Class: NewsCrawler::Processing::StructureAnalysis

Inherits:
Object
  • Object
show all
Includes:
CrawlerModule, URLHelper
Defined in:
lib/news_crawler/processing/structure_analysis.rb

Overview

Analyse website structure to extract content Database should only contains raw data from one website.

Instance Method Summary collapse

Methods included from URLHelper

#get_url_path, #same_domain?

Methods included from CrawlerModule

#find_all, #find_one, #find_unprocessed, #load_yaml, #mark_all_as_unprocessed, #mark_processed, #mark_unprocessed, #next_unprocessed, #save_yaml

Constructor Details

#initialize ⇒ StructureAnalysis

Returns a new instance of StructureAnalysis.



40
41
42
43
44
45
46
47
48
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 40

def initialize
  @url_stats = {}
  while (url = next_unprocessed)
    NCLogger.get_logger.info "[NC::P::SA] Processing #{url}"
    re = extract_content(url)
    @url_stats[url] = re
    save_yaml(url, re)
  end
end

Instance Method Details

#analyse(url) ⇒ Object

Get and analyse url for information



264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 264

def analyse(url)
  #        puts "processing #{url}"
  html_doc = RawData.find_by_url(url)
  doc = Nokogiri.HTML(html_doc)
  inner_url = doc.xpath('//a').collect { | a_el |
    temp_url = (a_el.attribute 'href').to_s
    if (!temp_url.nil?) && (temp_url[0] == '/')
      temp_url = URI.join(url, temp_url).to_s
    end
    temp_url
  }

  inner_url.delete_if { | url_0 |
    (url_0.nil?) || (url_0.size == 0) || (url_0 == '#') ||
    (url_0 == 'javascript:;')
  }

  inner_url.each do  | url |
    @url_stats[url] = (@url_stats[url] || 0) + 1
  end
  mark_processed(url)
end

#classify_h2(root, limit) ⇒ Symbol

Predict type of tree point by root is fragment of article or index page

Parameters:

  • root (Nokogiri::XML::Node)

Returns:

  • (Symbol) —

    one of :article, :list



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
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 103

def classify_h2(root, limit)
  current = root
  current = current.parent if current.text?

  depth = 0

  while true
    expect_hash = hash_node(current, 0)
    previous = current
    current = current.parent

    depth += 1
    lons = {}
    node_count = 0
    node_list = [previous]
    current.children.each do | child |
      hc = hash_node(child, depth - 1)
      if hc == expect_hash
        node_count += 1
        node_list << child
      end
    end

    if node_count > 1
      a_tag_len, non_a_tag_len = count_a_and_non_a_tag(current)
      if non_a_tag_len > a_tag_len
        return :article
      else
        return :list
      end
      break
    end

    if current == limit
      a_tag_len, non_a_tag_len = count_a_and_non_a_tag(current)
      if non_a_tag_len > a_tag_len
        return :article
      else
        return :list
      end
      break
    end
  end

  return :list
end

#count_a_and_non_a_tag(node) ⇒ [Fixnum, Fixnum]

Count a tag and non-a tag in tree pointed by node

Parameters:

  • node (Nokogiri::XML::Node)

Returns:

  • ([Fixnum, Fixnum]) —

    a tag and non-a tag



153
154
155
156
157
158
159
160
161
162
163
164
165
166
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 153

def count_a_and_non_a_tag(node)
  a_tag_list = node.xpath './/a'
  a_tag_len = a_tag_list.count # number of a tag

  non_a_tag_list = node.xpath './/text()[not (ancestor::a)]'
  non_a_tag_len = non_a_tag_list.to_a.inject(0) do | memo, node |
    if node.content.gsub(/\s+/, '').length > 15
      memo + 1
    else
      memo
    end
  end
  [ a_tag_len, non_a_tag_len ]
end

#extract_content(url) ⇒ Object 



50
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
95
96
97
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 50

def extract_content(url)
  html_doc = RawData.find_by_url(url)
  result = {}
  result[:type] == :article

  # Remove tag causing trouble to nokogiri
  html_doc = remove_tag(html_doc, 'script')
  html_doc = remove_tag(html_doc, 'iframe')
  html_doc = remove_tag(html_doc, 'style')

  doc = Nokogiri::HTML.parse(html_doc)
  longest = find_longest_node(doc)
  lowest_ancestor, path_to_longest = find_lowest_ancestor_has_id(longest)

