Class: CraigScrape::Listings

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

Overview

Listings represents a parsed Craigslist listing page and is generally returned by CraigScrape.scrape_listing

Constant Summary collapse

LABEL = 
/^(.+?)[ ]*[\-]?$/
LOCATION = 
/^[ ]*\((.*?)\)$/
IMG_TYPE = 
/^[ ]*(.+)[ ]*$/
HEADER_DATE = 
/^[ ]*(?:Sun|Mon|Tue|Wed|Thu|Fri|Sat)[ ]+(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Nov|Dec)[ ]+([0-9]{1,2})[ ]*$/i
SUMMARY_DATE = 
/^[ ]([^ ]+)[ ]+([^ ]+)[ ]*[\-][ ]*$/
NEXT_PAGE_LINK = 
/^[ ]*(?:next [\d]+ postings|Next \>\>)[ ]*$/
XPATH_POST_DATE = 
"*[@class='itemdate']"
XPATH_POST_IMGPIC = 
"*[@class='itempx']/*[@class='p']"
"//*[@class='ban']//a"
XPATHS_POST_PRICE =

There’s a couple places that the price hangs out. We search in this order

["*[@class='itempp']", "*[@class='itemph']"]

Constants inherited from Scraper

Scraper::HTML_ENCODING, Scraper::HTML_TAG, Scraper::HTTP_HEADERS, Scraper::URL_PARTS

Instance Attribute Summary

Attributes inherited from Scraper

#url

Instance Method Summary collapse

Methods inherited from Scraper

#attributes, #downloaded?, #initialize, #uri

Constructor Details

This class inherits a constructor from CraigScrape::Scraper

Instance Method Details

#next_page ⇒ Object

Returns a Listings object of the next_page_url on the current listings object



116
117
118
 
# File 'lib/listings.rb', line 116

def next_page
  CraigScrape::Listings.new URI.encode(next_page_url) if next_page_url
end

#next_page_href ⇒ Object

String, URL Path href-fragment of the next page link



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
98
99
100
101
102
103
104
105
106
107
108
 
# File 'lib/listings.rb', line 72

def next_page_href
  unless @next_page_href
   
    if html.at_xpath(XPATH_PAGENAV_LINKS)
      # Post 12/3
      next_link = html.xpath(XPATH_PAGENAV_LINKS).find{|link| NEXT_PAGE_LINK.match link.content}
      @next_page_href = next_link[:href] if next_link
    else 
      # Old style
      cursor = html.at 'p:last-of-type'
      
      cursor = cursor.at 'a' if cursor
      
      # Category Listings have their 'next 100 postings' link at the end of the doc in a p tag 
      next_link = cursor if cursor and NEXT_PAGE_LINK.match cursor.inner_html

      # Search listings put their next page in a link towards the top
      next_link = (html / 'a').find{ |a| he_decode(a.inner_html) == '<b>Next>></b>' } unless next_link
              
      # Some search pages have a bug, whereby a 'next page' link isn't displayed,
      # even though we can see that theres another page listed in the page-number links block at the top
      # and bottom of the listing page
      unless next_link
        cursor = html % 'div.sh:first-of-type > b:last-of-type'

        # If there's no 'a' in the next sibling, we'll have just performed a nil assignment, otherwise
        # We're looking good.
        next_link = cursor.next_element if cursor and /^[\d]+$/.match cursor.inner_html
      end
      
      # We have an anchor tag - so - let's assign the href:
      @next_page_href = next_link[:href] if next_link
    end
  end
  
  @next_page_href
end

#next_page_url ⇒ Object

String, Full URL Path of the ‘next page’ link



111
112
113
 
# File 'lib/listings.rb', line 111

def next_page_url
  (next_page_href) ? url_from_href(next_page_href) : nil
end

#posts ⇒ Object

Array, PostSummary objects found in the listing



25
26
27
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
61
62
63
64
65
66
67
68
69
 
# File 'lib/listings.rb', line 25

def posts
  unless @posts
    current_date = nil
    @posts = []

    # All we care about are p and h4 tags. This seemed to be the only way I could do this on Nokogiri:
    post_tags = html.search('*').reject{|n| !/^(?:p|h4)$/i.match n.name } 

    # The last p in the list is sometimes a 'next XXX pages' link. We don't want to include this in our PostSummary output:
    post_tags.pop if (
      post_tags.length > 0 and 
      post_tags.last.at('a') and 
      NEXT_PAGE_LINK.match post_tags.last.at('a').inner_html
    )

    # Now we iterate though the listings:
    post_tags.each do |el|
      case el.name
        when 'p'
         post_summary = parse_summary el, current_date

         # Validate that required fields are present:
         parse_error! unless [post_summary[:label],post_summary[:href]].all?{|f| f and f.length > 0}
  
         post_summary[:url] = url_from_href post_summary[:href]

         @posts << CraigScrape::Posting.new(post_summary)
       when 'h4'
        # Let's make sense of the h4 tag, and then read all the p tags below it
        if HEADER_DATE.match he_decode(el.inner_html)
          # Generally, the H4 tags contain valid dates. When they do - this is easy:
          current_date = most_recent_date $1, $2
        elsif html.at('h4:last-of-type') == el
          # There's a specific bug in craigslist, where these nonsense h4's just appear without anything relevant inside them.
          # They're safe to ignore if they're not the last h4 on the page. I fthey're the last h4 on the page, 
          # we need to pull up the full post in order to accurate tell the date.
          # Setting this to nil will achieve the eager-load.
          current_date = nil
        end
      end        
    end        
  end

  @posts
end