jruby · Oct 7, 2015 · Oct 7, 2015 · Oct 7, 2015
Showing with 15 additions and 18 deletions.

+1 −14 core/src/main/java/org/jruby/RubyString.java

+13 −0 spec/regression/GH-3267_utf8_regexp_string_match_compiles_as_utf8_spec.rb

+1 −2 test/mri/excludes/TestRegexp.rb

+0 −2 test/mri/excludes/TestString.rb
diff --git a/core/src/main/java/org/jruby/RubyString.java b/core/src/main/java/org/jruby/RubyString.java
@@ -3573,20 +3573,7 @@ private RubyString getStringForPattern(IRubyObject obj) {
      */
     private RubyRegexp getPattern(IRubyObject obj) {
         if (obj instanceof RubyRegexp) return (RubyRegexp)obj;
-        return RubyRegexp.newRegexp(getRuntime(), getStringForPattern(obj).value);
-    }
-
-    private Regex getStringPattern19(Ruby runtime, IRubyObject obj) {
-        RubyString str = getStringForPattern(obj);
-        if (str.scanForCodeRange() == CR_BROKEN) {
-            throw runtime.newRegexpError("invalid multybyte character: " +
-                    RegexpSupport.regexpDescription19(runtime, str.value, new RegexpOptions(), str.value.getEncoding()).toString());
-        }
-        if (str.value.getEncoding().isDummy()) {
-            throw runtime.newArgumentError("can't make regexp with dummy encoding");
-        }
-
-        return RubyRegexp.getQuotedRegexpFromCache19(runtime, str.value, new RegexpOptions(), str.isAsciiOnly());
+        return RubyRegexp.newRegexpFromStr(getRuntime(), getStringForPattern(obj), 0);
     }
 
     // MRI: get_pat_quoted

diff --git a/spec/regression/GH-3267_utf8_regexp_string_match_compiles_as_utf8_spec.rb b/spec/regression/GH-3267_utf8_regexp_string_match_compiles_as_utf8_spec.rb
@@ -0,0 +1,13 @@
+# jruby/jruby#3267
+# String passed to String#match was not following proper channels to compile to Regexp.
+# As a result, encoding was not negotiated properly.
+describe "A UTF-8 string matched against a US-ASCII string" do
+  it "compiles to regexp successfully" do
+    result = nil
+    expect(
+        lambda {result = "".force_encoding('US-ASCII').match("Període\\ de\\ retorn".force_encoding('UTF-8'))}
+    ).not_to raise_error
+
+    expect(result).to eq(nil)
+  end
+end
diff --git a/test/mri/excludes/TestRegexp.rb b/test/mri/excludes/TestRegexp.rb
@@ -1,12 +1,11 @@
 exclude :test_char_class, "needs investigation"
 exclude :test_dup_warn, "needs investigation"
-exclude :test_eq_tilde_can_be_overridden, "needs investigation"
 exclude :test_equal, "needs investigation"
 exclude :test_initialize, "needs investigation"
 exclude :test_invalid_escape_error, "needs investigation"
 exclude :test_invalid_fragment, "needs investigation"
 exclude :test_named_capture_nonascii, "needs investigation"
-exclude :test_once_multithread, "needs investigation"
+exclude :test_once_multithread, "fails intermittently"
 exclude :test_options_in_look_behind, "needs investigation"
 exclude :test_raw_hyphen_and_tk_char_type_after_range, "needs investigation"
 exclude :test_regsub_K, "needs investigation"

diff --git a/test/mri/excludes/TestString.rb b/test/mri/excludes/TestString.rb
@@ -1,7 +1,5 @@
 exclude :test_count, "needs investigation"
-exclude :test_eq_tilde_can_be_overridden, "needs investigation"
 exclude :test_partition, "needs investigation"
-exclude :test_regexp_match_subclass, "String subclass with overridden =~ should see obj =~ dispatch (#2157)"
 exclude :test_rpartition, "needs investigation"
 exclude :test_rstrip, "needs investigation"
 exclude :test_to_i, "needs investigation"