harfbuzz-ruby 1.0.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b56e40fbf6e75fcf2cecdcfce3d2dc685281a3283c60b96f6a6bac6c28f16fc1
-  data.tar.gz: 6c1137acb54689eedf6a05da99e47e44ea83b000cf0b98a129bdfbc4c60074ad
+  metadata.gz: ba8458f950bfa90bc1c6be8d9d23c8096424cf3eb01383aeedae3678cd49c0c9
+  data.tar.gz: 94631d375acb53a4ab4ae47a63fa6d0e1cae7fe38577fe0a27663086417cf3de
 SHA512:
-  metadata.gz: d4f119a1c98ec67296ec7351bd36943e1a067341305cd7c3ee7e9a1efb7d42c7a8515273951017af8dbe212f3a4252827108c40332297e2dfb2d323595c3ea25
-  data.tar.gz: 2bc37ddb7b27ef540fef758b52a2708d5df8ec41fd96763680e85adcbe43c667ede6f3dd9873b57f381847ff129c6a60398e1773fa499498e5933a146a8b9d5d
+  metadata.gz: f623f5a924225ae41e3c5d71f0a02cdb0d4270c9cf57067cb64fd670b6e87d3a7deacd51e382abc7c836b8ec30a5413da3726e969eb9b70b66cb800e3d6d3c14
+  data.tar.gz: b11384daed8e7d3f76deb8f6f21f5623af3e039543843ea0b649bbaaa39b028000aabc2bae6d19ecd73ed74ecb558bf20f3df1a582e522bb0ce9285cab74c92b

data/CHANGELOG.md

@@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## Unreleased
 
+- Add `Buffer#each_glyph_with_clusters` for yielding the next distinct cluster without changing `Buffer#each_glyph`'s existing six-value interface.
+
 ## v1.0.0 (2026-02-27)
 
-- Initial release.
+- Initial release.

data/README.md

@@ -1,4 +1,4 @@
-# HarfBuzz Ruby
+# HarfBuzz Ruby [![Gem Version](https://badge.fury.io/rb/harfbuzz-ruby.svg)](https://badge.fury.io/rb/harfbuzz-ruby) [![CI](https://github.com/ydah/harfbuzz-ruby/actions/workflows/main.yml/badge.svg)](https://github.com/ydah/harfbuzz-ruby/actions/workflows/main.yml)
 
 Ruby bindings for [HarfBuzz](https://harfbuzz.github.io/) text shaping engine.
 
@@ -90,20 +90,26 @@ face = HarfBuzz::Face.new(blob, 0)
 font = HarfBuzz::Font.new(face)
 
 # Create buffer and add text
+text = "Hello, World!"
 buffer = HarfBuzz::Buffer.new
-buffer.add_utf8("Hello, World!")
+buffer.add_utf8(text)
 buffer.guess_segment_properties
 
 # Shape
 HarfBuzz.shape(font, buffer)
 
-# Read results
-buffer.glyph_infos.zip(buffer.glyph_positions).each do |info, pos|
-  puts "glyph_id=#{info.glyph_id} cluster=#{info.cluster} " \
-       "x_advance=#{pos.x_advance} x_offset=#{pos.x_offset}"
+# Read results without allocating GlyphInfo/GlyphPosition wrappers
+last_cluster = nil
+buffer.each_glyph_with_clusters do |glyph_id, cluster, next_cluster, x_advance, _y_advance, x_offset, _y_offset|
+  source = cluster == last_cluster ? "" : text.byteslice(cluster...(next_cluster || text.bytesize))
+  puts "glyph_id=#{glyph_id} source=#{source.inspect} cluster=#{cluster} next_cluster=#{next_cluster} " \
+       "x_advance=#{x_advance} x_offset=#{x_offset}"
+  last_cluster = cluster
 end
 ```
 
+`Buffer#each_glyph` yields the original six values. `Buffer#each_glyph_with_clusters` adds `next_cluster` as the third value; it is the next distinct cluster in glyph order, or `nil` for the last cluster.
+
 ### Shaping with Features
 
 ```ruby

data/benchmark/shaping_bench.rb

@@ -28,6 +28,11 @@ face = HarfBuzz::Face.new(blob, 0)
 font = HarfBuzz::Font.new(face)
 font.make_immutable!
 
