merges r22505 and r22547 from trunk into ruby_1_9_1.

--
* string.c (tr_trans): should not be affected by the encoding of
  replacement unless actually modified.  [ruby-talk:328967]
--
* string.c (tr_trans): should recalculate coderange.
  [ruby-core:22326] (reopened at [ruby-core:22328])

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_1_9_1@22731 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

3 files changed, 51 insertions, 8 deletions

diff --git a/ChangeLog b/ChangeLog
index eb480282c2..7fb56be290 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,13 @@
+Sun Feb 22 22:42:20 2009  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+	* string.c (tr_trans): should recalculate coderange.
+	  [ruby-core:22326] (reopened at [ruby-core:22328])
+
+Sun Feb 22 14:33:06 2009  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+	* string.c (tr_trans): should not be affected by the encoding of
+	  replacement unless actually modified.  [ruby-talk:328967]
+
 Fri Feb  6 12:11:24 2009  NAKAMURA Usaku  <usa@ruby-lang.org>
 
 	* ruby.c (process_options): set initial default_external before -r.
diff --git a/string.c b/string.c
index 432fdc4c79..1a8b182ecc 100644
--- a/string.c
+++ b/string.c
@@ -4603,6 +4603,10 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
     int singlebyte = single_byte_optimizable(str);
     int cr;
 
+#define CHECK_IF_ASCII(c) \
+    (void)((cr == ENC_CODERANGE_7BIT && !rb_isascii(c)) ? \
+	   (cr = ENC_CODERANGE_VALID) : 0)
+
     StringValue(src);
     StringValue(repl);
     if (RSTRING_LEN(str) == 0 || !RSTRING_PTR(str)) return Qnil;
@@ -4674,6 +4678,8 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 	}
     }
 
+    if (cr == ENC_CODERANGE_VALID)
+	cr = ENC_CODERANGE_7BIT;
     str_modify_keep_cr(str);
     s = RSTRING_PTR(str); send = RSTRING_END(str);
     if (sflag) {
@@ -4682,8 +4688,10 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 	char *buf = ALLOC_N(char, max), *t = buf;
 
 	while (s < send) {
-	    c0 = c = rb_enc_codepoint(s, send, enc);
-	    tlen = clen = rb_enc_codelen(c, enc);
+	    int may_modify = 0;
+	    c0 = c = rb_enc_codepoint(s, send, e1);
+	    clen = rb_enc_codelen(c, e1);
+	    tlen = enc == e1 ? clen : rb_enc_codelen(c, enc);
 
 	    s += clen;
 	    if (c < 256) {
@@ -4702,7 +4710,10 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 		c = errc;
 	    }
 	    if (c != -1) {
-		if (save == c) continue;
+		if (save == c) {
+		    CHECK_IF_ASCII(c);
+		    continue;
+		}
 		save = c;
 		tlen = rb_enc_codelen(c, enc);
 		modify = 1;
@@ -4710,6 +4721,7 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 	    else {
 		save = -1;
 		c = c0;
+		if (enc != e1) may_modify = 1;
 	    }
 	    while (t - buf + tlen >= max) {
 		offset = t - buf;
@@ -4718,6 +4730,10 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 		t = buf + offset;
 	    }
 	    rb_enc_mbcput(c, t, enc);
+	    if (may_modify && memcmp(s, t, tlen) != 0) {
+		modify = 1;
+	    }
+	    CHECK_IF_ASCII(c);
 	    t += tlen;
 	}
 	*t = '\0';
@@ -4740,6 +4756,7 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 		    modify = 1;
 		}
 	    }
+	    CHECK_IF_ASCII(c);
 	    s++;
 	}
     }
@@ -4749,8 +4766,10 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 	char *buf = ALLOC_N(char, max), *t = buf;
 
 	while (s < send) {
-	    c0 = c = rb_enc_codepoint(s, send, enc);
-	    tlen = clen = rb_enc_codelen(c, enc);
+	    int may_modify = 0;
+	    c0 = c = rb_enc_codepoint(s, send, e1);
+	    clen = rb_enc_codelen(c, e1);
+	    tlen = enc == e1 ? clen : rb_enc_codelen(c, enc);
 
 	    if (c < 256) {
 		c = trans[c];
@@ -4772,8 +4791,8 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 		modify = 1;
 	    }
 	    else {
-		modify = 1;
 		c = c0;
+		if (enc != e1) may_modify = 1;
 	    }
 	    while (t - buf + tlen >= max) {
 		offset = t - buf;
@@ -4781,7 +4800,13 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
 		REALLOC_N(buf, char, max);
 		t = buf + offset;
 	    }
-	    if (s != t) rb_enc_mbcput(c, t, enc);
+	    if (s != t) {
+		rb_enc_mbcput(c, t, enc);
+		if (may_modify && memcmp(s, t, tlen) != 0) {
+		    modify = 1;
+		}
+	    }
+	    CHECK_IF_ASCII(c);
 	    s += clen;
 	    t += tlen;
 	}
@@ -4796,7 +4821,6 @@ tr_trans(VALUE str, VALUE src, VALUE repl, int sflag)
     }
     
     if (modify) {
-	cr = ENC_CODERANGE_AND(cr, ENC_CODERANGE(repl));
 	if (cr != ENC_CODERANGE_BROKEN)
 	    ENC_CODERANGE_SET(str, cr);
 	rb_enc_associate(str, enc);
diff --git a/test/ruby/test_string.rb b/test/ruby/test_string.rb
index eba0256439..2315a8c560 100644
--- a/test/ruby/test_string.rb
+++ b/test/ruby/test_string.rb
@@ -1395,6 +1395,9 @@ class TestString < Test::Unit::TestCase
     assert_equal(S("hippo"), S("hello").tr(S("el"), S("ip")))
     assert_equal(S("*e**o"), S("hello").tr(S("^aeiou"), S("*")))
     assert_equal(S("hal"),   S("ibm").tr(S("b-z"), S("a-z")))
+
+    a = "abc".force_encoding(Encoding::US_ASCII)
+    assert_equal(Encoding::US_ASCII, a.tr(S("z"), S("\u0101")).encoding)
   end
 
   def test_tr!
@@ -1415,11 +1418,17 @@ class TestString < Test::Unit::TestCase
     a = S("ibm")
     assert_nil(a.tr!(S("B-Z"), S("A-Z")))
     assert_equal(S("ibm"), a)
+
+    a = "abc".force_encoding(Encoding::US_ASCII)
+    assert_nil(a.tr!(S("z"), S("\u0101")))
+    assert_equal(Encoding::US_ASCII, a.encoding)
   end
 
   def test_tr_s
     assert_equal(S("hypo"), S("hello").tr_s(S("el"), S("yp")))
     assert_equal(S("h*o"),  S("hello").tr_s(S("el"), S("*")))
+    assert_equal("a".hash, "\u0101\u0101".tr_s("\u0101", "a").hash)
+    assert_equal(true, "\u3041\u3041".tr("\u3041", "a").ascii_only?)
   end
 
   def test_tr_s!