fix each_grapheme_cluster's size [Bug #14363]

From: Hugo Peixoto <hugo.peixoto@gmail.com>

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62892 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

string.c

@@ -8355,6 +8355,56 @@ rb_str_codepoints(VALUE str)
     return rb_str_enumerate_codepoints(str, ary);
 }
 
+static VALUE
+rb_str_each_grapheme_cluster_size(VALUE str, VALUE args, VALUE eobj)
+{
+    long grapheme_cluster_count = 0;
+    regex_t *reg_grapheme_cluster = NULL;
+    static regex_t *reg_grapheme_cluster_utf8 = NULL;
+    int encidx = ENCODING_GET(str);
+    rb_encoding *enc = rb_enc_from_index(encidx);
+    int unicode_p = rb_enc_unicode_p(enc);
+    const char *ptr, *end;
+
+    if (!unicode_p || single_byte_optimizable(str)) {
+	return rb_str_length(str);
+    }
+
+    /* synchronize */
+    if (encidx == rb_utf8_encindex() && reg_grapheme_cluster_utf8) {
+	reg_grapheme_cluster = reg_grapheme_cluster_utf8;
+    }
+    if (!reg_grapheme_cluster) {
+	const OnigUChar source[] = "\\X";
+	int r = onig_new(&reg_grapheme_cluster, source, source + sizeof(source) - 1,
+			 ONIG_OPTION_DEFAULT, enc, OnigDefaultSyntax, NULL);
+	if (r) {
+	    rb_bug("cannot compile grapheme cluster regexp");
+	}
+	if (encidx == rb_utf8_encindex()) {
+	    reg_grapheme_cluster_utf8 = reg_grapheme_cluster;
+	}
+    }
+
+    ptr = RSTRING_PTR(str);
+    end = RSTRING_END(str);
+
+    while (ptr < end) {
+	OnigPosition len = onig_match(reg_grapheme_cluster,
+				      (const OnigUChar *)ptr, (const OnigUChar *)end,
+				      (const OnigUChar *)ptr, NULL, 0);
+	if (len == 0) break;
+	if (len < 0) {
+	    break;
+	}
+	grapheme_cluster_count++;
+	ptr += len;
+    }
+    RB_GC_GUARD(str);
+
+    return LONG2NUM(grapheme_cluster_count);
+}
+
 static VALUE
 rb_str_enumerate_grapheme_clusters(VALUE str, VALUE ary)
 {
@@ -8426,7 +8476,7 @@ rb_str_enumerate_grapheme_clusters(VALUE str, VALUE ary)
 static VALUE
 rb_str_each_grapheme_cluster(VALUE str)
 {
-    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_char_size);
+    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_grapheme_cluster_size);
     return rb_str_enumerate_grapheme_clusters(str, 0);
 }
 

test/ruby/test_string.rb

@@ -982,11 +982,18 @@ CODE
       "\u{1f469 200d 2764 fe0f 200d 1f469}",
     ].each do |g|
       assert_equal [g], g.each_grapheme_cluster.to_a
+      assert_equal 1, g.each_grapheme_cluster.size
+    end
+
+    [
+      ["\u{a 308}", ["\u000A", "\u0308"]],
+      ["\u{d 308}", ["\u000D", "\u0308"]],
+      ["abc", ["a", "b", "c"]],
+    ].each do |str, grapheme_clusters|
+      assert_equal grapheme_clusters, str.each_grapheme_cluster.to_a
+      assert_equal grapheme_clusters.size, str.each_grapheme_cluster.size
     end
 
-    assert_equal ["\u000A", "\u0308"], "\u{a 308}".each_grapheme_cluster.to_a
-    assert_equal ["\u000D", "\u0308"], "\u{d 308}".each_grapheme_cluster.to_a
-    assert_equal ["a", "b", "c"], "abc".b.each_grapheme_cluster.to_a
     s = ("x"+"\u{10ABCD}"*250000)
     assert_empty(s.each_grapheme_cluster {s.clear})
   end