Refactored the code and added expect functions to make the code flow a bit nicer.

misc/test.asm

@@ -6,6 +6,7 @@
 
 ;load r2, label
 something_insance:
+load r1, unknown_symbol
 load byte r1, r3
-load r1, r3
-load r1, 5
+load r1, 5
+load r3, r4

src/parser.c

@@ -11,21 +11,19 @@ List* TokensList;
 int CurrentToken = 0;
 
 void PrintSymbols(void);
-void HandleOperation(void);
 void RemoveCurrentToken(void);
 void HandleAssemblerDirective(void);
-void HandleAssemblerDirective(void);
 void AdvanceParser(void);
 void IgnoreParserLine(void);
 Token* PeekToken(void);
 int ParserAtEnd(void);
-int Expect(TokenClass class, void* value);
 IRState MachineState;
 
 void ExpectPuncuation(TokenPunctuation punctuation, ExpectOptions options);
 void ExpectRegister(void);
 Symbol* ExpectIdentifier(int resolved, ExpectOptions options);
-//int ExpectIdentifier();
+void ExpectCharacterClass(Symbol* symbol);
+void ExpectLineEndOrFileEnd(ExpectOptions options);
 
 int HeapSize = 0;
 int PC = 0;
@@ -41,7 +39,6 @@ IRState* ParseTokens(List* tokens) {
     while(!ParserAtEnd()) {
         Token* t = PeekToken();
         if (!t || t->EndOfFile) break;
-        Symbol* tmp;
 
         switch(t->Class) {
             case DirectiveClass:
@@ -52,18 +49,20 @@ IRState* ParseTokens(List* tokens) {
                 break;
             case LabelClass:
                 {
-                    int found = TryGetSymbol(t->Lemexe, MachineState.SymbolsTable, &tmp);
+                    Symbol* symbol;
+                    int found = TryGetSymbol(t->Lemexe, MachineState.SymbolsTable, &symbol);
 
-                    if (found && tmp->Resolved) {
+                    if (found && symbol->Resolved) {
                         fprintf(stderr, "[Error] Line %d: Redefinition of symbol '%s'\n", t->LineNumber, t->Lemexe);
                         exit(1);
                     }
 
                     if (!found) AddSymbolToTable(t->Lemexe, HeapSize, 1, MachineState.SymbolsTable);
-                    else tmp->Resolved = 1;
+                    else symbol->Resolved = 1;
 
                     RemoveCurrentToken(); //label
-                    RemoveCurrentToken(); //newline
+                    
+                    ExpectLineEndOrFileEnd(RemoveExpected);
                 }
                 break;
             default:
@@ -72,11 +71,6 @@ IRState* ParseTokens(List* tokens) {
                 //IgnoreParserLine();
                 break;
         }
-
-        // if (last == CurrentToken) {
-        //     printf("Failed %d\n", CurrentToken);
-        //     exit(90);
-        // }
     }
 
     PrintSymbols();
@@ -94,39 +88,18 @@ void HandleAssemblerDirective() {
 
                 Symbol* symbol = ExpectIdentifier(1, RemoveExpected);
 
-                Token* value = PeekToken();
-
-                if (value->Class == CharacterClass) {
-                    symbol->Length = strlen(value->Lemexe);
-
-                    RemoveCurrentToken(); //Remove the string declared by this DB command.
-
-                    if (PeekToken()->Class == PunctuationClass && PeekToken()->Value.Punctuation == Comma) {
-                        ExpectPuncuation(Comma, RemoveExpected);
-
-                        if (PeekToken()->Class != NumberClass) {
-                            fprintf(stderr, "Syntax error line %d: Expected terminating byte.\n", PeekToken()->LineNumber);
-                            exit(1);
-                        }
-
-                        RemoveCurrentToken(); //Remove terminating byte
-
-                        symbol->Length++;
-                    }
-                    else if (PeekToken()->Class != PunctuationClass || PeekToken()->Value.Punctuation != NewLine) {
-                        fprintf(stderr, "Syntax error line %d: Expected COMMA.", PeekToken()->LineNumber);
-                        exit(1);
-                    }
+                if (PeekToken()->Class == CharacterClass) {
+                    ExpectCharacterClass(symbol);
                 }
                 else if (PeekToken()->Class == NumberClass) {
                     symbol->Length = 2;
 
