Implement rb_enc_codelen

Author: bjfish

CHANGELOG.md

@@ -65,6 +65,7 @@ Compatibility:
 * Implemented `ONIGENC_MBC_CASE_FOLD`.
 * Fixed `Thread#raise` to call the exception class' constructor with no arguments when given no message (#2045).
 * Fixed `refine + super` compatibility (#2039, @ssnickolay)
+* Implemented `rb_enc_codelen`.
 
 Performance:
 

lib/cext/include/ruby/onigmo.h

@@ -369,7 +369,12 @@ int onigenc_mbclen_approximate(const OnigUChar* p,const OnigUChar* e, const stru
 #define ONIGENC_MBC_MINLEN(enc)               ((enc)->min_enc_len)
 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end)      (enc)->is_mbc_newline((p),(end),enc)
 #define ONIGENC_MBC_TO_CODE(enc,p,end)         (enc)->mbc_to_code((p),(end),enc)
+#ifdef TRUFFLERUBY
+int rb_tr_code_to_mbclen(OnigCodePoint code, OnigEncodingType *encoding);
+#define ONIGENC_CODE_TO_MBCLEN(enc,code)       rb_tr_code_to_mbclen(code,enc)
+#else
 #define ONIGENC_CODE_TO_MBCLEN(enc,code)       (enc)->code_to_mbclen(code,enc)
+#endif
 #define ONIGENC_CODE_TO_MBC(enc,code,buf)      (enc)->code_to_mbc(code,buf,enc)
 #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \
   (enc)->property_name_to_ctype(enc,p,end)

spec/ruby/optional/capi/encoding_spec.rb

@@ -58,6 +58,15 @@
     end
   end
 
+  describe "rb_enc_codelen" do
+    it "returns the correct length for the given codepoint" do
+      @s.rb_enc_codelen(0x24, Encoding::UTF_8).should == 1
+      @s.rb_enc_codelen(0xA2, Encoding::UTF_8).should == 2
+      @s.rb_enc_codelen(0x20AC, Encoding::UTF_8).should == 3
+      @s.rb_enc_codelen(0x24B62, Encoding::UTF_8).should == 4
+    end
+  end
+
   describe "rb_enc_find" do
     it "returns the encoding of an Encoding" do
       @s.rb_enc_find("UTF-8").should == "UTF-8"

spec/ruby/optional/capi/ext/encoding_spec.c

@@ -278,6 +278,12 @@ static VALUE encoding_spec_ONIGENC_MBC_CASE_FOLD(VALUE self, VALUE str) {
   return rb_ary_new3(2, str_result, INT2FIX(bytes_used));
 }
 
+static VALUE encoding_spec_rb_enc_codelen(VALUE self, VALUE code, VALUE encoding) {
+  unsigned int c = FIX2UINT(code);
+  rb_encoding *enc = rb_to_encoding(encoding);
+  return INT2FIX(rb_enc_codelen(c, enc));
+}
+
 void Init_encoding_spec(void) {
   VALUE cls;
   native_rb_encoding_pointer = (rb_encoding**) malloc(sizeof(rb_encoding*));
@@ -311,6 +317,7 @@ void Init_encoding_spec(void) {
   rb_define_method(cls, "rb_enc_associate_index", encoding_spec_rb_enc_associate_index, 2);
   rb_define_method(cls, "rb_enc_compatible", encoding_spec_rb_enc_compatible, 2);
   rb_define_method(cls, "rb_enc_copy", encoding_spec_rb_enc_copy, 2);
+  rb_define_method(cls, "rb_enc_codelen", encoding_spec_rb_enc_codelen, 2);
   rb_define_method(cls, "rb_enc_find", encoding_spec_rb_enc_find, 1);
   rb_define_method(cls, "rb_enc_find_index", encoding_spec_rb_enc_find_index, 1);
   rb_define_method(cls, "rb_enc_isalnum", encoding_spec_rb_enc_isalnum, 2);

src/main/c/cext/encoding.c

@@ -79,6 +79,18 @@ int rb_enc_mbc_to_codepoint(char *p, char *e, rb_encoding *enc) {
       length));
 }
 
+int rb_tr_code_to_mbclen(OnigCodePoint code, OnigEncodingType *encoding) {
+  return polyglot_as_i32(polyglot_invoke(RUBY_CEXT, "code_to_mbclen", code, rb_tr_unwrap(rb_enc_from_encoding(encoding))));
+}
+
+int rb_enc_codelen(int c, rb_encoding *enc) {
+  int n = ONIGENC_CODE_TO_MBCLEN(enc,c);
+  if (n == 0) {
+    rb_raise(rb_eArgError, "invalid codepoint 0x%x in %s", c, rb_enc_name(enc));
+  }
+  return n;
+}
+
 rb_encoding* rb_enc_get(VALUE object) {
   return rb_to_encoding(RUBY_CEXT_INVOKE("rb_enc_get", object));
 }

src/main/java/org/truffleruby/cext/CExtNodes.java

@@ -522,6 +522,16 @@ protected DynamicObject clearCodeRange(DynamicObject string,
 
     }
 
+    @CoreMethod(names = "code_to_mbclen", onSingleton = true, required = 2, lowerFixnum = 1)
+    public abstract static class CodeToMbcLenNode extends CoreMethodArrayArgumentsNode {
+
+        @Specialization
+        protected int codeToMbcLen(int code, DynamicObject encoding) {
+            return EncodingOperations.getEncoding(encoding).codeToMbcLength(code);
+        }
+
+    }
+
     @CoreMethod(names = "rb_enc_codepoint_len", onSingleton = true, required = 2)
     public abstract static class RbEncCodePointLenNode extends CoreMethodArrayArgumentsNode {