* ext/-test-/string/coderange.c: add Bug::String.new#coderange_scan

to explicitly scan coderange.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@45649 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,8 @@
+Mon Apr 21 18:55:21 2014  NARUSE, Yui  <naruse@ruby-lang.org>
+
+	* ext/-test-/string/coderange.c: add Bug::String.new#coderange_scan
+	  to explicitly scan coderange.
+
 Mon Apr 21 18:19:35 2014  NARUSE, Yui  <naruse@ruby-lang.org>
 
 	* string.c (coderange_scan): remove unused logic.

ext/-test-/string/coderange.c

@@ -2,10 +2,11 @@
 #include "ruby/encoding.h"
 
 static VALUE sym_7bit, sym_valid, sym_unknown, sym_broken;
+
 static VALUE
-str_coderange(VALUE str)
+coderange_int2sym(int coderange)
 {
-    switch (ENC_CODERANGE(str)) {
+    switch (coderange) {
       case ENC_CODERANGE_7BIT:
 	return sym_7bit;
       case ENC_CODERANGE_VALID:
@@ -19,6 +20,21 @@ str_coderange(VALUE str)
     UNREACHABLE;
 }
 
+/* return coderange without scan */
+static VALUE
+str_coderange(VALUE str)
+{
+    return coderange_int2sym(ENC_CODERANGE(str));
+}
+
+/* scan coderange and return the result */
+static VALUE
+str_coderange_scan(VALUE str)
+{
+    ENC_CODERANGE_SET(str, ENC_CODERANGE_UNKNOWN);
+    return coderange_int2sym(rb_enc_str_coderange(str));
+}
+
 void
 Init_coderange(VALUE klass)
 {
@@ -27,4 +43,5 @@ Init_coderange(VALUE klass)
     sym_unknown = ID2SYM(rb_intern("unknown"));
     sym_broken = ID2SYM(rb_intern("broken"));
     rb_define_method(klass, "coderange", str_coderange, 0);
+    rb_define_method(klass, "coderange_scan", str_coderange_scan, 0);
 }

test/-ext-/string/test_coderange.rb

@@ -0,0 +1,59 @@
+# coding: ascii-8bit
+require 'test/unit'
+require "-test-/string/string"
+require "rbconfig/sizeof"
+
+class Test_StringCoderange < Test::Unit::TestCase
+  def setup
+    @sizeof_voidp = RbConfig::SIZEOF["void*"]
+    @a8 = Encoding::ASCII_8BIT
+    @a7 = Encoding::US_ASCII
+    @u8 = Encoding::UTF_8
+  end
+
+  def test_ascii8bit
+    enc = @a8
+    str = "a"
+    str.force_encoding(enc)
+    assert_equal :"7bit", Bug::String.new(str).coderange_scan
+
+    str = "a\xBE".force_encoding(enc)
+    assert_equal :valid, Bug::String.new(str).coderange_scan
+  end
+
+  def test_usascii
+    enc = @a7
+    str = "a"
+    str.force_encoding(enc)
+    assert_equal :"7bit", Bug::String.new(str).coderange_scan
+
+    str = "a" * (@sizeof_voidp * 2)
+    str << "\xBE"
+    str.force_encoding(enc)
+    assert_equal :broken, Bug::String.new(str).coderange_scan
+  end
+
+  def test_utf8
+    enc = @u8
+    str = "a"
+    str.force_encoding(enc)
+    assert_equal :"7bit", Bug::String.new(str).coderange_scan
+
+    str = "a" * (@sizeof_voidp * 3)
+    str << "aa\xC2\x80"
+    str.force_encoding(enc)
+    assert_equal :valid, Bug::String.new(str).coderange_scan
+
+    str = "a" * (@sizeof_voidp * 2)
+    str << "\xC2\x80"
+    str << "a" * (@sizeof_voidp * 2)
+    str.force_encoding(enc)
+    assert_equal :valid, Bug::String.new(str).coderange_scan
+
+    str = "a" * (@sizeof_voidp * 2)
+    str << "\xC1\x80"
+    str << "a" * (@sizeof_voidp * 2)
+    str.force_encoding(enc)
+    assert_equal :broken, Bug::String.new(str).coderange_scan
+  end
+end