feat: Initial parsing of instructions

Initial ability to parse basic instructions. Currently hex numbers are
broken and immediates with labels (movl $str, %ecx) are broken.

5 files changed, 177 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..6cbb1b5
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+jas
+src/*.o
+src/parser.tab.*
+src/lex.yy.c
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..c10ffc6
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,36 @@
+CC         = gcc-15
+SRC_DIR    = src
+SRCS       = $(SRC_DIR)/lex.yy.c \
+             $(SRC_DIR)/parser.tab.c
+PROG       = jas
+OBJS       = $(patsubst %.c, %.o, \
+             $(patsubst %.cpp, %.o, \
+             $(patsubst %.cxx, %.o, \
+             $(patsubst %.cc, %.o, $(SRCS)))))
+
+$(PROG): $(OBJS)
+	$(CC) $(CFLAGS) -o $@ $^ $(LDFLAGS)
+
+$(SRC_DIR)/parser.tab.c: $(SRC_DIR)/parser.y
+	bison $(BISONFLAGS) -d $< -o $@
+
+$(SRC_DIR)/lex.yy.c: $(SRC_DIR)/lexer.l $(SRC_DIR)/parser.tab.c
+	flex $(FLEXFLAGS) -o $@ -l $<
+
+%.o: %.c
+	$(CC) $(CFLAGS) -c $< -o $@
+
+%.o: %.cpp
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+%.o: %.cxx
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+%.o: %.cc
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
+clean:
+	rm -rf $(PROG) $(OBJS) $(SRC_DIR)/lex.yy.c $(SRC_DIR)/parser.tab.c $(SRC_DIR)/parser.tab.h
+
+.PHONY: all clean
+
diff --git a/examples/instructions.s b/examples/instructions.s
new file mode 100644
index 0000000..f16e6c1
--- /dev/null
+++ b/examples/instructions.s
@@ -0,0 +1,8 @@
+movl $4, %eax
+movl $1, %ebx
+movl $1, %eax
+movl $0, %ebx
+int $0x80
+
+movl $str, %ecx
+movl $str_len, %edx
diff --git a/src/lexer.l b/src/lexer.l
new file mode 100644
index 0000000..46f6bc6
--- /dev/null
+++ b/src/lexer.l
@@ -0,0 +1,28 @@
+%{
+#include <string.h>
+#include <stdlib.h>
+#include "parser.tab.h"
+extern int yylex();
+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_LABEL; }
+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; }
+\".*\"                          {
+    yylval.s_val = strdup(yytext+1);
+    char *const close_paren = strrchr(yylval.s_val, '"');
+    if (close_paren != NULL) {
+        *close_paren = '\0';
+    }
+    return T_STRING;
+}
+,                               { return T_COMMA; }
+\$                              { return T_IMMEDIATE; }
+[ \t]                           ;
+\n                              { ++line_number; return T_ENDL; }
+.                               ;
+%%
diff --git a/src/parser.y b/src/parser.y
new file mode 100644
index 0000000..719643d
--- /dev/null
+++ b/src/parser.y
@@ -0,0 +1,101 @@
+%{
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <getopt.h>
+
+size_t line_number = 1;
+
+extern int yylex();
+extern int yyparse();
+extern FILE *yyin;
+void yyerror(const char *s);
+%}
+
+%union {
+    int i_val;
+    float f_val;
+    char *s_val;
+    char *symbol;
+}
+
+%token <i_val> T_INTEGER
+%token <f_val> T_FLOAT
+%token <s_val> T_STRING
+%token <symbol> T_LABEL
+%token <symbol> T_INSTRUCTION
+%token T_ENDL
+%token T_COMMA
+%token T_IMMEDIATE
+%%
+asm:
+    statements
+    ;
+
+statements:
+    statements statement
+    | statement
+    ;
+
+statement:
+    instruction_one_param_symbol T_ENDL
+    | instruction_one_param_immediate T_ENDL
+    | instruction_two_param_symbols T_ENDL
+    | instruction_symbol_immediate T_ENDL
+    | instruction_immediate_symbol T_ENDL
+    | T_ENDL
+    ;
+
+instruction_one_param_symbol:
+    T_INSTRUCTION T_LABEL {
+        printf("Found instruction %s(%s)\n", $1, $2);
+    };
+
+instruction_one_param_immediate:
+    T_INSTRUCTION T_IMMEDIATE T_INTEGER {
+        printf("Found instruction %s(%d)\n", $1, $3);
+    };
+instruction_two_param_symbols:
+    T_INSTRUCTION T_LABEL T_COMMA T_LABEL {
+        printf("Found instruction %s(%s, %s)\n", $1, $2, $4);
+    };
+
+instruction_symbol_immediate:
+    T_INSTRUCTION T_LABEL T_COMMA T_IMMEDIATE T_INTEGER {
+        printf("Found instruction %s(%s, %d)\n", $1, $2, $5);
+    }
+
+instruction_immediate_symbol:
+    T_INSTRUCTION T_IMMEDIATE T_INTEGER T_COMMA T_LABEL {
+        printf("Found instruction %s(%d, %s)\n", $1, $3, $5);
+    }
+%%
+
+void yyerror(const char *s) {
+    printf("Parser Error: %lu: %s\n", line_number, s);
+}
+
+int main(int argc, char *argv[]) {
+    FILE *input = stdin;
+
+    int j = 0;
+    while (j != -1) {
+        j = getopt(argc, argv, "o:");
+
+        switch (j) {
+        case 'o':
+            input = fopen(optarg, "r");
+            if (input == NULL) {
+                perror(argv[0]);
+                return -1;
+            }
+            break;
+        default:
+            break;
+        }
+    }
+
+    yyin = input;
+    yyparse();
+}
+