                     RemoveCurrentToken(); //Clear the number token.
+                    
+                    ExpectLineEndOrFileEnd(RemoveExpected);
                 }
 
                 HeapSize += symbol->Length;
-
-                RemoveCurrentToken(); //Wipe the line end token
             }
             break;
         case Load:
@@ -135,7 +108,7 @@ void HandleAssemblerDirective() {
                 Token* reg = PeekToken();
 
                 if (reg->Class != RegisterClass && reg->Class != DirectiveClass) {
-                    fprintf(stderr, "Syntax error line %d: Expected register or directive 'byte', got %d.", reg->LineNumber, reg->Class);
+                    fprintf(stderr, "[Line %d] Syntax error, expected register or keyword 'byte'.\n", reg->LineNumber);
                     exit(1);
                 }
 
@@ -153,23 +126,25 @@ void HandleAssemblerDirective() {
                         directive->Class = MnemonicClass;
                         directive->Value.Mnemonic = LODWI;
                         directive->Lemexe = "LODWI";
+                        AdvanceParser(); //Number, register or identifier
                     }
                     else if (PeekToken()->Class == RegisterClass) {
                         directive->Class = MnemonicClass;
                         directive->Value.Mnemonic = LODW;
                         directive->Lemexe = "LODW";
+                        AdvanceParser(); //Number, register or identifier
                     }
                     else {
+                        ExpectIdentifier(0, ForwardParser);
+
                         directive->Class = MnemonicClass;
                         directive->Value.Mnemonic = LAA;
                         directive->Lemexe = "LAA";
                     }
-
-                    AdvanceParser(); //Number, register or identifier
                 }
                 else if (reg->Class == DirectiveClass) {
                     if (reg->Value.Directive != Byte) {
-                        fprintf(stderr, "Syntax error: Line %d expected keyword 'byte'.\n", reg->LineNumber);
+                        fprintf(stderr, "[Line %d] Syntax error, expected keyword 'byte'.\n", reg->LineNumber);
                         exit(3);
                     }
                     
@@ -186,7 +161,7 @@ void HandleAssemblerDirective() {
                     directive->Lemexe = "LODB";
                 }
 
-                AdvanceParser(); //Line end
+                ExpectLineEndOrFileEnd(ForwardParser);
             }
 
             break;
@@ -255,6 +230,57 @@ Symbol* ExpectIdentifier(int resolved, ExpectOptions options) {
     return symbol;
 }
 
+void ExpectCharacterClass(Symbol* symbol) {
+    Token* token = PeekToken();
+
+    if (token->Class != CharacterClass) {
+        fprintf(stderr, "[Line %d] Syntax error, expected character string.\n", token->LineNumber);
+        exit(5);
+    }
+
+    symbol->Length = strlen(token->Lemexe);
+
+    RemoveCurrentToken(); //Remove the string declared by this DB command.
+
+    token = PeekToken();
+
+    if (token->EndOfFile) return;
+
+    if (token->Class == PunctuationClass && token->Value.Punctuation == NewLine) {
+        RemoveCurrentToken();
+        return;
+    }
+    
+    ExpectPuncuation(Comma, RemoveExpected);
+
+    token = PeekToken();
+
+    if (token->Class != NumberClass) {
+        fprintf(stderr, "[Line %d] Syntax error, expected terminating byte.\n", token->LineNumber);
+        exit(7);
+    }
+
+    RemoveCurrentToken(); //Remove terminating byte.
+    //TODO: add the raw value of the byte to the end of the string, but for now just pretend all numbers are zero.
+    symbol->Length++;
+
+    ExpectLineEndOrFileEnd(RemoveExpected);
+}
+
+void ExpectLineEndOrFileEnd(ExpectOptions options) {
+    Token* token = PeekToken();
+
+    if (token->EndOfFile) return;
+
+    if (token->Class != PunctuationClass || token->Value.Punctuation != NewLine) {
+        fprintf(stderr, "[Line %d] Syntax error, expected line break.\n", token->LineNumber);
+        exit(9);
+    }
+
+    if (options & RemoveExpected) RemoveCurrentToken();
+    if (options & ForwardParser) AdvanceParser();
+}
+
 void PrintSymbols(void) {
     //char mn[12];
     printf("-----SYMBOLS-----\n");