Class: Readability::Document

Inherits:
Object
  • Object
show all
Defined in:
lib/readability.rb

Constant Summary collapse

DEFAULT_OPTIONS = 
{
  :retry_length => 250,
  :min_text_length => 25,
  :remove_unlikely_candidates => true,
  :weight_classes => true,
  :clean_conditionally => true,
  :remove_empty_nodes => true
}.freeze
REGEXES = 
{
    :unlikelyCandidatesRe => /combx|comment|community|disqus|extra|foot|header|menu|remark|rss|shoutbox|sidebar|sponsor|ad-break|agegate|pagination|pager|popup/i,
    :okMaybeItsACandidateRe => /and|article|body|column|main|shadow/i,
    :positiveRe => /article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i,
    :negativeRe => /combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget/i,
    :divToPElementsRe => /<(a|blockquote|dl|div|img|ol|p|pre|table|ul)/i,
    :replaceBrsRe => /(<br[^>]*>[ \n\r\t]*){2,}/i,
    :replaceFontsRe => /<(\/?)font[^>]*>/i,
    :trimRe => /^\s+|\s+$/,
    :normalizeRe => /\s{2,}/,
    :killBreaksRe => /(<br\s*\/?>(\s|&nbsp;?)*){1,}/,
    :videoRe => /http:\/\/(www\.)?(youtube|vimeo)\.com/i
}

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(input, options = {}) ⇒ Document

Returns a new instance of Document.



18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
 
# File 'lib/readability.rb', line 18

def initialize(input, options = {})
  @options = DEFAULT_OPTIONS.merge(options)
  @input = input

  if RUBY_VERSION =~ /^1\.9\./ && !@options[:encoding]
    @input = GuessHtmlEncoding.encode(@input, @options[:html_headers]) unless @options[:do_not_guess_encoding]
    @options[:encoding] = @input.encoding.to_s
  end

  @input = @input.gsub(REGEXES[:replaceBrsRe], '</p><p>').gsub(REGEXES[:replaceFontsRe], '<\1span>')
  @remove_unlikely_candidates = @options[:remove_unlikely_candidates]
  @weight_classes = @options[:weight_classes]
  @clean_conditionally = @options[:clean_conditionally]
  make_html
end

Instance Attribute Details

#html ⇒ Object

Returns the value of attribute html.



16
17
18
 
# File 'lib/readability.rb', line 16

def html
  @html
end

#options ⇒ Object

Returns the value of attribute options.



16
17
18
 
# File 'lib/readability.rb', line 16

def options
  @options
end

Instance Method Details

#class_weight(e) ⇒ Object 



170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
 
# File 'lib/readability.rb', line 170

def class_weight(e)
  weight = 0
  return weight unless @weight_classes

  if e[:class] && e[:class] != ""
    if e[:class] =~ REGEXES[:negativeRe]
      weight -= 25
    end

    if e[:class] =~ REGEXES[:positiveRe]
      weight += 25
    end
  end

  if e[:id] && e[:id] != ""
    if e[:id] =~ REGEXES[:negativeRe]
      weight -= 25
    end

    if e[:id] =~ REGEXES[:positiveRe]
      weight += 25
    end
  end

  weight
end

#clean_conditionally(node, candidates, selector) ⇒ Object 



297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
 
# File 'lib/readability.rb', line 297

def clean_conditionally(node, candidates, selector)
  return unless @clean_conditionally
  node.css(selector).each do |el|
    weight = class_weight(el)
    content_score = candidates[el] ? candidates[el][:content_score] : 0
    name = el.name.downcase

    if weight + content_score < 0
      el.remove
      debug("Conditionally cleaned #{name}##{el[:id]}.#{el[:class]} with weight #{weight} and content score #{content_score} because score + content score was less than zero.")
    elsif el.text.count(",") < 10
      counts = %w[p img li a embed input].inject({}) { |m, kind| m[kind] = el.css(kind).length; m }
      counts["li"] -= 100

      content_length = el.text.strip.length  # Count the text length excluding any surrounding whitespace
      link_density = get_link_density(el)
      to_remove = false
      reason = ""

      if counts["img"] > counts["p"]
        reason = "too many images"
        to_remove = true
      elsif counts["li"] > counts["p"] && name != "ul" && name != "ol"
        reason = "more <li>s than <p>s"
        to_remove = true
      elsif counts["input"] > (counts["p"] / 3).to_i
        reason = "less than 3x <p>s than <input>s"
        to_remove = true
      elsif content_length < (options[:min_text_length] || TEXT_LENGTH_THRESHOLD) && (counts["img"] == 0 || counts["img"] > 2)
        reason = "too short a content length without a single image"
        to_remove = true
      elsif weight < 25 && link_density > 0.2
        reason = "too many links for its weight (#{weight})"
        to_remove = true
      elsif weight >= 25 && link_density > 0.5
        reason = "too many links for its weight (#{weight})"
        to_remove = true
      elsif (counts["embed"] == 1 && content_length < 75) || counts["embed"] > 1
        reason = "<embed>s with too short a content length, or too many <embed>s"
        to_remove = true
      end

      if to_remove
        debug("Conditionally cleaned #{name}##{el[:id]}.#{el[:class]} with weight #{weight} and content score #{content_score} because it has #{reason}.")
        el.remove
      end
    end
  end
end