+iteration_buffer = HarfBuzz::Buffer.new
+iteration_buffer.add_utf8("Hello World " * 20)
+iteration_buffer.guess_segment_properties
+HarfBuzz.shape(font, iteration_buffer)
+
 Benchmark.ips do |x|
   x.config(time: 5, warmup: 2)
 
@@ -55,6 +60,58 @@ Benchmark.ips do |x|
     end
   end
 
+  x.report("iterate glyph wrappers") do
+    total = 0
+    infos = iteration_buffer.glyph_infos
+    positions = iteration_buffer.glyph_positions
+    count = [infos.length, positions.length].min
+
+    i = 0
+    while i < count
+      cluster = infos[i].cluster
+      next_cluster = nil
+      next_run_index = i + 1
+
+      while next_run_index < count
+        candidate_cluster = infos[next_run_index].cluster
+        if candidate_cluster != cluster
+          next_cluster = candidate_cluster
+          break
+        end
+
+        next_run_index += 1
+      end
+
+      while i < next_run_index
+        info = infos[i]
+        position = positions[i]
+        total += info.glyph_id
+        total += cluster
+        total += next_cluster || 0
+        total += position.x_advance
+        total += position.y_advance
+        total += position.x_offset
+        total += position.y_offset
+        i += 1
+      end
+    end
+    total
+  end
+
+  x.report("iterate each_glyph_with_clusters") do
+    total = 0
+    iteration_buffer.each_glyph_with_clusters do |glyph_id, cluster, next_cluster, x_advance, y_advance, x_offset, y_offset|
+      total += glyph_id
+      total += cluster
+      total += next_cluster || 0
+      total += x_advance
+      total += y_advance
+      total += x_offset
+      total += y_offset
+    end
+    total
+  end
+
   x.report("create + destroy buffer") do
     HarfBuzz::Buffer.new
   end

data/lib/harfbuzz/buffer.rb

@@ -3,6 +3,23 @@
 module HarfBuzz
   # Wraps hb_buffer_t — text buffer for shaping
   class Buffer
+    GLYPH_INFO_SIZE = C::HbGlyphInfoT.size
+    GLYPH_INFO_CODEPOINT_OFFSET = C::HbGlyphInfoT.offset_of(:codepoint)
+    GLYPH_INFO_CLUSTER_OFFSET = C::HbGlyphInfoT.offset_of(:cluster)
+    GLYPH_POSITION_SIZE = C::HbGlyphPositionT.size
+    GLYPH_POSITION_X_ADVANCE_OFFSET = C::HbGlyphPositionT.offset_of(:x_advance)
+    GLYPH_POSITION_Y_ADVANCE_OFFSET = C::HbGlyphPositionT.offset_of(:y_advance)
+    GLYPH_POSITION_X_OFFSET_OFFSET = C::HbGlyphPositionT.offset_of(:x_offset)
+    GLYPH_POSITION_Y_OFFSET_OFFSET = C::HbGlyphPositionT.offset_of(:y_offset)
+    private_constant :GLYPH_INFO_SIZE,
+                     :GLYPH_INFO_CODEPOINT_OFFSET,
+                     :GLYPH_INFO_CLUSTER_OFFSET,
+                     :GLYPH_POSITION_SIZE,
+                     :GLYPH_POSITION_X_ADVANCE_OFFSET,
+                     :GLYPH_POSITION_Y_ADVANCE_OFFSET,
+                     :GLYPH_POSITION_X_OFFSET_OFFSET,
+                     :GLYPH_POSITION_Y_OFFSET_OFFSET
+
     attr_reader :ptr
 
