fix UTF-32 valid_encoding?

* enc/utf_32be.c (utf32be_mbc_enc_len): check arguments precisely.
  [ruby-core:79966] [Bug #13292]

* enc/utf_32le.c (utf32le_mbc_enc_len): ditto.

* regenc.h (UNICODE_VALID_CODEPOINT_P): predicate for valid
  Unicode codepoints.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@57816 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

1 files changed, 68 insertions, 0 deletions

diff --git a/test/ruby/enc/test_utf32.rb b/test/ruby/enc/test_utf32.rb
index 23e8aeb315..76379abca0 100644
--- a/test/ruby/enc/test_utf32.rb
+++ b/test/ruby/enc/test_utf32.rb
@@ -90,5 +90,73 @@ EOT
     assert_equal(sl, "a".ord.chr("utf-32le"))
     assert_equal(sb, "a".ord.chr("utf-32be"))
   end
+
+  def test_utf32be_valid_encoding
+    all_assertions do |a|
+      [
+        "\x00\x00\x00\x00",
+        "\x00\x00\x00a",
+        "\x00\x00\x30\x40",
+        "\x00\x00\xd7\xff",
+        "\x00\x00\xe0\x00",
+        "\x00\x00\xff\xff",
+        "\x00\x10\xff\xff",
+      ].each {|s|
+        s.force_encoding("utf-32be")
+        a.for(s) {
+          assert_predicate(s, :valid_encoding?, "#{encdump s}.valid_encoding?")
+        }
+      }
+      [
+        "a",
+        "\x00a",
+        "\x00\x00a",
+        "\x00\x00\xd8\x00",
+        "\x00\x00\xdb\xff",
+        "\x00\x00\xdc\x00",
+        "\x00\x00\xdf\xff",
+        "\x00\x11\x00\x00",
+      ].each {|s|
+        s.force_encoding("utf-32be")
+        a.for(s) {
+          assert_not_predicate(s, :valid_encoding?, "#{encdump s}.valid_encoding?")
+        }
+      }
+    end
+  end
+
+  def test_utf32le_valid_encoding
+    all_assertions do |a|
+      [
+        "\x00\x00\x00\x00",
+        "a\x00\x00\x00",
+        "\x40\x30\x00\x00",
+        "\xff\xd7\x00\x00",
+        "\x00\xe0\x00\x00",
+        "\xff\xff\x00\x00",
+        "\xff\xff\x10\x00",
+      ].each {|s|
+        s.force_encoding("utf-32le")
+        a.for(s) {
+          assert_predicate(s, :valid_encoding?, "#{encdump s}.valid_encoding?")
+        }
+      }
+      [
+        "a",
+        "a\x00",
+        "a\x00\x00",
+        "\x00\xd8\x00\x00",
+        "\xff\xdb\x00\x00",
+        "\x00\xdc\x00\x00",
+        "\xff\xdf\x00\x00",
+        "\x00\x00\x11\x00",
+      ].each {|s|
+        s.force_encoding("utf-32le")
+        a.for(s) {
+          assert_not_predicate(s, :valid_encoding?, "#{encdump s}.valid_encoding?")
+        }
+      }
+    end
+  end
 end