M xywasm.c → xywasm.c

@@ -339,23 +339,18 @@ token_type = TOKEN_TYPE_SYMBOL;
     return 0;
 }
 
+// Returns: 1 token processed, 0 EOF, -1 error
 static int next_token(xyw_context *ctx);
 
 static int process_directive_label(xyw_context *ctx)
 {
-    (void)ctx;
-    /*
-    if (next_token(ctx) != 0)
-    {
-        return -1;
-    }
-    if (token_type != TOKEN_TYPE_SYMBOL)
+    int r = next_token(ctx);
+    if (r != 1 || token_type != TOKEN_TYPE_SYMBOL)
     {
         token_error("Expected symbol after .label directive");
         return -1;
     }
-    */
-    // TODO: store label in token.
+    // TODO: store label info
     return 0;
 }
 
@@ -424,103 +419,131 @@ 
 static int next_token(xyw_context *ctx)
 {
     unsigned char c;
+    token_length = 0;
+    token[0] = '\0';
+    int in_string = 0;
+
     while (fread(&c, 1, 1, ctx->fp) == 1)
     {
+        if (!in_string && c == '\n')
+        {
+            if (token_length > 0)
+            {
+                if (process_token(ctx) != 0)
+                    return -1;
+                token_length = 0;
+                token[0] = '\0';
+                ctx->line++;
+                ctx->column = 1;
+                reset_directive();
+                return 1;
+            }
+            ctx->line++;
+            ctx->column = 1;
+            reset_directive();
+            continue;
+        }
+        if (!in_string && token_length == 0 && c == ';')
+        {
+            token[token_length++] = (char)c;
+            token[token_length] = '\0';
+            ctx->column++;
+            if (process_token(ctx) != 0)
+                return -1;
+            token_length = 0;
+            token[0] = '\0';
+            return 1;
+        }
         if (c == '"')
         {
-            // Special handling for string tokens
-            if (token_length < STRING_MAX_LENGTH - 1)
+            ctx->column++;
+            if (!in_string)
             {
-                token[token_length++] = (char)c;
-                token[token_length] = '\0';
-                ctx->column++;
-                while (fread(&c, 1, 1, ctx->fp) == 1)
+                in_string = 1;
+                if (token_length < STRING_MAX_LENGTH - 1)
                 {
-                    ctx->column++;
-                    if (token_length < STRING_MAX_LENGTH - 1)
-                    {
-                        token[token_length++] = (char)c;
-                        token[token_length] = '\0';
-                    }
-                    else
-                    {
-                        token_error("Token too long");
-                        return -1;
-                    }
-                    if (c == '"')
-                    {
-                        break;
-                    }
+                    token[token_length++] = (char)c;
+                    token[token_length] = '\0';
                 }
-                // Process string token
-                if (process_token(ctx) != 0)
+                else
                 {
+                    token_error("Token too long");
                     return -1;
                 }
-                token_length = 0;
-                token[0] = '\0';
+                continue;
             }
             else
             {
-                token_error("Token too long");
-                return -1;
+                if (token_length < STRING_MAX_LENGTH - 1)
+                {
+                    token[token_length++] = (char)c;
+                    token[token_length] = '\0';
+                }
+                else
+                {
+                    token_error("Token too long");
+                    return -1;
+                }
+                if (process_token(ctx) != 0)
+                    return -1;
+                token_length = 0;
+                token[0] = '\0';
+                in_string = 0;
+                return 1;
             }
-            continue;
         }
-        if (c > 0x20 && c != 0x7f) // printable non-space
+        else if (in_string)
         {
-            if (token_length < TOKEN_MAX_LENGTH - 1)
+            ctx->column++;
+            if (token_length < STRING_MAX_LENGTH - 1)
             {
                 token[token_length++] = (char)c;
                 token[token_length] = '\0';
-                ctx->column++;
-                // Save directive data if in a directive
-                if (directive_length > 0)
-                {
-                    if (directive_data_length >= DIRECTIVE_CONTENT_MAX_LENGTH - 1)
-                    {
-                        token_error("Directive data too long");
-                        return -1;
-                    }
-                    directive_data[directive_data_length++] = (char)c;
-                    directive_data[directive_data_length] = '\0';
-                }
             }
             else
             {
                 token_error("Token too long");
                 return -1;
             }
+            continue;
         }
         else if (c == ' ' || c == '\t')
         {
-            // whitespace separator: finalize token if present
             if (token_length > 0)
             {
                 if (process_token(ctx) != 0)
-                {
                     return -1;
-                }
                 token_length = 0;
                 token[0] = '\0';
+                ctx->column++;
+                return 1;
             }
             ctx->column++;
+            continue;
         }
-        else if (c == '\n')
+        else if (c > 0x20 && c != 0x7f)
         {
-            if (token_length > 0)
+            ctx->column++;
+            if (token_length < TOKEN_MAX_LENGTH - 1)
             {
-                if (process_token(ctx) != 0)
+                token[token_length++] = (char)c;
+                token[token_length] = '\0';
+                if (directive_length > 0)
                 {
-                    return -1;
+                    if (directive_data_length >= DIRECTIVE_CONTENT_MAX_LENGTH - 1)
+                    {
+                        token_error("Directive data too long");
+                        return -1;
+                    }
+                    directive_data[directive_data_length++] = (char)c;
+                    directive_data[directive_data_length] = '\0';
                 }
-                token_length = 0;
-                token[0] = '\0';
+            }
+            else
+            {
+                token_error("Token too long");
+                return -1;
             }
-            ctx->line++;
-            ctx->column = 1;
-            // Directives are line-based, so reset directive data
-            reset_directive();
         }
         else
         {
@@ -528,7 +551,15 @@ token_error("Invalid character encountered");
             return -1;
         }
     }
-    return 0;
+    if (token_length > 0)
+    {
+        if (process_token(ctx) != 0)
+            return -1;
+        token_length = 0;
+        token[0] = '\0';
+        return 1; // last token at EOF
+    }
+    return 0; // EOF no token
 }
 
 // Read a file and process its tokens
@@ -541,24 +572,13 @@ error("Could not open file");
         return -1;
     }
 
-    if (next_token(ctx) != 0)
-    {
-        fclose(ctx->fp);
-        return -1;
-    }
-
-    // Final token at EOF
-    if (token_length > 0)
+    int r;
+    while ((r = next_token(ctx)) == 1)
     {
-        if (process_token(ctx) != 0)
-        {
-            fclose(ctx->fp);
-            return -1;
-        }
+        // tokens processed inline
     }
-
     fclose(ctx->fp);
-    return 0;
+    return r;
 }
 
 ///// Public API