     # Creates a new empty buffer
@@ -310,6 +327,78 @@ module HarfBuzz
       end
     end
 
+    # Iterates over shaped glyph data without allocating GlyphInfo/GlyphPosition objects
+    # @yield [glyph_id, cluster, x_advance, y_advance, x_offset, y_offset]
+    # @return [Enumerator, self]
+    def each_glyph
+      return enum_for(__method__) { length } unless block_given?
+
+      with_glyph_data do |infos_ptr, positions_ptr, count|
+        i = 0
+        while i < count
+          info_offset = i * GLYPH_INFO_SIZE
+          position_offset = i * GLYPH_POSITION_SIZE
+
+          yield infos_ptr.get_uint32(info_offset + GLYPH_INFO_CODEPOINT_OFFSET),
+                infos_ptr.get_uint32(info_offset + GLYPH_INFO_CLUSTER_OFFSET),
+                positions_ptr.get_int32(position_offset + GLYPH_POSITION_X_ADVANCE_OFFSET),
+                positions_ptr.get_int32(position_offset + GLYPH_POSITION_Y_ADVANCE_OFFSET),
+                positions_ptr.get_int32(position_offset + GLYPH_POSITION_X_OFFSET_OFFSET),
+                positions_ptr.get_int32(position_offset + GLYPH_POSITION_Y_OFFSET_OFFSET)
+
+          i += 1
+        end
+      end
+
+      self
+    end
+
+    # Iterates over shaped glyph data and yields the next distinct cluster without allocating
+    # GlyphInfo/GlyphPosition objects.
+    # @yield [glyph_id, cluster, next_cluster, x_advance, y_advance, x_offset, y_offset]
+    # @return [Enumerator, self]
+    def each_glyph_with_clusters
+      return enum_for(__method__) { length } unless block_given?
+
+      with_glyph_data do |infos_ptr, positions_ptr, count|
+        i = 0
+        while i < count
+          info_offset = i * GLYPH_INFO_SIZE
+          cluster = infos_ptr.get_uint32(info_offset + GLYPH_INFO_CLUSTER_OFFSET)
+          next_cluster = nil
+
+          next_run_index = i + 1
+          while next_run_index < count
+            next_cluster_offset = (next_run_index * GLYPH_INFO_SIZE) + GLYPH_INFO_CLUSTER_OFFSET
+            candidate_cluster = infos_ptr.get_uint32(next_cluster_offset)
+            if candidate_cluster != cluster
+              next_cluster = candidate_cluster
+              break
+            end
+
+            next_run_index += 1
+          end
+
+          while i < next_run_index
+            info_offset = i * GLYPH_INFO_SIZE
+            position_offset = i * GLYPH_POSITION_SIZE
+
+            yield infos_ptr.get_uint32(info_offset + GLYPH_INFO_CODEPOINT_OFFSET),
+                  cluster,
+                  next_cluster,
+                  positions_ptr.get_int32(position_offset + GLYPH_POSITION_X_ADVANCE_OFFSET),
+                  positions_ptr.get_int32(position_offset + GLYPH_POSITION_Y_ADVANCE_OFFSET),
+                  positions_ptr.get_int32(position_offset + GLYPH_POSITION_X_OFFSET_OFFSET),
+                  positions_ptr.get_int32(position_offset + GLYPH_POSITION_Y_OFFSET_OFFSET)
+
+            i += 1
+          end
+        end
+      end
+
+      self
+    end
+
     # @return [Boolean] true if the buffer has glyph position data
     def has_positions?
       C.from_hb_bool(C.hb_buffer_has_positions(@ptr))
@@ -483,6 +572,20 @@ module HarfBuzz
 
     private
 
+    def with_glyph_data
+      info_length_ptr = FFI::MemoryPointer.new(:uint)
+      infos_ptr = C.hb_buffer_get_glyph_infos(@ptr, info_length_ptr)
+      position_length_ptr = FFI::MemoryPointer.new(:uint)
+      positions_ptr = C.hb_buffer_get_glyph_positions(@ptr, position_length_ptr)
+
+      return if infos_ptr.null? || positions_ptr.null?
+
+      count = [info_length_ptr.read_uint, position_length_ptr.read_uint].min
+      return if count.zero?
+
+      yield infos_ptr, positions_ptr, count
+    end
+
     def register_finalizer
       return if instance_variable_defined?(:@borrowed) && @borrowed
 

data/lib/harfbuzz/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HarfBuzz
-  VERSION = "1.0.0"
+  VERSION = "1.1.0"
 end

data/sig/harfbuzz.rbs

@@ -140,6 +140,10 @@ module HarfBuzz
 
     def glyph_infos: () -> Array[GlyphInfo]
     def glyph_positions: () -> Array[GlyphPosition]
+    def each_glyph: () { (Integer glyph_id, Integer cluster, Integer x_advance, Integer y_advance, Integer x_offset, Integer y_offset) -> void } -> self
+                  | () -> Enumerator[[Integer, Integer, Integer, Integer, Integer, Integer], self]
+    def each_glyph_with_clusters: () { (Integer glyph_id, Integer cluster, Integer? next_cluster, Integer x_advance, Integer y_advance, Integer x_offset, Integer y_offset) -> void } -> self
+                                  | () -> Enumerator[[Integer, Integer, Integer?, Integer, Integer, Integer, Integer], self]
     def has_positions?: () -> bool
 
     def normalize_glyphs: () -> self

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: harfbuzz-ruby
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Yudai Takada