#content(remove_unlikely_candidates = :default) ⇒ Object 



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
 
# File 'lib/readability.rb', line 57

def content(remove_unlikely_candidates = :default)
  @remove_unlikely_candidates = false if remove_unlikely_candidates == false

  @html.css("script, style").each(&:remove)

  remove_unlikely_candidates! if @remove_unlikely_candidates
  transform_misused_divs_into_paragraphs!
  candidates = score_paragraphs(options[:min_text_length])
  best_candidate = select_best_candidate(candidates)
  article = get_article(candidates, best_candidate)

  cleaned_article = sanitize(article, candidates, options)
  if article.text.strip.length < options[:retry_length]
    if @remove_unlikely_candidates
      @remove_unlikely_candidates = false
    elsif @weight_classes
      @weight_classes = false
    elsif @clean_conditionally
      @clean_conditionally = false
    else
      # nothing we can do
      return cleaned_article
    end

    make_html
    content
  else
    cleaned_article
  end
end

#debug(str) ⇒ Object 



212
213
214
 
# File 'lib/readability.rb', line 212

def debug(str)
  puts str if options[:debug]
end

#get_article(candidates, best_candidate) ⇒ Object 



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
 
# File 'lib/readability.rb', line 88

def get_article(candidates, best_candidate)
  # Now that we have the top candidate, look through its siblings for content that might also be related.
  # Things like preambles, content split by ads that we removed, etc.

  sibling_score_threshold = [10, best_candidate[:content_score] * 0.2].max
  output = Nokogiri::XML::Node.new('div', @html)
  best_candidate[:elem].parent.children.each do |sibling|
    append = false
    append = true if sibling == best_candidate[:elem]
    append = true if candidates[sibling] && candidates[sibling][:content_score] >= sibling_score_threshold

    if sibling.name.downcase == "p"
      link_density = get_link_density(sibling)
      node_content = sibling.text
      node_length = node_content.length

      if node_length > 80 && link_density < 0.25
        append = true
      elsif node_length < 80 && link_density == 0 && node_content =~ /\.( |$)/
        append = true
      end
    end

    if append
      sibling.name = "div" unless %w[div p].include?(sibling.name.downcase)
      output << sibling
    end
  end

  output
end

#get_link_density(elem) ⇒ Object 



134
135
136
137
138
 
# File 'lib/readability.rb', line 134

def get_link_density(elem)
  link_length = elem.css("a").map(&:text).join("").length
  text_length = elem.text.length
  link_length / text_length.to_f
end

#make_html ⇒ Object 



34
35
36
 
# File 'lib/readability.rb', line 34

def make_html
  @html = Nokogiri::HTML(@input, nil, @options[:encoding])
end

#remove_unlikely_candidates! ⇒ Object 



216
217
218
219
220
221
222
223
224
 
# File 'lib/readability.rb', line 216

def remove_unlikely_candidates!
  @html.css("*").each do |elem|
    str = "#{elem[:class]}#{elem[:id]}"
    if str =~ REGEXES[:unlikelyCandidatesRe] && str !~ REGEXES[:okMaybeItsACandidateRe] && elem.name.downcase != 'body'
      debug("Removing unlikely candidate - #{str}")
      elem.remove
    end
  end
end

#sanitize(node, candidates, options = {}) ⇒ Object 



246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
 
# File 'lib/readability.rb', line 246

def sanitize(node, candidates, options = {})
  node.css("h1, h2, h3, h4, h5, h6").each do |header|
    header.remove if class_weight(header) < 0 || get_link_density(header) > 0.33
  end

  node.css("form, object, iframe, embed").each do |elem|
    elem.remove
  end

  if @options[:remove_empty_nodes]
    # remove <p> tags that have no text content - this will also remove p tags that contain only images.
    node.css("p").each do |elem|
      elem.remove if elem.content.strip.empty?
    end
  end

