Fixed to_binary for Bit

lib/pgvector/bit.rb

@@ -27,7 +27,10 @@ def to_a
 
  def to_binary
  buffer = [@data.length].pack("l>")
- @data.split(/.{8}/).pack("B*", buffer: buffer)
+ # TODO improve
+ @data.scan(/.{1,8}/).each do |b|
+ [b].pack("B*", buffer: buffer)
+ end
  buffer
  end
  end

test/pg_test.rb

@@ -33,7 +33,7 @@ def test_halfvec_text
  end
 
  def test_bit_text
- embedding = "101"
+ embedding = "1010000001"
  conn.exec_params("INSERT INTO pg_items (binary_embedding) VALUES ($1), (NULL)", [embedding])
 
  res = conn.exec("SELECT * FROM pg_items ORDER BY id").to_a
@@ -42,7 +42,7 @@ def test_bit_text
  end
 
  def test_bit_binary
- embedding = "101"
+ embedding = "1010000001"
  conn.exec_params("INSERT INTO pg_items (binary_embedding) VALUES ($1), (NULL)", [embedding])
 
  res = conn.exec_params("SELECT * FROM pg_items ORDER BY id", [], 1).to_a
@@ -99,7 +99,7 @@ def test_type_map_binary
  def test_copy_text
  embedding = Pgvector::Vector.new([1, 2, 3])
  half_embedding = Pgvector::HalfVector.new([1, 2, 3])
- binary_embedding = Pgvector::Bit.new([true, false, true])
+ binary_embedding = Pgvector::Bit.new([true, false, true, false, false, false, false, false, false, true])
  sparse_embedding = Pgvector::SparseVector.new([1, 2, 3])
  coder = PG::TextEncoder::CopyRow.new
  conn.copy_data("COPY pg_items (embedding, half_embedding, binary_embedding, sparse_embedding) FROM STDIN", coder) do
@@ -108,21 +108,21 @@ def test_copy_text
  res = conn.exec("SELECT * FROM pg_items").first
  assert_equal [1, 2, 3], res["embedding"]
  assert_equal [1, 2, 3], res["half_embedding"].to_a
- assert_equal "101", res["binary_embedding"]
+ assert_equal "1010000001", res["binary_embedding"]
  assert_equal [1, 2, 3], res["sparse_embedding"].to_a
  end
 
  def test_copy_binary
  embedding = Pgvector::Vector.new([1, 2, 3])
- binary_embedding = Pgvector::Bit.new([true, false, true])
+ binary_embedding = Pgvector::Bit.new([true, false, true, false, false, false, false, false, false, true])
  sparse_embedding = Pgvector::SparseVector.new([1, 2, 3])
  coder = PG::BinaryEncoder::CopyRow.new
  conn.copy_data("COPY pg_items (embedding, binary_embedding, sparse_embedding) FROM STDIN WITH (FORMAT BINARY)", coder) do
  conn.put_copy_data([embedding.to_binary, binary_embedding.to_binary, sparse_embedding.to_binary])
  end
  res = conn.exec("SELECT * FROM pg_items").first
  assert_equal [1, 2, 3], res["embedding"]
- assert_equal "101", res["binary_embedding"]
+ assert_equal "1010000001", res["binary_embedding"]
  assert_equal [1, 2, 3], res["sparse_embedding"].to_a
  end
 
@@ -134,7 +134,7 @@ def conn
  conn.exec("CREATE EXTENSION IF NOT EXISTS vector")
  end
  conn.exec("DROP TABLE IF EXISTS pg_items")
- conn.exec("CREATE TABLE pg_items (id bigserial PRIMARY KEY, embedding vector(3), half_embedding halfvec(3), binary_embedding bit(3), sparse_embedding sparsevec(3))")
+ conn.exec("CREATE TABLE pg_items (id bigserial PRIMARY KEY, embedding vector(3), half_embedding halfvec(3), binary_embedding bit(10), sparse_embedding sparsevec(3))")
 
  registry = PG::BasicTypeRegistry.new.define_default_types
  Pgvector::PG.register_vector(registry)