refactor masm to add codegen step

1 files changed, 67 insertions, 19 deletions

diff --git a/masm/lex.h b/masm/lex.h
index e08d0a3..8da6558 100644
--- a/masm/lex.h
+++ b/masm/lex.h
@@ -7,41 +7,89 @@
 #include <stdio.h>
 #include <stdint.h>
 
-struct lexer {
-	FILE *file;
-	int peek;
-	int x;
-	int y;
+/// represents a non null
+/// terminated string
+struct string {
+	char *str;
+	uint32_t len;
+	uint32_t size;
+	bool allocated;
 };
 
-struct lexer_state {
-	long offset;
-	int peek;
-	int x;
-	int y;
-};
+/* initalize a string */
+void string_init(struct string *string);
+/* free a string */
+void string_free(struct string *string);
+/* clone a string, leave the old one */
+int string_clone(struct string *dst, const struct string *const src);
+/* move a string, delete the old one */
+void string_move(struct string *dst, struct string *src);
+/* pushes a char onto a string */
+int string_push(struct string *string, char c);
+/* load a string from the bss (not allocated) */
+void string_bss(struct string *string, char *src);
 
 enum token_type {
-	TOK_IDENT,
-	TOK_REG,
-	TOK_LABEL,
-	TOK_STRING,
+	/// has no associated
+	/// data
 	TOK_COMMA,
 	TOK_EQUAL,
 	TOK_LPAREN,
 	TOK_RPAREN,
-	TOK_NUMBER,
 	TOK_EOF,
 	TOK_NL,
+
+	/// uses number
+	TOK_NUMBER,
+
+	/// uses string
+	TOK_REG,
+	TOK_IDENT,
+	TOK_LABEL,
+	TOK_STRING,
 	TOK_DIRECTIVE,
 };
 
+/// represents a token
+/// returned from the lexer
 struct token {
+	/// type
 	enum token_type type;
+
+	/// position
+	int x, y;
+	/// pos in bytes
+	int off;
+
+	/// data
 	union {
 		int64_t number;
-		char text[MAX_LEX_LENGTH];
+		struct string string;
 	};
+};
+
+/* frees a token*/
+void token_free(struct token *token);
+
+/// holds the data
+/// for the current lexer
+struct lexer {
+	// the currently
+	// open file
+	FILE *file;
+
+	// the last character peeked
+	int peek;
+
+	// the current position
+	int x, y;
+};
+
+/// holds a previous state of a
+/// lexer, which allows rebounding
+struct lexer_state {
+	long offset;
+	int peek;
 	int x;
 	int y;
 };
@@ -49,8 +97,8 @@ struct token {
 /* initalize a lexer */
 int lexer_init(const char *file, struct lexer *lexer);
 
-/* free the lxer */
-int lexer_free(struct lexer *lexer);
+/* free the lexer */
+void lexer_free(struct lexer *lexer);
 
 /* lexes the next token, returns M_ERROR on error,
  * and TOK_EOF on EOF */