feat: Parsing modifiers and directives

Support added for parsing a limited set of directives and for all access
modifiers.

2 files changed, 110 insertions, 20 deletions

diff --git a/src/lexer.l b/src/lexer.l
index 26be828..ac579e7 100644
--- a/src/lexer.l
+++ b/src/lexer.l
@@ -7,11 +7,53 @@ extern size_t line_number;
 %}
 %option noyywrap
 %%
-^[a-zA-Z]+                      { yylval.symbol = strdup(yytext); return T_INSTRUCTION; }
-%*[a-zA-Z\_\-]+[a-zA-Z0-9\_\-]* { yylval.symbol = strdup(yytext); return T_SYMBOL; }
 0x[0-9A-Fa-f]+                  { yylval.i_val = atoi(yytext); return T_INTEGER; }
-[0-9]+                          { yylval.i_val = atoi(yytext); return T_INTEGER; }
--*[0-9]*\.*[0-9]*               { yylval.i_val = atof(yytext); return T_FLOAT; }
+-*[0-9]+                        { yylval.i_val = atoi(yytext); return T_INTEGER; }
+^[a-zA-Z0-9\.\_]+:              {
+    yylval.symbol = strdup(yytext);
+    char *const colon = strrchr(yylval.symbol, ':');
+    if (colon != NULL) {
+        *colon = '\0';
+    }
+    return T_LABEL;
+}
+.align                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.globl                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.local                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.ident                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.section                        { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.size                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.text                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.data                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.rodata                         { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.bss                            { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.string                         { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.asciz                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.equ                            { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.byte                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.2byte                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.half                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.short                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.4byte                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.word                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.long                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.8byte                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.dword                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.quad                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.float                          { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.double                         { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+.zero                           { yylval.directive = strdup(yytext); return T_DIRECTIVE; }
+%hi                             { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%lo                             { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%pcrel_hi                       { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%pcrel_lo                       { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%tprel_hi                       { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%tprel_lo                       { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%tprel_add                      { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%tls_ie_pcrel_hi                { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%tls_gd_pcrel_hi                { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+%got_pcrel_hi                   { yylval.modifier = strdup(yytext); return T_MODIFIER; }
+[a-zA-Z\_\.]+[a-zA-Z0-9\_\.]*   { yylval.modifier = strdup(yytext); return T_SYMBOL; }
 \".*\"                          {
     yylval.s_val = strdup(yytext+1);
     char *const close_paren = strrchr(yylval.s_val, '"');
diff --git a/src/parser.y b/src/parser.y
index e63935a..5f7b4bb 100644
--- a/src/parser.y
+++ b/src/parser.y
@@ -14,20 +14,22 @@ void yyerror(const char *s);
 
 %union {
     int i_val;
-    float f_val;
     char *s_val;
     char *symbol;
+    char *modifier;
+    char *directive;
 }
 
-%token <i_val> T_INTEGER
-%token <f_val> T_FLOAT
-%token <s_val> T_STRING
-%token <symbol> T_SYMBOL
-%token <symbol> T_INSTRUCTION
-%token T_ENDL
-%token T_COMMA
-%token T_OPENPAREN
-%token T_CLOSEPAREN
+%token <i_val>     T_INTEGER
+%token <s_val>     T_STRING
+%token <symbol>    T_SYMBOL
+%token <directive> T_DIRECTIVE
+%token <symbol>    T_LABEL
+%token <modifier>  T_MODIFIER
+%token             T_ENDL
+%token             T_COMMA
+%token             T_OPENPAREN
+%token             T_CLOSEPAREN
 %%
 asm:
     statements
@@ -39,48 +41,94 @@ statements:
     ;
 
 statement:
-    instructions
+    label
+    | directives
+    | instructions
     | T_ENDL
     ;
 
+label:
+    T_LABEL T_ENDL
+    {
+        printf("Found Label: %s\n", $1);
+    };
+
 instructions:
     rb_type
     | i_type
     | s_type
     | u_type
     | j_type
+    | i_type_modifier
+    | s_type_modifier
+    | pseudo_type
     ;
 
 rb_type:
-    T_INSTRUCTION T_SYMBOL T_COMMA T_SYMBOL T_COMMA T_SYMBOL T_ENDL
+    T_SYMBOL T_SYMBOL T_COMMA T_SYMBOL T_COMMA T_SYMBOL T_ENDL
     {
         printf("Read instruction: %s(%s, %s, %s)\n", $1, $2, $4, $6);
     };
 
 i_type:
-    T_INSTRUCTION T_SYMBOL T_COMMA T_SYMBOL T_COMMA T_INTEGER T_ENDL
+    T_SYMBOL T_SYMBOL T_COMMA T_SYMBOL T_COMMA T_INTEGER T_ENDL
     {
         printf("Read instruction: %s(%s, %s, %d)\n", $1, $2, $4, $6);
     };
 
+i_type_modifier:
+    T_SYMBOL T_SYMBOL T_COMMA T_SYMBOL T_COMMA T_MODIFIER T_OPENPAREN T_SYMBOL T_CLOSEPAREN T_ENDL
+    {
+        printf("Read instruction: %s(%s, %s, %s of %s)\n", $1, $2, $4, $6, $8);
+    };
+
 s_type:
-    T_INSTRUCTION T_SYMBOL T_COMMA T_INTEGER T_OPENPAREN T_SYMBOL T_CLOSEPAREN T_ENDL
+    T_SYMBOL T_SYMBOL T_COMMA T_INTEGER T_OPENPAREN T_SYMBOL T_CLOSEPAREN T_ENDL
     {
         printf("Read instruction: %s(%s, %s + %d)\n", $1, $2, $6, $4);
     };
 
+s_type_modifier:
+    T_SYMBOL T_SYMBOL T_COMMA T_MODIFIER T_OPENPAREN T_SYMBOL T_CLOSEPAREN T_ENDL
+    {
+        printf("Read instruction: %s(%s, %s of %s)\n", $1, $2, $4, $6);
+    };
+
 u_type:
-    T_INSTRUCTION T_SYMBOL T_COMMA T_INTEGER T_ENDL
+    T_SYMBOL T_SYMBOL T_COMMA T_INTEGER T_ENDL
     {
         printf("Read instruction: %s(%s, %d)\n", $1, $2, $4);
     };
 
 j_type:
-    T_INSTRUCTION T_SYMBOL T_COMMA T_SYMBOL T_ENDL
+    T_SYMBOL T_SYMBOL T_COMMA T_SYMBOL T_ENDL
     {
         printf("Read instruction: %s(%s, %s)\n", $1, $2, $4);
     };
 
+pseudo_type:
+    T_SYMBOL T_SYMBOL T_ENDL
+    {
+        printf("Read Pseudo Instruction: %s(%s)\n", $1, $2);
+    };
+
+directives:
+    directive_int
+    | directive_string
+    ;
+
+directive_int:
+    T_DIRECTIVE T_INTEGER T_ENDL
+    {
+        printf("Read directive %s %d\n", $1, $2);
+    };
+
+directive_string:
+    T_DIRECTIVE T_STRING T_ENDL
+    {
+        printf("Read directive %s %s\n", $1, $2);
+    };
+
 %%
 
 void yyerror(const char *s) {