[GR-19220] Adding rb_hash_new_capa to CEXT api (#3103)

PullRequest: truffleruby/3902

Author: andrykonchin

CHANGELOG.md

@@ -28,6 +28,7 @@ Compatibility:
 * Handle `long long` and aliases in `Fiddle` (#3128, @eregon).
 * Add `Module#refinements` (#3039, @itarato).
 * Add `Refinement#refined_class` (#3039, @itarato).
+* Add `rb_hash_new_capa` function (#3039, @itarato).
 
 Performance:
 

lib/cext/ABI_version.txt

@@ -1 +1 @@
-8
+9

lib/cext/include/ruby/internal/intern/hash.h

@@ -106,6 +106,17 @@ VALUE rb_hash(VALUE obj);
  */
 VALUE rb_hash_new(void);
 
+/**
+ * Identical to rb_hash_new(), except it additionally specifies how many keys
+ * it is expected to contain. This way you can create a hash that is large enough
+ * for your need. For large hashes it means it won't need to be reallocated and
+ * rehashed as much, improving performance.
+ *
+ * @param[in]  capacity  Designed capacity of the hash.
+ * @return     An empty Hash, whose capacity is `capacity`.
+ */
+VALUE rb_hash_new_capa(long capacity);
+
 /**
  * Duplicates a hash.
  *

lib/truffle/truffle/cext.rb

@@ -972,6 +972,11 @@ def rb_hash_new
     {}
   end
 
+  def rb_hash_new_capa(capacity)
+    raise RuntimeError, 'st_table too big' if capacity < 0
+    {}
+  end
+
   def rb_ident_hash_new
     {}.compare_by_identity
   end

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

@@ -105,6 +105,10 @@ VALUE hash_spec_rb_hash_new(VALUE self) {
   return rb_hash_new();
 }
 
+VALUE hash_spec_rb_hash_new_capa(VALUE self, VALUE capacity) {
+  return rb_hash_new_capa(NUM2LONG(capacity));
+}
+
 VALUE rb_ident_hash_new(void); /* internal.h, used in ripper */
 
 VALUE hash_spec_rb_ident_hash_new(VALUE self) {
@@ -149,6 +153,7 @@ void Init_hash_spec(void) {
   rb_define_method(cls, "rb_hash_lookup2", hash_spec_rb_hash_lookup2, 3);
   rb_define_method(cls, "rb_hash_lookup2_default_undef", hash_spec_rb_hash_lookup2_default_undef, 2);
   rb_define_method(cls, "rb_hash_new", hash_spec_rb_hash_new, 0);
+  rb_define_method(cls, "rb_hash_new_capa", hash_spec_rb_hash_new_capa, 1);
   rb_define_method(cls, "rb_ident_hash_new", hash_spec_rb_ident_hash_new, 0);
   rb_define_method(cls, "rb_hash_size", hash_spec_rb_hash_size, 1);
   rb_define_method(cls, "rb_hash_set_ifnone", hash_spec_rb_hash_set_ifnone, 2);

spec/ruby/optional/capi/hash_spec.rb

@@ -50,6 +50,22 @@
     end
   end
 
+  ruby_version_is '3.2' do
+    describe "rb_hash_new_capa" do
+      it "returns a new hash" do
+        @s.rb_hash_new_capa(3).should == {}
+      end
+
+      it "creates a hash with no default proc" do
+        @s.rb_hash_new_capa(3) {}.default_proc.should be_nil
+      end
+
+      it "raises RuntimeError when negative index is provided" do
+        -> { @s.rb_hash_new_capa(-1) }.should raise_error(RuntimeError, "st_table too big")
+      end
+    end
+  end
+
   describe "rb_ident_hash_new" do
     it "returns a new compare by identity hash" do
       result = @s.rb_ident_hash_new

spec/truffleruby.next-specs

@@ -33,3 +33,4 @@ spec/ruby/core/sizedqueue/shift_spec.rb
 spec/ruby/core/module/refinements_spec.rb
 spec/ruby/core/refinement/refined_class_spec.rb
 spec/ruby/core/module/used_refinements_spec.rb
+spec/ruby/optional/capi/hash_spec.rb

src/main/c/cext/hash.c

@@ -24,6 +24,10 @@ VALUE rb_hash_new() {
   return RUBY_CEXT_INVOKE("rb_hash_new");
 }
 
+VALUE rb_hash_new_capa(long capacity) {
+  return rb_tr_wrap(polyglot_invoke(RUBY_CEXT, "rb_hash_new_capa", capacity));
+}
+
 VALUE rb_ident_hash_new() {
   return RUBY_CEXT_INVOKE("rb_ident_hash_new");
 }