[GR-20329] Implement rb_enc_mbcput.

PullRequest: truffleruby/1782

Author: eregon

lib/cext/include/ruby/onigmo.h

@@ -373,10 +373,12 @@ int onigenc_mbclen_approximate(const OnigUChar* p,const OnigUChar* e, const stru
 #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)
+int rb_tr_code_to_mbc(OnigCodePoint code, UChar *buf, OnigEncoding enc);
+#define ONIGENC_CODE_TO_MBC(enc,code,buf)      rb_tr_code_to_mbc(code,buf,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)
+#endif
 #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \
   (enc)->property_name_to_ctype(enc,p,end)
 

spec/ruby/optional/capi/encoding_spec.rb

@@ -139,6 +139,18 @@
     end
   end
 
+  describe "rb_enc_mbcput" do
+    it "writes the correct bytes to the buffer" do
+      @s.rb_enc_mbcput(0x24, Encoding::UTF_8).should == "$"
+      @s.rb_enc_mbcput(0xA2, Encoding::UTF_8).should == "¢"
+      @s.rb_enc_mbcput(0x20AC, Encoding::UTF_8).should == "€"
+      @s.rb_enc_mbcput(0x24B62, Encoding::UTF_8).should == "𤭢"
+
+      @s.rb_enc_mbcput(0x24, Encoding::UTF_16BE).bytes.should == [0, 0x24]
+      @s.rb_enc_mbcput(0x24B62, Encoding::UTF_16LE).bytes.should == [82, 216, 98, 223]
+    end
+  end
+
   describe "rb_usascii_encoding" do
     it "returns the encoding for Encoding::US_ASCII" do
       @s.rb_usascii_encoding.should == "US-ASCII"
@@ -630,6 +642,10 @@
       @s.ONIGENC_MBC_CASE_FOLD("lower".force_encoding("binary")).should == ["l", 1]
       @s.ONIGENC_MBC_CASE_FOLD("Upper".force_encoding("binary")).should == ["u", 1]
       @s.ONIGENC_MBC_CASE_FOLD("É").should == ["é", 2]
+
+      str, length = @s.ONIGENC_MBC_CASE_FOLD('$'.encode(Encoding::UTF_16BE))
+      length.should == 2
+      str.bytes.should == [0, 0x24]
     end
   end
 end

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

@@ -127,6 +127,18 @@ static VALUE encoding_spec_rb_enc_mbc_to_codepoint(VALUE self, VALUE str, VALUE
   return INT2FIX(rb_enc_mbc_to_codepoint(p, e, rb_enc_get(str)));
 }
 
+static VALUE encoding_spec_rb_enc_mbcput(VALUE self, VALUE code, VALUE encoding) {
+  unsigned int c = FIX2UINT(code);
+  rb_encoding *enc = rb_to_encoding(encoding);
+  char buf[ONIGENC_CODE_TO_MBC_MAXLEN];
+  memset(buf, '\1', sizeof(buf));
+  int len = rb_enc_mbcput(c, buf, enc);
+  if (buf[len] != '\1') {
+    rb_raise(rb_eRuntimeError, "should not change bytes after len");
+  }
+  return rb_enc_str_new(buf, len, enc);
+}
+
 static VALUE encoding_spec_rb_enc_from_encoding(VALUE self, VALUE name) {
   return rb_enc_from_encoding(rb_enc_find(RSTRING_PTR(name)));
 }
@@ -271,8 +283,12 @@ static VALUE encoding_spec_ONIGENC_MBC_CASE_FOLD(VALUE self, VALUE str) {
   char *beg_initial = beg;
   char *end = beg + 2;
   OnigUChar fold[ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM];
+  memset(fold, '\1', sizeof(fold));
   rb_encoding *enc = rb_enc_get(str);
   int r = ONIGENC_MBC_CASE_FOLD(enc, ONIGENC_CASE_FOLD, &beg, (const OnigUChar *)end, fold);
+  if (r > 0 && fold[r] != '\1') {
+    rb_raise(rb_eRuntimeError, "should not change bytes after len");
+  }
   VALUE str_result = r <= 0 ? Qnil : rb_enc_str_new((char *)fold, r, enc);
   long bytes_used = beg - beg_initial;
   return rb_ary_new3(2, str_result, INT2FIX(bytes_used));
@@ -324,6 +340,7 @@ void Init_encoding_spec(void) {
   rb_define_method(cls, "rb_enc_isspace", encoding_spec_rb_enc_isspace, 2);
   rb_define_method(cls, "rb_enc_from_index", encoding_spec_rb_enc_from_index, 1);
   rb_define_method(cls, "rb_enc_mbc_to_codepoint", encoding_spec_rb_enc_mbc_to_codepoint, 2);
+  rb_define_method(cls, "rb_enc_mbcput", encoding_spec_rb_enc_mbcput, 2);
   rb_define_method(cls, "rb_enc_from_encoding", encoding_spec_rb_enc_from_encoding, 1);
   rb_define_method(cls, "rb_enc_get", encoding_spec_rb_enc_get, 1);
   rb_define_method(cls, "rb_enc_precise_mbclen", encoding_spec_rb_enc_precise_mbclen, 2);

