More robust labels, implement symbol resolution.

src/assembler.c

@@ -14,8 +14,6 @@
 #include "rom.h"
 #include "util.h"
 
-#define IS_LABEL(line) (line->data[line->length - 1] == ':')
-
 /* Sentinel values for overlap table */
 const ASMLine header_sentinel, bounds_sentinel;
 
@@ -34,7 +32,7 @@ static ErrorInfo* add_label_to_table(
 
     const ASMSymbol *current = asm_symtable_find(symtable, symbol);
     if (current) {
-        ErrorInfo *ei = error_info_create(line, ET_LAYOUT, ED_LYT_DUPE_LABELS);
+        ErrorInfo *ei = error_info_create(line, ET_SYMBOL, ED_SYM_DUPE_LABELS);
         error_info_append(ei, current->line);
         return ei;
     }
@@ -43,7 +41,8 @@ static ErrorInfo* add_label_to_table(
     if (!label)
         OUT_OF_MEMORY()
 
-    label->offset = offset;
+    // TODO: don't assume all ROM gets mapped to slot 2
+    label->offset = (offset >= 0xC000) ? ((offset & 0x3FFF) + 0x8000) : offset;
     label->symbol = symbol;
     label->line = line;
     asm_symtable_insert(symtable, label);
@@ -144,7 +143,7 @@ static ErrorInfo* tokenize(AssemblerState *state)
         overlap_table[state->header.offset + i] = &header_sentinel;
 
     while (line) {
-        if (IS_LABEL(line)) {
+        if (line->is_label) {
             if ((ei = add_label_to_table(state->symtable, line, offset)))
                 goto cleanup;
         }
@@ -261,9 +260,31 @@ static ErrorInfo* resolve_defaults(AssemblerState *state)
 */
 static ErrorInfo* resolve_symbols(AssemblerState *state)
 {
-    // TODO
+    ErrorInfo *ei;
+    ASMInstruction *inst = state->instructions;
+    const ASMSymbol *symbol;
+
+    while (inst) {
+        if (inst->symbol) {
+            symbol = asm_symtable_find(state->symtable, inst->symbol);
+            if (!symbol) {
+                ei = error_info_create(inst->line, ET_SYMBOL, ED_SYM_NO_LABEL);
+                return ei;
+            }
 
-    (void) state;
+            if (inst->loc.length == 3) {
+                inst->b2 = symbol->offset & 0xFF;
+                inst->b3 = symbol->offset >> 8;
+            } else {
+                inst->b3 = symbol->offset & 0xFF;
+                inst->b4 = symbol->offset >> 8;
+            }
+
+            free(inst->symbol);
+            inst->symbol = NULL;
+        }
+        inst = inst->next;
+    }
     return NULL;
 }
 

src/assembler/errors.c

@@ -17,6 +17,7 @@ static const char *asm_error_types[] = {
     "include directive",    // ET_INCLUDE
     "preprocessor",         // ET_PREPROC
     "memory layout",        // ET_LAYOUT
+    "symbol table",         // ET_SYMBOL
     "instruction parser"    // ET_PARSER
 };
 
@@ -33,11 +34,13 @@ static const char *asm_error_descs[] = {
 
     "header offset exceeds given ROM size",             // ED_LYT_HEADER_RANGE
     "declared ROM size in header exceeds actual size",  // ED_LYT_DECLARE_RANGE
-    "duplicate definitions for label",                  // ED_LYT_DUPE_LABELS
     "location is out of bounds for the ROM size",       // ED_LYT_BOUNDS
     "location overlaps with instruction or data",       // ED_LYT_OVERLAP
     "location overlaps with ROM header",                // ED_LYT_OVERLAP_HEAD
 
+    "duplicate definitions for label",  // ED_SYM_DUPE_LABELS
+    "undefined reference to label",     // ED_SYM_NO_LABEL
+
     "syntax error"  // ED_PARSE_SYNTAX
 };
 

src/assembler/errors.h

@@ -13,6 +13,7 @@ typedef enum {
     ET_INCLUDE,
     ET_PREPROC,
     ET_LAYOUT,
+    ET_SYMBOL,
     ET_PARSER
 } ASMErrorType;
 
