Module: PdfExtract::References

Defined in:
lib/references/references.rb

Class Method Summary collapse

Class Method Details

.frequencies(lines, delimit_key) ⇒ Object 



46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
 
# File 'lib/references/references.rb', line 46

def self.frequencies lines, delimit_key
  fs = {}
  lines.each do |line|
    val = line[delimit_key].floor
    fs[val] ||= 0
    fs[val] = fs[val].next
  end

  ary = []
  fs.each_pair do |key, val|
    ary << {:value => key, :count => val}
  end

  ary.sort_by { |item| item[:count] }.reverse
end

.include_in(pdf) ⇒ Object 



186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
 
# File 'lib/references/references.rb', line 186

def self.include_in pdf
  pdf.spatials :references, :depends_on => [:sections] do |parser|

    sections = []

    parser.objects :sections do |section|
      sections << section
    end

    parser.after do
      max_score = sections.map {|s| s[:reference_score]}.max
      min_permittable = max_score - (max_score * pdf.settings[:reference_flex])

      refs = []

      sections = sections.reject do |s|
        # A section without any years is definitely not a list of
        # references. So too a section that appears in the first
        # half of an article.
        s[:lateness] < pdf.settings[:min_lateness] || s[:year_ratio].zero?
      end

      sections.each do |section|
        if section[:reference_score] >= min_permittable
        # TODO Enable classification once we have a reasonable model.
        #if Score.reference?(section)
          content = Spatial.get_text_content(section)
          if numeric_sequence? pdf, content
            refs += split_by_delimiter pdf, content
          elsif multi_margin? section[:lines]
            refs += split_by_margin section[:lines]
          elsif multi_spacing? section[:lines]
            refs += split_by_line_spacing section[:lines]
          end
        end
      end

      # TODO Ideally we wouldn't see the ref headers here.
      # Unfortunately publication details can look a lot like references.
      refs.reject do |ref|
        norm = ref[:content].downcase.strip
        norm =~ /references?/ || norm =~ /submitted for publication/ || norm =~ /additional contributions/
      end

    end

  end
end

.multi_margin?(lines) ⇒ Boolean

Returns:

  • (Boolean) 


155
156
157
 
# File 'lib/references/references.rb', line 155

def self.multi_margin? lines
  lines.uniq { |line| line[:x_offset].floor }.count > 1
end

.multi_spacing?(lines) ⇒ Boolean

Returns:

  • (Boolean) 


159
160
161
 
# File 'lib/references/references.rb', line 159

def self.multi_spacing? lines
  lines.uniq { |line| line[:spacing].floor }.count > 1
end

.numeric_sequence?(pdf, content) ⇒ Boolean

Returns:

  • (Boolean) 


163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
 
# File 'lib/references/references.rb', line 163

def self.numeric_sequence? pdf, content
  last_n = -1
  first_n = -1
  seq_count = 0
  content.scan /\d+/ do |m|
     # Avoid misinterpreting years as sequence
    if m.to_i < pdf.settings[:max_reference_order]
      if last_n == -1
        last_n = m.to_i
        first_n = m.to_i if first_n == -1
      elsif last_n.next == m.to_i
        last_n = last_n.next
        seq_count = seq_count.next
      end
    end
  end

  # Sequence must be long enough and first number of sequence
  # must appear near the very start of content.
  large_enough = seq_count >= pdf.settings[:min_sequence_count]
  large_enough && content[0..30] =~ /#{first_n.to_s}/
end

.partition_by(ary, &block) ⇒ Object 



32
33
34
35
36
37
38
39
40
41
42
43
44
 
# File 'lib/references/references.rb', line 32

def self.partition_by ary, &block
  matching = []
  parts = []
  ary.each do |item|
    if yield(item)
      parts << matching
      matching = []
    end
    matching << item
  end
  parts << matching
  parts.reject { |p| p.empty? }
end

.select_delimiter(lines, delimit_key) ⇒ Object 



62
63
64
 
# File 'lib/references/references.rb', line 62

def self.select_delimiter lines, delimit_key
  frequencies(lines, delimit_key)[1][:value]
end

.split_by_delimiter(pdf, s) ⇒ Object 



80
81
82
83
84
85
86
87
88
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
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
 
# File 'lib/references/references.rb', line 80

def self.split_by_delimiter pdf, s
  # Find sequential numbers and use them as partition points.

  # Determine the charcaters that are most likely part of numeric
  # delimiters.

  after = {}
  before = {}
  last_n = -1

  s.scan /[^\d]?\d+[^\d]/ do |m|
    n = m[/\d+/].to_i
    if n < pdf.settings[:max_reference_order]
      if last_n == -1
        before[m[0]] ||= 0
        before[m[0]] = before[m[0]].next
        after[m[-1]] ||= 0
        after[m[-1]] = after[m[-1]].next
        last_n = n
      elsif n == last_n.next
        before[m[0]] ||= 0
        before[m[0]] = before[m[0]].next
        after[m[-1]] ||= 0
        after[m[-1]] = after[m[-1]].next
        last_n = last_n.next
      end
    end
  end

  b_s = "" if before.length.zero?
  b_s = "\\" + before.max_by { |_, v| v }[0] unless before.length.zero?
  a_s = "" if after.length.zero?
  a_s = "\\" + after.max_by { |_, v| v }[0] unless after.length.zero?

  # TODO Turn into settings. Needs typed settings
  if ["", "\\[", "\\ "].include?(b_s) && ["", "\\.", "\\]", "\\ "].include?(a_s)

    # Split by the delimiters and record separate refs.

    last_n = -1
    current_ref = ""
    refs = []
    parts = s.partition(Regexp.new "#{b_s}?\\d+#{a_s}")

    while not parts[1].length.zero?
      n = parts[1][/\d+/].to_i
      if n < pdf.settings[:max_reference_order] && last_n == -1
        last_n = n
      elsif n == last_n.next
        current_ref += parts[0]
        refs << {
          :content => current_ref.strip,
          :order => last_n
        }
        current_ref = ""
        last_n = last_n.next
      else
        current_ref += parts[0] + parts[1]
      end

      parts = parts[2].partition(Regexp.new "#{b_s}?\\d+#{a_s}")
    end

    refs << {
      :content => (current_ref + parts[0]).strip,
      :order => last_n
    }

    refs

  else
    []
  end
end

.split_by_line_spacing(lines) ⇒ Object 



73
74
75
76
77
78
 
# File 'lib/references/references.rb', line 73

def self.split_by_line_spacing lines
  delimiting_spacing = select_delimiter lines, :spacing
  lines = lines.drop_while { |l| l[:spacing].floor != delimiting_spacing }
  parts = partition_by(lines) { |line| line[:spacing].floor == delimiting_spacing }
  parts.map { |part| {:content => part.map { |line| line[:content] }.join(" ")} }
end

.split_by_margin(lines) ⇒ Object 



66
67
68
69
70
71
 
# File 'lib/references/references.rb', line 66

def self.split_by_margin lines
  delimiting_x_offset = select_delimiter lines, :x_offset
  lines = lines.drop_while { |l| l[:x_offset].floor != delimiting_x_offset }
  parts = partition_by(lines) { |line| line[:x_offset].floor == delimiting_x_offset }
  parts.map { |part| {:content => part.map { |line| line[:content] }.join(" ")} }
end