Implemented CAPI rb_enc_codepoint_len.

rubinius · Jun 3, 2013 · 08bff7d · 08bff7d
1 parent b3579cf
commit 08bff7d
Show file tree

Hide file tree

Showing 5 changed files with 78 additions and 0 deletions.
diff --git a/spec/ruby/optional/capi/encoding_spec.rb b/spec/ruby/optional/capi/encoding_spec.rb
@@ -429,5 +429,46 @@
       end
     end
 
+    describe "rb_enc_codepoint_len" do
+      it "raises ArgumentError if an empty string is given" do
+        lambda do
+          @s.rb_enc_codepoint_len("")
+        end.should raise_error(ArgumentError)
+      end
+
+      it "raises ArgumentError if an invalid byte sequence is given" do
+        lambda do
+          @s.rb_enc_codepoint_len("\xa0\xa1") # Invalid sequence identifier
+        end.should raise_error(ArgumentError)
+      end
+
+      it "returns codepoint 0x24 and length 1 for character '$'" do
+        codepoint, length = @s.rb_enc_codepoint_len("$")
+
+        codepoint.should == 0x24
+        length.should == 1
+      end
+
+      it "returns codepoint 0xA2 and length 2 for character '¢'" do
+        codepoint, length = @s.rb_enc_codepoint_len("¢")
+
+        codepoint.should == 0xA2
+        length.should == 2
+      end
+
+      it "returns codepoint 0x20AC and length 3 for character '€'" do
+        codepoint, length = @s.rb_enc_codepoint_len("€")
+
+        codepoint.should == 0x20AC
+        length.should == 3
+      end
+
+      it "returns codepoint 0x24B62 and length 4 for character '𤭢'" do
+        codepoint, length = @s.rb_enc_codepoint_len("𤭢")
+
+        codepoint.should == 0x24B62
+        length.should == 4
+      end
+    end
   end
 end
diff --git a/spec/ruby/optional/capi/ext/encoding_spec.c b/spec/ruby/optional/capi/ext/encoding_spec.c
@@ -257,6 +257,18 @@ static VALUE encoding_spec_rb_enc_nth(VALUE self, VALUE str, VALUE index) {
 }
 #endif
 
+#ifdef HAVE_RB_ENC_CODEPOINT_LEN
+static VALUE encoding_spec_rb_enc_codepoint_len(VALUE self, VALUE str) {
+  char* start = RSTRING_PTR(str);
+  char* end = start + RSTRING_LEN(str);
+
+  int len;
+  unsigned int codepoint = rb_enc_codepoint_len(start, end, &len, rb_enc_get(str));
+
+  return rb_ary_new3(2, LONG2NUM(codepoint), LONG2NUM(len));
+}
+#endif
+
 void Init_encoding_spec() {
   VALUE cls;
   cls = rb_define_class("CApiEncodingSpecs", rb_cObject);
@@ -401,6 +413,10 @@ void Init_encoding_spec() {
 #ifdef HAVE_RB_ENC_NTH
   rb_define_method(cls, "rb_enc_nth", encoding_spec_rb_enc_nth, 2);
 #endif
+
+#ifdef HAVE_RB_ENC_CODEPOINT_LEN
+  rb_define_method(cls, "rb_enc_codepoint_len", encoding_spec_rb_enc_codepoint_len, 1);
+#endif
 }
 
 #ifdef __cplusplus

diff --git a/spec/ruby/optional/capi/ext/rubyspec.h b/spec/ruby/optional/capi/ext/rubyspec.h
@@ -191,6 +191,7 @@
 #define HAVE_RB_ENCDB_ALIAS                1
 #define HAVE_RB_ENC_ASSOCIATE              1
 #define HAVE_RB_ENC_ASSOCIATE_INDEX        1
+#define HAVE_RB_ENC_CODEPOINT_LEN          1
 #define HAVE_RB_ENC_COMPATIBLE             1
 #define HAVE_RB_ENC_COPY                   1
 #define HAVE_RB_ENC_FIND                   1

diff --git a/vm/capi/19/include/ruby/encoding.h b/vm/capi/19/include/ruby/encoding.h
@@ -89,6 +89,8 @@ rb_encoding* rb_to_encoding(VALUE obj);
 
 VALUE rb_enc_from_encoding(rb_encoding *enc);
 
+unsigned int rb_enc_codepoint_len(const char *p, const char *e, int *len_p, rb_encoding *enc);
+
 int rb_define_dummy_encoding(const char *);
 int rb_enc_find_index(const char *name);
 int rb_enc_to_index(rb_encoding* enc);

diff --git a/vm/capi/encoding.cpp b/vm/capi/encoding.cpp
@@ -398,4 +398,22 @@ extern "C" {
 
     Transcoding::define(env->state(), (OnigTranscodingType*)trans);
   }
+
+  unsigned int rb_enc_codepoint_len(const char *p, const char *e, int *len_p, rb_encoding *enc) {
+    if(e <= p) {
+      rb_raise(rb_eArgError, "empty string");
+    }
+
+    int r = rb_enc_precise_mbclen(p, e, enc);
+
+    if(ONIGENC_MBCLEN_CHARFOUND_P(r)) {
+      if(len_p) {
+        *len_p = ONIGENC_MBCLEN_CHARFOUND_LEN(r);
+      }
+
+      return ONIGENC_MBC_TO_CODE(enc, (UChar*)p, (UChar*)e);
+    } else {
+      rb_raise(rb_eArgError, "invalid byte sequence in %s", rb_enc_name(enc));
+    }
+  }
 }