[Bug #19882] Reject tokens invalid as symbols

nobu · nobu · commit 89802078f9f4 · 2023-09-15T17:27:36.000+09:00
diff --git a/bootstraptest/test_syntax.rb b/bootstraptest/test_syntax.rb
@@ -535,8 +535,8 @@ def assert_syntax_error expected, code, message = ''
 assert_equal %q{[]}, %q{$&;[]}, '[ruby-dev:31068]'
 assert_syntax_error "syntax error, unexpected *, expecting '}'", %q{{*0}}, '[ruby-dev:31072]'
 assert_syntax_error "`@0' is not allowed as an instance variable name", %q{@0..0}, '[ruby-dev:31095]'
-assert_syntax_error "identifier $00 is not valid to get", %q{$00..0}, '[ruby-dev:31100]'
-assert_syntax_error "identifier $00 is not valid to set", %q{0..$00=1}
+assert_syntax_error "`$00' is not allowed as a global variable name", %q{$00..0}, '[ruby-dev:31100]'
+assert_syntax_error "`$00' is not allowed as a global variable name", %q{0..$00=1}
 assert_equal %q{0}, %q{[*0];0}, '[ruby-dev:31102]'
 assert_syntax_error "syntax error, unexpected ')'", %q{v0,(*,v1,) = 0}, '[ruby-dev:31104]'
 assert_equal %q{1}, %q{
diff --git a/parse.y b/parse.y
@@ -573,6 +573,7 @@ static void numparam_name(struct parser_params *p, ID id);
 #define STR_NEW2(ptr) rb_enc_str_new((ptr),strlen(ptr),p->enc)
 #define STR_NEW3(ptr,len,e,func) parser_str_new(p, (ptr),(len),(e),(func),p->enc)
 #define TOK_INTERN() intern_cstr(tok(p), toklen(p), p->enc)
+#define VALID_SYMNAME_P(s, l, enc, type) (rb_enc_symname_type(s, l, enc, (1U<<(type))) == (int)(type))
 
 static st_table *
 push_pvtbl(struct parser_params *p)
@@ -9640,7 +9641,13 @@ parse_gvar(struct parser_params *p, const enum lex_state_e last_state)
 
     if (tokadd_ident(p, c)) return 0;
     SET_LEX_STATE(EXPR_END);
-    tokenize_ident(p);
+    if (VALID_SYMNAME_P(tok(p), toklen(p), p->enc, ID_GLOBAL)) {
+        tokenize_ident(p);
+    }
+    else {
+        compile_error(p, "`%.*s' is not allowed as a global variable name", toklen(p), tok(p));
+        set_yylval_noname();
+    }
     return tGVAR;
 }
 
@@ -13661,7 +13668,7 @@ rb_reg_named_capture_assign_iter_impl(struct parser_params *p, const char *s, lo
     NODE *node, *succ;
 
     if (!len) return ST_CONTINUE;
-    if (rb_enc_symname_type(s, len, enc, (1U<<ID_LOCAL)) != ID_LOCAL)
+    if (!VALID_SYMNAME_P(s, len, enc, ID_LOCAL))
         return ST_CONTINUE;
 
     var = intern_cstr(s, len, enc);
diff --git a/symbol.h b/symbol.h
@@ -109,10 +109,9 @@ is_global_name_punct(const int c)
     return (ruby_global_name_punct_bits[(c - 0x20) / 32] >> (c % 32)) & 1;
 }
 
-int rb_enc_symname_type(const char *name, long len, rb_encoding *enc, unsigned int allowed_attrset);
-
 RUBY_SYMBOL_EXPORT_BEGIN
 
+int rb_enc_symname_type(const char *name, long len, rb_encoding *enc, unsigned int allowed_attrset);
 size_t rb_sym_immortal_count(void);
 
 RUBY_SYMBOL_EXPORT_END
diff --git a/test/ruby/test_parse.rb b/test/ruby/test_parse.rb
@@ -661,6 +661,7 @@ def test_symbol
     assert_syntax_error(':@@1', /is not allowed/)
     assert_syntax_error(':@', /is not allowed/)
     assert_syntax_error(':@1', /is not allowed/)
+    assert_syntax_error(':$01234', /is not allowed/)
   end
 
   def test_parse_string
@@ -1478,8 +1479,8 @@ def set(arg)
   end
 
   def test_ungettable_gvar
-    assert_syntax_error('$01234', /not valid to get/)
-    assert_syntax_error('"#$01234"', /not valid to get/)
+    assert_syntax_error('$01234', /not allowed/)
+    assert_syntax_error('"#$01234"', /not allowed/)
   end
 
 =begin

Original file line number	Diff line number	Diff line change
`@@ -109,10 +109,9 @@ is_global_name_punct(const int c)`
`109`	`109`	`return (ruby_global_name_punct_bits[(c - 0x20) / 32] >> (c % 32)) & 1;`
`110`	`110`	`}`
`111`	`111`
`112`		`-int rb_enc_symname_type(const char name, long len, rb_encoding enc, unsigned int allowed_attrset);`
`113`		`-`
`114`	`112`	`RUBY_SYMBOL_EXPORT_BEGIN`
`115`	`113`
	`114`	`+int rb_enc_symname_type(const char name, long len, rb_encoding enc, unsigned int allowed_attrset);`
`116`	`115`	`size_t rb_sym_immortal_count(void);`
`117`	`116`
`118`	`117`	`RUBY_SYMBOL_EXPORT_END`