 
 void main (int argc, char **argv)
@@ -55,4 +116,6 @@ void main (int argc, char **argv)
   //henshin_lex();
   yyin = fopen(argv[1], "r");
   yyparse();
+
+  printf("%s", yyast);
 }
diff --git a/lex.l b/lex.l
index 2f79474..64c3765 100644
--- a/lex.l
+++ b/lex.l
@@ -1,4 +1,5 @@
 %{
+#include "ast.h"
 #include "grammar.tab.h"
 %}
 
@@ -14,6 +15,10 @@
 "const"                 { return CONST; }
 "var"                   { return VAR; }
 
+"integer"               { return TYPE_INTEGER; }
+"string"                { return TYPE_STRING; }
+"void"                  { return TYPE_VOID; }
+
 "("                     { return PARENTHESIS_LEFT; }
 ")"                     { return PARENTHESIS_RIGHT; }
 "{"                     { return BRACE_LEFT; }
@@ -22,9 +27,10 @@
 "]"                     { return BRACKET_RIGHT; }
 
 ","                     { return COMMA; }
+":"                     { return COLON; }
 
-[0-9]+                  { yylval = atoi(yytext); return NUMBER; }
-[a-zA-Z]+[a-zA-Z0-9]*   { return IDENTIFIER; }
+[a-zA-Z][a-zA-Z0-9]*    { yylval.string = yytext; return IDENTIFIER; }
+[0-9]+                  { yylval.number = atoi(yytext); return NUMBER; }
 "//".*                  { return COMMENT; }
 
 \n                      { return END_OF_LINE; }
diff --git a/test/test.test b/test/test.test
new file mode 100644
index 0000000..44580c7
--- /dev/null
+++ b/test/test.test
@@ -0,0 +1,6 @@
+function henshin(hen: integer, shin: integer): void {
+  const x: integer = 2
+  const b: string = x
+
+  return b
+}