* string.c (rb_str_sub_bang): fix coderange.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19662 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,7 @@
+Thu Oct  2 20:46:17 2008  Tanaka Akira  <akr@fsij.org>
+
+	* string.c (rb_str_sub_bang): fix coderange.
+
 Wed Oct  1 22:54:09 2008  Yusuke Endoh  <mame@tsg.ne.jp>
 
 	* lib/rexml/parsers/lightparser.rb, lib/debug.rb: remove unused when

string.c

@@ -3307,7 +3307,11 @@ rb_str_sub_bang(int argc, VALUE *argv, VALUE str)
 	if (OBJ_UNTRUSTED(repl)) untrusted = 1;
 	if (ENC_CODERANGE_UNKNOWN < cr && cr < ENC_CODERANGE_BROKEN) {
 	    int cr2 = ENC_CODERANGE(repl);
-	    if (cr2 == ENC_CODERANGE_UNKNOWN || cr2 > cr) cr = cr2;
+            if (cr2 == ENC_CODERANGE_BROKEN ||
+                (cr == ENC_CODERANGE_VALID && cr2 == ENC_CODERANGE_7BIT))
+                cr = ENC_CODERANGE_UNKNOWN;
+            else
+                cr = cr2;
 	}
 	plen = end0 - beg0;
 	if (RSTRING_LEN(repl) > plen) {

test/ruby/test_m17n.rb

@@ -959,6 +959,21 @@ class TestM17N < Test::Unit::TestCase
     assert_equal(Encoding::EUC_JP, "\xa4\xa2".force_encoding("euc-jp").gsub(/./, '\&').encoding)
   end
 
+  def test_sub2
+    s = "\x80".force_encoding("ASCII-8BIT")     
+    r = Regexp.new("\x80".force_encoding("ASCII-8BIT")) 
+    s2 = s.sub(r, "")
+    assert(s2.empty?)
+    assert(s2.ascii_only?)
+  end
+
+  def test_sub3
+    repl = "\x81".force_encoding("sjis")
+    assert_equal(false, repl.valid_encoding?)
+    s = "a@".sub(/a/, repl)
+    assert(s.valid_encoding?)
+  end
+
   def test_insert
     s = e("\xa3\xb0\xa3\xb1\xa3\xb2\xa3\xb3\xa3\xb4")
     assert_equal(e("\xa3\xb0\xa3\xb1\xa3\xb2\xa3\xb3\xa3\xb4a"), s.insert(-1, "a"))

test/ruby/test_m17n_comb.rb

@@ -115,6 +115,7 @@ class TestM17NComb < Test::Unit::TestCase
 
   def combination(*args, &b)
     AllPairs.each(*args, &b)
+    #AllPairs.exhaustive_each(*args, &b)
   end
 
   def encdump(str)
@@ -1395,21 +1396,25 @@ class TestM17NComb < Test::Unit::TestCase
       [
         [
           "#{encdump s1}.sub(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { s1.sub(r2, s3) }
+          lambda { s1.sub(r2, s3) },
+          false
         ],
         [
           "#{encdump s1}.sub(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { s1.sub(r2) { s3 } }
+          lambda { s1.sub(r2) { s3 } },
+          false
         ],
         [
           "#{encdump s1}.gsub(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { s1.gsub(r2, s3) }
+          lambda { s1.gsub(r2, s3) },
+          true
         ],
         [
           "#{encdump s1}.gsub(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { s1.gsub(r2) { s3 } }
+          lambda { s1.gsub(r2) { s3 } },
+          true
         ]
-      ].each {|desc, doit|
+      ].each {|desc, doit, g|
         if !s1.valid_encoding?
           assert_raise(ArgumentError, desc) { doit.call }
           next
@@ -1422,7 +1427,7 @@ class TestM17NComb < Test::Unit::TestCase
           assert_equal(s1, doit.call)
           next
         end
-        if !str_enc_compatible?(s1.gsub(r2, ''), s3)
+        if !str_enc_compatible?(g ? s1.gsub(r2, '') : s1.sub(r2, ''), s3)
           assert_raise(Encoding::CompatibilityError, desc) { doit.call }
           next
         end
@@ -1449,21 +1454,25 @@ class TestM17NComb < Test::Unit::TestCase
       [
         [
           "t=#{encdump s1}.dup;t.sub!(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { t=s1.dup; [t, t.sub!(r2, s3)] }
+          lambda { t=s1.dup; [t, t.sub!(r2, s3)] },
+          false
         ],
         [
           "t=#{encdump s1}.dup;t.sub!(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { t=s1.dup; [t, t.sub!(r2) { s3 }] }
+          lambda { t=s1.dup; [t, t.sub!(r2) { s3 }] },
+          false
         ],
         [
           "t=#{encdump s1}.dup;t.gsub!(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { t=s1.dup; [t, t.gsub!(r2, s3)] }
+          lambda { t=s1.dup; [t, t.gsub!(r2, s3)] },
+          true
         ],
         [
           "t=#{encdump s1}.dup;t.gsub!(Regexp.new(#{encdump s2}), #{encdump s3})",
-          lambda { t=s1.dup; [t, t.gsub!(r2) { s3 }] }
+          lambda { t=s1.dup; [t, t.gsub!(r2) { s3 }] },
+          true
         ]
-      ].each {|desc, doit|
+      ].each {|desc, doit, g|
         if !s1.valid_encoding?
           assert_raise(ArgumentError, desc) { doit.call }
           next
@@ -1476,7 +1485,7 @@ class TestM17NComb < Test::Unit::TestCase
           assert_equal([s1, nil], doit.call)
           next
         end
-        if !str_enc_compatible?(s1.gsub(r2, ''), s3)
+        if !str_enc_compatible?(g ? s1.gsub(r2, '') : s1.sub(r2, ''), s3)
           assert_raise(Encoding::CompatibilityError, desc) { doit.call }
           next
         end