Support parsing labels; add checks for label length.

src/assembler/errors.c

@@ -42,6 +42,7 @@ static const char *error_descs[] = {
 
     [ED_SYM_DUPE_LABELS]  = "duplicate definitions for label",
     [ED_SYM_NO_LABEL]     = "undefined reference to label",
+    [ED_SYM_TOO_LONG]     = "label name is too long",
     [ED_SYM_IS_REGISTER]  = "labels cannot share names with registers",
     [ED_SYM_IS_CONDITION] = "labels cannot share names with condition codes",
 

src/assembler/errors.h

@@ -41,6 +41,7 @@ typedef enum {
 
     ED_SYM_DUPE_LABELS,
     ED_SYM_NO_LABEL,
+    ED_SYM_TOO_LONG,
     ED_SYM_IS_REGISTER,
     ED_SYM_IS_CONDITION,
 

src/assembler/inst_args.h

@@ -5,6 +5,8 @@
 
 #include <stdint.h>
 
+#include "state.h"
+
 typedef enum {
     AT_REGISTER,
     AT_IMMEDIATE,
@@ -50,7 +52,9 @@ typedef struct {
     int8_t offset;
 } ASMArgIndexed;
 
-typedef char* ASMArgLabel;
+typedef struct {
+    char text[MAX_SYMBOL_SIZE];
+} ASMArgLabel;
 
 typedef enum {
     COND_NZ, COND_N, COND_NC, COND_C, COND_PO, COND_PE, COND_P, COND_M

src/assembler/instructions.c

@@ -143,28 +143,24 @@ static uint8_t fill_bytes_variadic(uint8_t *bytes, size_t len, ...)
 
     Return ED_NONE (0) on success or an error code on failure.
 */
-static ASMErrorDesc parse_arg(
-    ASMInstArg *arg, const char *str, size_t size, char **symbol)
+static ASMErrorDesc parse_arg(ASMInstArg *arg, const char *str, size_t size)
 {
-#define USE_PARSER(func, argtype, field)                                      \
+#define TRY_PARSER(func, argtype, field)                                      \
     if (argparse_##func(&arg->data.field, str, size)) {                       \
         arg->type = argtype;                                                  \
         return ED_NONE;                                                       \
     }
 
     DEBUG("parse_arg(): -->%.*s<-- %zu", (int) size, str, size)
-    USE_PARSER(register, AT_REGISTER, reg)
-    USE_PARSER(immediate, AT_IMMEDIATE, imm)
-    USE_PARSER(indirect, AT_INDIRECT, indirect)
-    USE_PARSER(indexed, AT_INDEXED, index)
-    USE_PARSER(condition, AT_CONDITION, cond)
-
-    // AT_LABEL
-    // ASMArgLabel label;
-
+    TRY_PARSER(register, AT_REGISTER, reg)
+    TRY_PARSER(immediate, AT_IMMEDIATE, imm)
+    TRY_PARSER(indirect, AT_INDIRECT, indirect)
+    TRY_PARSER(indexed, AT_INDEXED, index)
+    TRY_PARSER(condition, AT_CONDITION, cond)
+    TRY_PARSER(label, AT_LABEL, label)
     return ED_PS_ARG_SYNTAX;
 
-#undef USE_PARSER
+#undef TRY_PARSER
 }
 
 /*
@@ -176,7 +172,6 @@ static ASMErrorDesc parse_args(
     ASMInstArg args[3], size_t *nargs, const char *str, size_t size)
 {
     ASMErrorDesc err;
-    static char *symbol = NULL;
     size_t start = 0, i = 0;
 
     while (i < size) {
@@ -184,7 +179,7 @@ static ASMErrorDesc parse_args(
         if (c == ',') {
             if (i == start)
                 return ED_PS_ARG_SYNTAX;
-            if ((err = parse_arg(&args[*nargs], str + start, i - start, &symbol)))
+            if ((err = parse_arg(&args[*nargs], str + start, i - start)))
                 return err;
             (*nargs)++;
 
@@ -207,7 +202,7 @@ static ASMErrorDesc parse_args(
     }
 
     if (i > start) {
-        if ((err = parse_arg(&args[*nargs], str + start, i - start, &symbol)))
+        if ((err = parse_arg(&args[*nargs], str + start, i - start)))
             return err;
         (*nargs)++;
     }

src/assembler/parse_util.c

@@ -412,6 +412,26 @@ bool argparse_indexed(ASMArgIndexed *result, const char *arg, ssize_t size)
 }
 
 /*
+    Read in a label argument and store it in *result.
+*/
+bool argparse_label(ASMArgLabel *result, const char *arg, ssize_t size)
+{
+    if (size >= MAX_SYMBOL_SIZE)
+        return false;
+
+    for (const char *i = arg; i < arg + size; i++) {
+        char c = *i;
+        if (!((c >= 'a' && c <= 'z') || (i != arg && c >= '0' && c <= '9') ||
+              c == '_' || c == '.'))
+            return false;
+    }
+
+    strncpy(result->text, arg, size);
+    result->text[size] = '\0';
+    return true;
+}
+
+/*
     Read in a boolean argument from the given line and store it in *result.
 */
 DIRECTIVE_PARSE_FUNC(bool, bool)

src/assembler/parse_util.h

@@ -25,6 +25,7 @@ bool argparse_condition(ASMArgCondition*, const char*, ssize_t);
 bool argparse_immediate(ASMArgImmediate*, const char*, ssize_t);
 bool argparse_indirect(ASMArgIndirect*, const char*, ssize_t);
 bool argparse_indexed(ASMArgIndexed*, const char*, ssize_t);
+bool argparse_label(ASMArgLabel*, const char*, ssize_t);
 
 /* Preprocessor directive parsers */
 bool dparse_bool(bool*, const ASMLine*, const char*);

src/assembler/state.h

@@ -13,6 +13,7 @@
 #define DEFAULT_REGION 6  // GG Export
 #define DEFAULT_DECLSIZE 0xC  // 32 KB
 
+#define MAX_SYMBOL_SIZE 256
 #define SYMBOL_TABLE_BUCKETS 128
 
 /* Structs */

src/assembler/tokenizer.c

@@ -62,6 +62,9 @@ static inline int8_t default_bank_slot(uint8_t bank)
 static ErrorInfo* add_label_to_table(
     ASMSymbolTable *symtable, const ASMLine *line, size_t offset, int8_t slot)
 {
+    if (line->length - 1 >= MAX_SYMBOL_SIZE)
+        return error_info_create(line, ET_SYMBOL, ED_SYM_TOO_LONG);
+
     ASMArgRegister reg;
     if (argparse_register(&reg, line->data, line->length - 1))
         return error_info_create(line, ET_SYMBOL, ED_SYM_IS_REGISTER);