Class: R::Vector

Inherits:
Object show all
Includes:
Enumerable, BinaryOperators, ExecBinOp, ExecUniOp, IndexedObject, LogicalOperators, UnaryOperators
Defined in:
lib/R_interface/rvector.rb

Constant Summary collapse

CHUNK_SIZE =

1M rows per page

1_000_000

Constants inherited from Object

Object::EXPLICIT_RUBY_SURFACE

Instance Attribute Summary

Attributes inherited from Object

#expression, #r_interop

Instance Method Summary collapse

Methods included from LogicalOperators

#&, #|

Methods included from ExecUniOp

#exec_uni_oper

Methods included from UnaryOperators

#!, #+@, #-@

Methods included from ExecBinOp

#coerce, #exec_bin_oper

Methods included from BinaryOperators

#!=, #%, #*, #**, #+, #-, #/, #<, #<=, #>, #>=, #_, #eq, #int_div, #til

Methods included from IndexedObject

#[], #[]=, #size

Methods inherited from Object

#==, build, build_class_probe_count, build_from_r_class_string, build_from_wrapper_tag, #call, class_probe_fetch_r_class, increment_build_class_probe_count!, #inspect, #instance_of?, #instance_variable_get, #instance_variable_set, #is_a?, #method_missing, #nil?, #pretty_print, #rclass, reset_build_counters!, #respond_to?, #send, #to_i, #to_ruby, #to_s, #typeof

Constructor Details

#initialize(r_interop) ⇒ Vector 



43
44
45
 
# File 'lib/R_interface/rvector.rb', line 43

def initialize(r_interop)
  super(r_interop)
end

Dynamic Method Handling

This class handles dynamic methods through the method_missing method in the class R::Object

Instance Method Details

#<=>(other_vector) ⇒ Object

@TODO: SHOULD DEFINE COMPARISON BETWEEN TWO VECTORS



204
205
206
 
# File 'lib/R_interface/rvector.rb', line 204

def <=>(other_vector)
  ::Kernel.raise(::NotImplementedError, "R::Vector#<=> is not implemented")
end

#class ⇒ Object 



35
36
37
 
# File 'lib/R_interface/rvector.rb', line 35

def class
  ::R::Vector
end

#each(mode = :vec) ⇒ Object 



165
166
167
168
169
170
171
172
173
174
175
176
177
178
 
# File 'lib/R_interface/rvector.rb', line 165

def each(mode = :vec)
  case mode
  when :vec
    (1..length.unboxed_get(0)).each do |i|
      yield self[i]
    end
  when :native
    (0...length.unboxed_get(0)).each do |i|
      yield unboxed_get(i)
    end
  else
    ::Kernel.raise("Type #{mode.inspect} is unknown for method :each")
  end
end

#each_with_index(result = :vec) ⇒ Object

Need to override each_with_index, as R indexing starts at 1



184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
 
# File 'lib/R_interface/rvector.rb', line 184

def each_with_index(result = :vec)
  case result
  when :vec
    (1..length.unboxed_get(0)).each do |i|
      yield self[i], i
    end
  when :native
    (0...length.unboxed_get(0)).each do |i|
      yield unboxed_get(i), i
    end
  else
    ::Kernel.raise("Type #{result} is unknown for method :each")
  end
  
end

#map(&block) ⇒ Object 



159
160
161
162
163
 
# File 'lib/R_interface/rvector.rb', line 159

def map(&block)
  stitch(halo: 0) do |data|
    data.map(&block)
  end
end

#pop ⇒ Object 



110
111
112
113
 
# File 'lib/R_interface/rvector.rb', line 110

def pop
  # Return unboxed first element as a Ruby scalar
  unboxed_get(0)
end

#stitch(halo: 0, &block) ⇒ Object 



117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
 
# File 'lib/R_interface/rvector.rb', line 117

def stitch(halo: 0, &block)
  len = atomic_vector_length

  # 2. Allocate result vector in R
  res_name = ::R::Support.generate_var_name
  ::R.bridge.eval_r("#{res_name} <- numeric(#{len})")