@@ -29,11 +30,13 @@ typedef enum {
 
     ED_LYT_HEADER_RANGE,
     ED_LYT_DECLARE_RANGE,
-    ED_LYT_DUPE_LABELS,
     ED_LYT_BOUNDS,
     ED_LYT_OVERLAP,
     ED_LYT_OVERLAP_HEAD,
 
+    ED_SYM_DUPE_LABELS,
+    ED_SYM_NO_LABEL,
+
     ED_PARSE_SYNTAX
 } ASMErrorDesc;
 

src/assembler/preprocessor.c

@@ -88,7 +88,7 @@
 static inline bool is_valid_label_char(char c, bool first)
 {
     return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') ||
-           (!first && c >= '0' && c <= '9') || c == '_';
+           (!first && c >= '0' && c <= '9') || c == '_' || c == '.';
 }
 
 /*
@@ -125,6 +125,7 @@ static size_t read_labels(
 
     strncpy(line->data, source + start, i - start + 1);
     line->length = i - start + 1;
+    line->is_label = true;
 
     nexti = read_labels(source + i + 1, length - i - 1, &line->next, tail_ptr);
     *head_ptr = line;
@@ -136,7 +137,7 @@ static size_t read_labels(
 /*
     Preprocess a single source line (source, length) into one or more ASMLines.
 
-    Only the data, length, and next fields of the ASMLine objects are
+    Only the data, length, is_label, and next fields of the ASMLine objects are
     populated. The normalization process strips comments, makes various
     adjustments outside of string literals (converts tabs to spaces, lowercases
     all alphabetical characters, and removes runs of multiple spaces), among
@@ -209,6 +210,7 @@ static ASMLine* normalize_line(const char *source, size_t length)
 
     line->data = data;
     line->length = di;
+    line->is_label = false;
     line->next = NULL;
 
     if (head) {  // Line has labels, so link the main part up
@@ -418,7 +420,7 @@ ErrorInfo* preprocess(AssemblerState *state, const LineBuffer *source)
 
     while ((prev = line, line = next)) {
         next = line->next;
-        if (line->data[0] != DIRECTIVE_MARKER)
+        if (line->is_label || line->data[0] != DIRECTIVE_MARKER)
             continue;
         if (IS_LOCAL_DIRECTIVE(line))
             continue;  // "Local" directives are handled by the tokenizer

src/assembler/state.c

@@ -144,7 +144,7 @@ static inline size_t hash_key(const char *key)
 /*
     Search for a key in the symbol table.
 
-    Return the key on success and NULL on failure.
+    Return the corresponding symbol on success and NULL on failure.
 */
 const ASMSymbol* asm_symtable_find(const ASMSymbolTable *tab, const char *key)
 {
@@ -159,6 +159,8 @@ const ASMSymbol* asm_symtable_find(const ASMSymbolTable *tab, const char *key)
 
 /*
     Insert a symbol into the table.
+
+    TODO: return boolean on success instead of void.
 */
 void asm_symtable_insert(ASMSymbolTable *tab, ASMSymbol *symbol)
 {

src/assembler/state.h

@@ -22,6 +22,7 @@ struct ASMLine {
     size_t length;
     const Line *original;
     const char *filename;
+    bool is_label;
     struct ASMLine *next;
 };
 typedef struct ASMLine ASMLine;
@@ -40,8 +41,8 @@ typedef struct {
 struct ASMInstruction {
     ASMLocation loc;
     uint8_t b1, b2, b3, b4;
-    uint8_t virtual_byte;
     char *symbol;
+    const ASMLine *line;
     struct ASMInstruction *next;
 };
 typedef struct ASMInstruction ASMInstruction;
@@ -54,7 +55,7 @@ struct ASMData {
 typedef struct ASMData ASMData;
 
 struct ASMSymbol {
-    size_t offset;
+    uint16_t offset;
     char *symbol;
     const ASMLine *line;
     struct ASMSymbol *next;