  # Heuristic 1
  # Longest content is a element as id attribute
  if path_to_longest.length == 2
    return { :type => :list }
  end

  parent = path_to_longest[1..-1]
  parent = parent.reverse
  xpath_path = parent.join('/')
  xpath_path = '//' + xpath_path + '//text()'

  guest_type = classify_h2(longest, lowest_ancestor)
  result = { :type => guest_type }

  if (result[:type] == :article)
    title_ = lowest_ancestor.css('h1')
    if title_.count == 1
      result[:title] = title_.to_a[0].content
    else
      # if cann't guest title then assume it isn't an article
      result[:type] = :list
    end

    main_content = ''
    lowest_ancestor.xpath(xpath_path).each do | node |
      main_content += node.content
    end

    result[:content] = main_content
  end

  mark_processed(url)
  result
end

#find_longest_node(doc) ⇒ Object

Find longest text node that doesn’t have a in ancestors list

Parameters:

  • doc (Nokogiri::XML::Node) 


198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 198

def find_longest_node(doc)
  xpath_query = '//*[@id]//text()[not (ancestor::a)]'

  a_l = doc.xpath xpath_query

  longest = nil
  longest_len = 0

  a_l.each do | en |
    node_content_wo_space = en.content.gsub(/\s/, '') # trick here
    if node_content_wo_space.length > longest_len
      longest_len = node_content_wo_space.length
      longest = en
    end
  end

  return longest
end

#find_lowest_ancestor_has_id(node) ⇒ Nokogiri::XML::Node

Find the lowest node’s ancestor has id attribute

Parameters:

  • node (Nokogiri::XML::Node)

Returns:

  • (Nokogiri::XML::Node) 


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

def find_lowest_ancestor_has_id(node)
  found_id = false

  closest_ancestor = node

  path_to_closest = []

  while (!found_id)
    if closest_ancestor.has_attribute?('id')
      path_to_closest << "#{closest_ancestor.node_name}[@id='#{closest_ancestor.attribute('id')}']"
      found_id = true
    else
      if closest_ancestor.has_attribute?('class')
        node_class = "@class = '#{closest_ancestor.attribute('class')}'"
      else
        node_class = 'not(@class)'
      end
      path_to_closest << "#{closest_ancestor.node_name}[#{node_class}]"
      closest_ancestor = closest_ancestor.parent
    end
  end

  return [ closest_ancestor, path_to_closest ]
end

#get_result ⇒ Object 



294
295
296
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 294

def get_result
  @url_stats
end

#hash_node(node, limit = -1)) ⇒ String

Calculate hash of a node by its and children info

Parameters:

  • node (Nokogiri::XML::Node)
  • limit (Fixnum) (defaults to: -1)) —

    limit depth of children (-1 for unlimited)

Returns:

  • (String) —

    Hash of node in base 64 encode



239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 239

def hash_node(node, limit = -1)
  node_sign = node.node_name
  node_sign += "##{node['id']}" unless node['id'].nil?
  node_sign += ".#{node['class']}" unless node['class'].nil?

  hash_sum = node_sign

  if limit != 0
    child_hash = Set.new
    node.children.each do | child_node |
      child_hash.add(hash_node(child_node, limit - 1))
    end

    child_hash.each do | ch |
      hash_sum += ch
    end
  else

  end

  Digest::SHA2.new.base64digest(hash_sum)
end

#is_url?(url) ⇒ Boolean

Check if it is really ‘url’

Parameters:

  • url (String)

Returns:

  • (Boolean) 


290
291
292
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 290

def is_url?(url)
  (url.size != 0) && (url != '#') && (url != 'javascript:;')
end

#node_info(node) ⇒ String

Return String represents node’s name, node’s id and node’s class

Parameters:

  • node (Nokogiri::XML::Node)

Returns:

  • (String) 


228
229
230
231
232
233
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 228

def node_info(node)
  node_pp = node.node_name
  node_pp += '#' + node.attribute('id') if node.has_attribute?('id')
  node_pp += '.' + node.attribute('class') if node.has_attribute?('class')
  node_pp
end

#remove_tag(html_doc, tag) ⇒ Object

Remove unwanted HTML tag

Parameters:

  • html_doc (String) —

    HTML document

  • tag (String) —

    tag to be removed



220
221
222
223
 
# File 'lib/news_crawler/processing/structure_analysis.rb', line 220

def remove_tag(html_doc, tag)
  pattern = Regexp.new("<#{tag}.*?>.*?</#{tag}>", Regexp::MULTILINE)
  html_doc.gsub(pattern, '')
end