* include/ruby/intern.h (rb_str_buf_cat_ascii): declared.

* string.c (rb_str_buf_cat_ascii): defined.

* re.c (rb_reg_s_union): use rb_str_buf_cat_ascii to support ASCII
  incompatible encoding.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@15232 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,12 @@
+Fri Jan 25 16:31:47 2008  Tanaka Akira  <akr@fsij.org>
+
+	* include/ruby/intern.h (rb_str_buf_cat_ascii): declared.
+
+	* string.c (rb_str_buf_cat_ascii): defined.
+
+	* re.c (rb_reg_s_union): use rb_str_buf_cat_ascii to support ASCII
+	  incompatible encoding.
+
 Fri Jan 25 16:11:16 2008  Nobuyoshi Nakada  <nobu@ruby-lang.org>
 
 	* ruby.c (process_options, load_file, rb_load_file): propagates script

include/ruby/intern.h

@@ -504,6 +504,7 @@ void rb_str_shared_replace(VALUE, VALUE);
 VALUE rb_str_buf_append(VALUE, VALUE);
 VALUE rb_str_buf_cat(VALUE, const char*, long);
 VALUE rb_str_buf_cat2(VALUE, const char*);
+VALUE rb_str_buf_cat_ascii(VALUE, const char*);
 VALUE rb_obj_as_string(VALUE);
 VALUE rb_check_string_type(VALUE);
 VALUE rb_str_dup(VALUE);

re.c

@@ -2668,7 +2668,7 @@ rb_reg_s_union(VALUE self, VALUE args0)
 	    VALUE e = rb_ary_entry(args0, i);
 
 	    if (0 < i)
-		rb_str_buf_cat2(source, "|"); /* xxx: UTF-16 */
+		rb_str_buf_cat_ascii(source, "|");
 
 	    v = rb_check_regexp_type(e);
 	    if (!NIL_P(v)) {
@@ -2726,6 +2726,9 @@ rb_reg_s_union(VALUE self, VALUE args0)
                 }
             }
 
+            if (i == 0) {
+                rb_enc_copy(source, v);
+            }
 	    rb_str_append(source, v);
 	}
 

string.c

@@ -1116,6 +1116,26 @@ rb_str_cat2(VALUE str, const char *ptr)
     return rb_str_cat(str, ptr, strlen(ptr));
 }
 
+VALUE
+rb_str_buf_cat_ascii(VALUE str, const char *ptr)
+{
+    rb_encoding *enc = rb_enc_get(str);
+    if (rb_enc_asciicompat(enc)) {
+        return rb_str_buf_cat(str, ptr, strlen(ptr));
+    }
+    else {
+        char *buf = ALLOCA_N(char, rb_enc_mbmaxlen(enc));
+        while (*ptr) {
+            int c = (unsigned char)*ptr;
+            int len = rb_enc_codelen(c, enc);
+            rb_enc_mbcput(c, buf, enc);
+            rb_str_buf_cat(str, buf, len);
+            ptr++;
+        }
+        return str;
+    }
+}
+
 static VALUE
 rb_enc_cr_str_buf_cat(VALUE str, const char *ptr, long len,
     int ptr_encindex, int ptr_cr, int *ptr_cr_ret)