spec/ruby/optional/capi/spec_helper.rb

@@ -76,6 +76,7 @@ def compile_extension(name)
           $ruby = ENV.values_at('RUBY_EXE', 'RUBY_FLAGS').join(' ')
           # MRI magic to consider building non-bundled extensions
           $extout = nil
+          $warnflags << ' -Wno-declaration-after-statement'
           create_makefile(#{ext.inspect})
         RUBY
         output = ruby_exe("extconf.rb")

src/main/c/cext/encoding.c

@@ -362,7 +362,17 @@ int rb_tr_enc_mbc_case_fold(rb_encoding *enc, int flag, const UChar** p, const U
           p));
    int result_len = RSTRING_LEN(result_str);
    if (result_len > 0) {
-     strncpy((char *)result, RSTRING_PTR(result_str), result_len);
+     memcpy(result, RSTRING_PTR(result_str), result_len);
+   }
+   return result_len;
+}
+
+int rb_tr_code_to_mbc(OnigCodePoint code, UChar *buf, OnigEncoding enc) {
+    VALUE result_str = rb_tr_wrap(polyglot_invoke(RUBY_CEXT, "rb_tr_code_to_mbc",
+            rb_tr_unwrap(rb_enc_from_encoding(enc)), code));
+   int result_len = RSTRING_LEN(result_str);
+   if (result_len > 0) {
+     memcpy(buf, RSTRING_PTR(result_str), result_len);
    }
    return result_len;
 }

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

@@ -1310,6 +1310,25 @@ protected int getCacheLimit() {
 
     }
 
+    @CoreMethod(names = "rb_tr_code_to_mbc", onSingleton = true, required = 2, lowerFixnum = 2)
+    public abstract static class RbTrMbcPutNode extends CoreMethodArrayArgumentsNode {
+
+        @Specialization(guards = "isRubyEncoding(enc)")
+        protected Object rbTrEncMbcPut(DynamicObject enc, int code) {
+            final Encoding encoding = EncodingOperations.getEncoding(enc);
+            final byte buf[] = new byte[org.jcodings.Config.ENC_CODE_TO_MBC_MAXLEN];
+            final int resultLength = encoding.codeToMbc(code, buf, 0);
+            final byte result[] = new byte[resultLength];
+            if (resultLength > 0) {
+                System.arraycopy(buf, 0, result, 0, resultLength);
+            }
+            return StringOperations.createString(
+                    getContext(),
+                    RopeOperations.create(result, USASCIIEncoding.INSTANCE, CodeRange.CR_UNKNOWN));
+        }
+
+    }
+
     @CoreMethod(names = "rb_enc_mbmaxlen", onSingleton = true, required = 1)
     public abstract static class RbEncMaxLenNode extends CoreMethodArrayArgumentsNode {