  # Conditionally clean <table>s, <ul>s, and <div>s
  clean_conditionally(node, candidates, "table, ul, div")

  # We'll sanitize all elements using a whitelist
  base_whitelist = @options[:tags] || %w[div p]
  # We'll add whitespace instead of block elements,
  # so a<br>b will have a nice space between them
  base_replace_with_whitespace = %w[br hr h1 h2 h3 h4 h5 h6 dl dd ol li ul address blockquote center]

  # Use a hash for speed (don't want to make a million calls to include?)
  whitelist = Hash.new
  base_whitelist.each {|tag| whitelist[tag] = true }
  replace_with_whitespace = Hash.new
  base_replace_with_whitespace.each { |tag| replace_with_whitespace[tag] = true }

  ([node] + node.css("*")).each do |el|
    # If element is in whitelist, delete all its attributes
    if whitelist[el.node_name]
      el.attributes.each { |a, x| el.delete(a) unless @options[:attributes] && @options[:attributes].include?(a.to_s) }

      # Otherwise, replace the element with its contents
    else
      if replace_with_whitespace[el.node_name]
        el.swap(Nokogiri::XML::Text.new(' ' << el.text << ' ', el.document))
      else
        el.swap(Nokogiri::XML::Text.new(el.text, el.document))
      end
    end

  end

  # Get rid of duplicate whitespace
  node.to_html.gsub(/[\r\n\f]+/, "\n" ).gsub(/[\t  ]+/, " ")
end

#score_node(elem) ⇒ Object 



197
198
199
200
201
202
203
204
205
206
207
208
209
210
 
# File 'lib/readability.rb', line 197

def score_node(elem)
  content_score = class_weight(elem)
  case elem.name.downcase
    when "div"
      content_score += 5
    when "blockquote"
      content_score += 3
    when "form"
      content_score -= 3
    when "th"
      content_score -= 5
  end
  { :content_score => content_score, :elem => elem }
end

#score_paragraphs(min_text_length) ⇒ Object 



140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
 
# File 'lib/readability.rb', line 140

def score_paragraphs(min_text_length)
  candidates = {}
  @html.css("p,td").each do |elem|
    parent_node = elem.parent
    grand_parent_node = parent_node.respond_to?(:parent) ? parent_node.parent : nil
    inner_text = elem.text

    # If this paragraph is less than 25 characters, don't even count it.
    next if inner_text.length < min_text_length

    candidates[parent_node] ||= score_node(parent_node)
    candidates[grand_parent_node] ||= score_node(grand_parent_node) if grand_parent_node

    content_score = 1
    content_score += inner_text.split(',').length
    content_score += [(inner_text.length / 100).to_i, 3].min

    candidates[parent_node][:content_score] += content_score
    candidates[grand_parent_node][:content_score] += content_score / 2.0 if grand_parent_node
  end

  # Scale the final candidates score based on link density. Good content should have a
  # relatively small link density (5% or less) and be mostly unaffected by this operation.
  candidates.each do |elem, candidate|
    candidate[:content_score] = candidate[:content_score] * (1 - get_link_density(elem))
  end

  candidates
end

#select_best_candidate(candidates) ⇒ Object 



120
121
122
123
124
125
126
127
128
129
130
131
132
 
# File 'lib/readability.rb', line 120

def select_best_candidate(candidates)
  sorted_candidates = candidates.values.sort { |a, b| b[:content_score] <=> a[:content_score] }

  debug("Top 5 canidates:")
  sorted_candidates[0...5].each do |candidate|
    debug("Candidate #{candidate[:elem].name}##{candidate[:elem][:id]}.#{candidate[:elem][:class]} with score #{candidate[:content_score]}")
  end

  best_candidate = sorted_candidates.first || { :elem => @html.css("body").first, :content_score => 0 }
  debug("Best candidate #{best_candidate[:elem].name}##{best_candidate[:elem][:id]}.#{best_candidate[:elem][:class]} with score #{best_candidate[:content_score]}")

  best_candidate
end

#title ⇒ Object 



52
53
54
55
 
# File 'lib/readability.rb', line 52

def title
  title = @html.css("title").first
  title ? title.text : nil
end

#transform_misused_divs_into_paragraphs! ⇒ Object 



226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
 
# File 'lib/readability.rb', line 226

def transform_misused_divs_into_paragraphs!
  @html.css("*").each do |elem|
    if elem.name.downcase == "div"
      # transform <div>s that do not contain other block elements into <p>s
      if elem.inner_html !~ REGEXES[:divToPElementsRe]
        debug("Altering div(##{elem[:id]}.#{elem[:class]}) to p");
        elem.name = "p"
      end
    else
      # wrap text nodes in p tags
#          elem.children.each do |child|
#            if child.text?
#              debug("wrapping text node with a p")
#              child.swap("<p>#{child.text}</p>")
#            end
#          end
    end
  end
end