Module: Transformed

Defined in:
lib/rbbt/ner/segment/transformed.rb

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Instance Attribute Details

#transformation_stackObject

Returns the value of attribute transformation_stack.



26
27
28
# File 'lib/rbbt/ner/segment/transformed.rb', line 26

def transformation_stack
  @transformation_stack
end

#transformed_segmentsObject

Returns the value of attribute transformed_segments.



26
27
28
# File 'lib/rbbt/ner/segment/transformed.rb', line 26

def transformed_segments
  @transformed_segments
end

Class Method Details

.sort(segments) ⇒ Object



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
# File 'lib/rbbt/ner/segment/transformed.rb', line 62

def self.sort(segments)
  segments.compact.sort do |a,b|
    case
    when ((a.nil? and b.nil?) or (a.offset.nil? and b.offset.nil?))
      0
    when (a.nil? or a.offset.nil?)
      -1
    when (b.nil? or b.offset.nil?)
      +1
      # Non-overlap
    when (a.end < b.offset.to_i or b.end < a.offset.to_i)
      b.offset <=> a.offset
      # b includes a
    when (a.offset.to_i >= b.offset.to_i and a.end <= b.end)
      -1
      # b includes a
    when (b.offset.to_i >= a.offset.to_i and b.end <= a.end)
      +1
      # Overlap
    when (a.offset.to_i > b.offset.to_i and a.end > b.end or b.offset.to_i < a.offset.to_i and b.end > a.end)
      a.length <=> b.length
    else
      raise "Unexpected case in sort: #{a.range} - #{b.range}"
    end
  end
end

.transform(text, segments, replacement = nil, &block) ⇒ Object



6
7
8
9
10
11
12
# File 'lib/rbbt/ner/segment/transformed.rb', line 6

def self.transform(text, segments, replacement = nil, &block)

  text.extend Transformed
  text.replace(segments, replacement, &block)

  text
end

.with_transform(text, segments, replacement = nil) ⇒ Object



14
15
16
17
18
19
20
21
22
23
24
# File 'lib/rbbt/ner/segment/transformed.rb', line 14

def self.with_transform(text, segments, replacement = nil)

  text.extend Transformed
  text.replace(segments, replacement)

  segments = yield text

  segments = nil unless Array === segments

  text.restore(segments, true)
end

Instance Method Details

#fix_segment(segment, range, diff) ⇒ Object



134
135
136
137
138
139
140
141
142
143
144
145
146
147
# File 'lib/rbbt/ner/segment/transformed.rb', line 134

def fix_segment(segment, range, diff)
  case
    # Before
  when segment.end < range.begin
    # After
  when segment.offset.to_i > range.end + diff
    segment.offset.to_i -= diff
    # Includes
  when (segment.offset.to_i <= range.begin and segment.end >= range.end + diff)
    segment.replace self[segment.offset.to_i..segment.end - diff]
  else
    raise "Segment Overlaps"
  end
end

#replace(segments, replacement = nil, &block) ⇒ Object



89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
# File 'lib/rbbt/ner/segment/transformed.rb', line 89

def replace(segments, replacement = nil, &block)
  @transformed_segments ||= {}
  @transformation_stack ||= []
  stack = []

  Transformed.sort(segments).each do |segment|
    next if segment.offset.nil?
    shift = shift segment.range

    next if shift.nil?

    shift_begin, shift_end = shift

    text_offset = self.respond_to?(:offset)? self.offset.to_i : 0
    updated_begin = segment.offset.to_i + shift_begin - text_offset
    updated_end   = segment.range.last + shift_end - text_offset

    updated_range = (updated_begin..updated_end)

    updated_text = self[updated_begin..updated_end]

    original_text = segment.dup
    segment.replace updated_text

    case
    when block_given?
      new =  block.call(segment)
    when String === replacement
      new = replacement
    when Proc === replacement
      new = replacement.call(segment)
    end

    diff = new.length - segment.segment_length

    self[updated_begin..updated_end] = new

    @transformed_segments[segment.object_id] = [segment.range, diff, updated_text, updated_range, @transformed_segments.size]

    segment.replace original_text
    stack << segment.object_id
  end
  @transformation_stack << stack
end

#restore(segments = [], first_only = false) ⇒ Object

Restore the sentence from all transformation. Segments that are passed as parameters are restored from transformed space to original space



151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
# File 'lib/rbbt/ner/segment/transformed.rb', line 151

def restore(segments = [], first_only = false)
  return segments if @transformation_stack.empty?

  if first_only
    @transformation_stack.pop.reverse.each do |id|
      orig_range, diff, text, range = @transformed_segments.delete id

      new_range = (range.begin..range.last + diff)
      self[new_range] = text
      segments.each do |segment|
        next unless Segment === segment
        fix_segment(segment, range, diff)
      end if Array === segments
    end
    segments
  else
    while @transformation_stack.any?
      restore(segments, true)
    end
    segments
  end
end

#shift(segment_o) ⇒ Object



28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
# File 'lib/rbbt/ner/segment/transformed.rb', line 28

def shift(segment_o)
  begin_shift = 0
  end_shift = 0

  @transformed_segments.sort_by{|id, info| info.last}.each{|id,info| 
    pseg_o, diff = info

    case
      # Before
    when segment_o.last + end_shift < pseg_o.begin
      # After
    when (segment_o.begin + begin_shift > pseg_o.last)
      begin_shift += diff
      end_shift += diff
      # Includes
    when (segment_o.begin + begin_shift <= pseg_o.begin and segment_o.last + end_shift >= pseg_o.last)
      end_shift += diff
      # Inside
    when (segment_o.begin + begin_shift >= pseg_o.begin and segment_o.last + end_shift <= pseg_o.last)
      return nil
      # Overlaps start
    when (segment_o.begin + begin_shift <= pseg_o.begin and segment_o.last + end_shift <= pseg_o.last)
      return nil
      # Overlaps end
    when (segment_o.begin + begin_shift >= pseg_o.begin and segment_o.last + end_shift >= pseg_o.last)
      return nil
   else
      raise "Unknown overlaps: #{segment_o.inspect} - #{pseg_o.inspect}"
    end
  }

  [begin_shift, end_shift]
end