  # 3. Process in chunks with Halo
  offset = 0
  while offset < len
    current_chunk_size = [CHUNK_SIZE, len - offset].min
    
    # Calculate extended range for Halo
    # R indices start at 1, but pull_double_vector uses 0-based offset
    pull_offset = [0, offset - halo].max
    pull_end = [len, offset + current_chunk_size + halo].min
    actual_pull_size = pull_end - pull_offset
    
    # 3.1 Pull chunk with Halo
    data = ::R.bridge.pull_double_vector(@r_interop, len, pull_offset, actual_pull_size)
    
    # 3.2 Process in Ruby
    # The block receives the data with halo. 
    # It's up to the block to handle the context, 
    # but the result should be the same size as data.
    result_data_with_halo = block.call(data)
    
    # 3.3 Extract the "inner" part (discard halo results)
    # The offset of our chunk within the pulled data is (offset - pull_offset)
    inner_start = offset - pull_offset
    inner_result = result_data_with_halo[inner_start, current_chunk_size]
    
    # 3.4 Push only the inner chunk back to R
    ::R.bridge.push_double_vector(inner_result, res_name, offset, len)
    
    offset += current_chunk_size
  end
  
  # 4. Return as R::Object (wrapped as Vector)
  ::R::Object.build(res_name)
end

#to_ary ⇒ Object

Return Ruby array so RSpec/eq and array conversion don't forward to_ary to R. Must always return an Array (length-1 vector unboxes to scalar via >> nil, so wrap in [v]).



101
102
103
104
 
# File 'lib/R_interface/rvector.rb', line 101

def to_ary
  v = self >> nil
  v.is_a?(::Array) ? v : [v]
end

#unboxed_get(index = nil, depth = 0) ⇒ Object Also known as: >>, <<

Unbox vector to Ruby. Atomic vectors: length 1 → scalar, length > 1 → Array of scalars. Recursion depth is checked; vectors do not recurse into R::Object (elements are scalars).



51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
 
# File 'lib/R_interface/rvector.rb', line 51

def unboxed_get(index = nil, depth = 0)
  if depth >= ::R::Support::MAX_UNBOX_DEPTH
    ::Kernel.raise(::R::UnboxDepthError, "unbox: list too deep (max depth #{::R::Support::MAX_UNBOX_DEPTH} exceeded)")
  end
  if index.nil?
    arr = ::R.bridge.pull_vector(@r_interop)
    return (arr.length == 1 ? arr[0] : arr)
  end

  # For indexed unboxing, use binary transport + result protocol (no printed-text parsing).
  idx = index
  len = atomic_vector_length
  ::Kernel.raise(::IndexError.new("index #{idx} out of array bounds: 0...#{len - 1}")) if idx >= len

  type = atomic_vector_typeof

  case type
  when 'integer'
    data = ::R.bridge.pull_integer_vector(@r_interop, 1, idx, 1)
    data[0]
  when 'character'
    var_name = ::R::Support.generate_var_name
    assignment = "#{var_name} <- #{@r_interop}[[#{idx + 1}]]"
    envelope = ::R.bridge.eval_r_with_result(assignment)
    raise "Result protocol: no envelope for character element #{@r_interop}[[#{idx + 1}]]" unless envelope
    raise "Result protocol: expected scalar_character, got #{envelope[:type]}" unless envelope[:type] == :scalar_character
    v = envelope[:value]
    (v.is_a?(::String) && v =~ /^rb_obj_\d+$/) ? ::R::Support.get_ruby_object(v) : v
  when 'logical'
    var_name = ::R::Support.generate_var_name
    assignment = "#{var_name} <- #{@r_interop}[[#{idx + 1}]]"
    envelope = ::R.bridge.eval_r_with_result(assignment)
    raise "Result protocol: no envelope for logical element #{@r_interop}[[#{idx + 1}]]" unless envelope
    raise "Result protocol: expected scalar_logical, got #{envelope[:type]}" unless envelope[:type] == :scalar_logical
    case envelope[:value]
    when true then true
    when false then false
    else nil
    end
  else
    data = ::R.bridge.pull_double_vector(@r_interop, 1, idx, 1)
    data[0]
  end
end