* re.c (rb_reg_initialize_str): forbid raw non ASCII character

for ASCII-8BIT regexp in non ASCII-8BIT script.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@14911 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,8 @@
+Sun Jan  6 21:14:12 2008  Tanaka Akira  <akr@fsij.org>
+
+	* re.c (rb_reg_initialize_str): forbid raw non ASCII character
+	  for ASCII-8BIT regexp in non ASCII-8BIT script.
+
 Sun Jan  6 18:19:12 2008  Tanaka Akira  <akr@fsij.org>
 
 	* include/ruby/encoding.h (rb_enc_str_buf_cat): declared.

re.c

@@ -2037,8 +2037,16 @@ rb_reg_initialize_str(VALUE obj, VALUE str, int options, onig_errmsg_buffer err)
 {
     int ret;
     rb_encoding *enc = rb_enc_get(str);
-    if (options & ARG_ENCODING_NONE)
-        enc = rb_ascii8bit_encoding();
+    if (options & ARG_ENCODING_NONE) {
+        rb_encoding *ascii8bit = rb_ascii8bit_encoding();
+        if (enc != ascii8bit) {
+            if (rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) {
+                strcpy(err, "/.../n has a non escaped non ASCII character in non ASCII-8BIT script");
+                return -1;
+            }
+            enc = ascii8bit;
+        }
+    }
     ret = rb_reg_initialize(obj, RSTRING_PTR(str), RSTRING_LEN(str), enc,
 			    options, err);
     RB_GC_GUARD(str);

test/ruby/test_m17n.rb

@@ -320,6 +320,8 @@ class TestM17N < Test::Unit::TestCase
     assert_regexp_fixed_ascii8bit(eval(a(%{/\xc2\xa1/n})))
     assert_regexp_fixed_ascii8bit(eval(a(%q{/\xc2\xa1/})))
 
+    assert_raise(SyntaxError) { eval("/\xa1\xa1/n".force_encoding("euc-jp")) }
+
     [/\xc2\xa1/n, eval(a(%{/\xc2\xa1/})), eval(a(%{/\xc2\xa1/n}))].each {|r|
       assert_equal(nil, r =~ a("a"))
       assert_equal(nil, r =~ e("a"))