From 66bfe8a24d9b1ca83d45396a9f9c962379d1895e Mon Sep 17 00:00:00 2001
From: Oxore <oxore@protonmail.com>
Date: Sun, 25 Jun 2023 15:31:25 +0300
Subject: Fix parsing size specifier

---
 main.c | 61 +++++++++++++++++++++++++++++--------------------------------
 1 file changed, 29 insertions(+), 32 deletions(-)

diff --git a/main.c b/main.c
index dee5171..77df6cb 100644
--- a/main.c
+++ b/main.c
@@ -1344,21 +1344,22 @@ static size_t find_line_length(const char *const str)
 }
 
 static int pars_yield_error_str(
-        struct pars *const self, const struct line_pos_info l, char *const str)
+        struct pars *const self,
+        const struct line_pos_info l,
+        const char *const found,
+        const size_t found_length)
 {
     fprintf(
             stderr,
-            "<stdin>:%lu:%lu: parsing error: expected %s, found '%s'\n",
+            "<stdin>:%lu:%lu: parsing error: expected %s, found '%.*s'\n",
             l.line_num + 1,
             l.column_num + 1,
-            "<TODO>",
-            str);
-    free(str);
-    const size_t line_length = find_line_length(self->lex->input + l.line_offset);
-    char *const line = calloc(1, line_length + 1);
-    memcpy(line, self->lex->input + l.line_offset, line_length);
-    fprintf(stderr, "%5lu | %s\n", l.line_num + 1, line);
-    free(line);
+            "<_unspecified_token_list>",
+            (int)found_length,
+            found);
+    const char *const line = self->lex->input + l.line_offset;
+    const size_t line_length = find_line_length(line);
+    fprintf( stderr, "%5lu | %.*s\n", l.line_num, (int)line_length, line);
     fputs("      | ", stderr);
     for (size_t i = 0; i < l.column_num; i++) {
         if (self->lex->input[l.line_offset + i] == '\t') {
@@ -1376,9 +1377,8 @@ static int pars_yield_error(struct pars *const self, const size_t token_id)
     const struct token token = self->lex->tokbuf[token_id];
     const struct line_pos_info l =
         lex_get_line_pos_info(self->lex, token.offset);
-    char *const found = calloc(1, token.length + 10);
-    snprintf(found, token.length + 1, "%s", self->lex->input + token.offset);
-    return pars_yield_error_str(self, l, found);
+    const char *const found = self->lex->input + token.offset;
+    return pars_yield_error_str(self, l, found, token.length);
 }
 
 static int pars_yield_error_nesting(
@@ -1451,19 +1451,27 @@ static int pars_parse_arg(
         (sizeof *self->lex->tokbuf);
     const size_t first_token_id = self->cur_tok_id;
     int nesting = 0;
+    int commas = 0;
     enum arg_type arg_type = ARG_EXPR;
     while (self->cur_tok_id < tokens_count) {
         const size_t token_id = self->cur_tok_id; // Peek
         const struct token token = self->lex->tokbuf[token_id];
-        if (token.type == TT_LPAREN) {
+        if (nesting == 1 && token.type == TT_COMMA) {
+            if (commas >= 2) {
+                return pars_yield_error(self, self->cur_tok_id);
+            } else {
+                commas++;
+            }
+        } else if (token.type == TT_LPAREN) {
             nesting++;
         } else if (token.type == TT_RPAREN) {
             nesting--;
         } else if (is_expression_token(token.type)) {
             // TODO parse expression
-        } else if (nesting > 0 && token.type == TT_COMMA) {
-            // Comma inside parentheses is allowed
         } else {
+            if (nesting > 0) {
+                return pars_yield_error(self, self->cur_tok_id);
+            }
             break;
         }
         self->cur_tok_id++; // Commit
@@ -1644,30 +1652,19 @@ static int pars_parse_instruction(
     if (pars_is_eof_reached(self)) {
         return pars_yield_error_eof(self);
     }
-    const size_t token2_id = self->cur_tok_id; // Peek
-    const struct token token2 = self->lex->tokbuf[token2_id];
-    if (token2.type == TT_DOT) {
+    const size_t size_spec_id = self->cur_tok_id; // Peek
+    const struct token size_spec = self->lex->tokbuf[size_spec_id];
+    if (size_spec.type == TT_DOT_ID) {
         self->cur_tok_id++; // Commit
-        if (pars_is_eof_reached(self)) {
-            return pars_yield_error_eof(self);
-        }
-        const size_t size_spec_id = self->cur_tok_id++;
-        const struct token size_spec = self->lex->tokbuf[size_spec_id];
-        if (size_spec.type != TT_ID) {
-            return pars_yield_error(self, size_spec_id);
-        }
         // Size specifier
-        if (size_spec.length != 1) {
+        if (size_spec.length != 2) {
             return pars_yield_error(self, size_spec_id);
         }
         const size_t opsize =
-            get_opsize_from_specifier(self->lex->input[size_spec.offset]);
+            get_opsize_from_specifier(self->lex->input[size_spec.offset + 1]);
         if (opsize == OPSIZE_NONE) {
             return pars_yield_error(self, size_spec_id);
         }
-        if (pars_is_eof_reached(self)) {
-            return pars_yield_error_eof(self);
-        }
         return pars_parse_instruction_args(self, label_id, mnemonic_id, opsize);
     }
     return pars_parse_instruction_args(
-- 
cgit v1.2.3