From: Nobuyoshi Nakada <nobu@...>
Date: 2009-02-13T22:19:41+09:00
Subject: [ruby-dev:37998] [Feature:1.9] {Array,Enumerable}#uniq_by, #uniq_by!

なかだです。

ブロックの値にしたがって一意のものを選ぶ、Array#uniq_byというの
はどうでしょうか。


Index: array.c
===================================================================
--- array.c	(revision 22100)
+++ array.c	(working copy)
@@ -2875,13 +2875,40 @@ ary_add_hash(VALUE hash, VALUE ary)
 }
 
-static VALUE
-ary_make_hash(VALUE ary)
+static inline VALUE
+ary_tmp_hash_new(void)
 {
     VALUE hash = rb_hash_new();
-
     RBASIC(hash)->klass = 0;
+    return hash;
+}
+
+static VALUE
+ary_make_hash(VALUE ary)
+{
+    VALUE hash = ary_tmp_hash_new();
     return ary_add_hash(hash, ary);
 }
 
+static VALUE
+ary_add_hash_by(VALUE hash, VALUE ary)
+{
+    long i;
+
+    for (i = 0; i < RARRAY_LEN(ary); ++i) {
+	VALUE v = rb_ary_elt(ary, i), k = rb_yield(v);
+	if (rb_hash_lookup2(hash, k, Qundef) == Qundef) {
+	    rb_hash_aset(hash, k, v);
+	}
+    }
+    return hash;
+}
+
+static VALUE
+ary_make_hash_by(VALUE ary)
+{
+    VALUE hash = ary_tmp_hash_new();
+    return ary_add_hash_by(hash, ary);
+}
+
 static inline void
 ary_recycle_hash(VALUE hash)
@@ -3063,4 +3090,43 @@ rb_ary_uniq(VALUE ary)
 }
 
+static int
+push_value(st_data_t key, st_data_t val, st_data_t ary)
+{
+    rb_ary_push((VALUE)ary, (VALUE)val);
+    return ST_DELETE;
+}
+
+static VALUE
+rb_ary_uniq_by_bang(VALUE ary)
+{
+    VALUE hash;
+    long size;
+
+    RETURN_ENUMERATOR(ary, 0, 0);
+    hash = ary_make_hash_by(ary);
+    size = RHASH_SIZE(hash);
+    if (RARRAY_LEN(ary) == size) return Qnil;
+    ary_resize_capa(ary, size);
+    ARY_SET_LEN(ary, 0);
+    st_foreach(RHASH_TBL(hash), push_value, ary);
+    ary_recycle_hash(hash);
+    return ary;
+}
+
+static VALUE
+rb_ary_uniq_by(VALUE ary)
+{
+    VALUE hash;
+    long size;
+
+    RETURN_ENUMERATOR(ary, 0, 0);
+    hash = ary_make_hash_by(ary);
+    size = RHASH_SIZE(hash);
+    ary = ary_new(rb_obj_class(ary), size);
+    st_foreach(RHASH_TBL(hash), push_value, ary);
+    ary_recycle_hash(hash);
+    return ary;
+}
+
 /* 
  *  call-seq:
@@ -3920,4 +3986,6 @@ Init_Array(void)
     rb_define_method(rb_cArray, "uniq", rb_ary_uniq, 0);
     rb_define_method(rb_cArray, "uniq!", rb_ary_uniq_bang, 0);
+    rb_define_method(rb_cArray, "uniq_by", rb_ary_uniq_by, 0);
+    rb_define_method(rb_cArray, "uniq_by!", rb_ary_uniq_by_bang, 0);
     rb_define_method(rb_cArray, "compact", rb_ary_compact, 0);
     rb_define_method(rb_cArray, "compact!", rb_ary_compact_bang, 0);
Index: enum.c
===================================================================
--- enum.c	(revision 22100)
+++ enum.c	(working copy)
@@ -1794,4 +1794,29 @@ enum_cycle(int argc, VALUE *argv, VALUE 
 }
 
+static VALUE
+enum_uniq_by_i(VALUE i, VALUE hash, int argc, VALUE *argv)
+{
+    return rb_hash_aset(hash, rb_yield_values(argc, argv), i);
+}
+
+static int
+push_value(st_data_t key, st_data_t val, st_data_t ary)
+{
+    rb_ary_push((VALUE)ary, (VALUE)val);
+    return ST_DELETE;
+}
+
+static VALUE
+enum_uniq_by(VALUE obj)
+{
+    VALUE hash = rb_hash_new(), uniq;
+
+    RBASIC(hash)->klass = 0;
+    rb_block_call(obj, id_each, 0, 0, enum_uniq_by_i, hash);
+    uniq = rb_ary_new2(RHASH_SIZE(hash));
+    st_foreach(RHASH_TBL(hash), push_value, uniq);
+    return uniq;
+}
+
 /*
  *  The <code>Enumerable</code> mixin provides collection classes with
@@ -1853,4 +1878,5 @@ Init_Enumerable(void)
     rb_define_method(rb_mEnumerable, "drop_while", enum_drop_while, 0);
     rb_define_method(rb_mEnumerable, "cycle", enum_cycle, -1);
+    rb_define_method(rb_mEnumerable, "uniq_by", enum_uniq_by, 0);
 
     id_eqq  = rb_intern("===");


-- 
--- 僕の前にBugはない。
--- 僕の後ろにBugはできる。
    中田 伸悦