Parsing out total results.

anthonator · Oct 29, 2013 · 212a365 · 212a365
1 parent 250e8b9
commit 212a365
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 3 deletions.
diff --git a/lib/ar_book_finder/search_results_parser.rb b/lib/ar_book_finder/search_results_parser.rb
@@ -1,9 +1,13 @@
 module ARBookFinder
   class SearchResultsParser
+    SEARCH_ROOT_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_pnlSearchFormPanel"]'
     SEARCH_PAGE_COUNT_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_ucSeachResults_lblResultsSummaryTop"]'
+    SEARCH_RESULTS_COUNT = '//*[@id="ctl00_ContentPlaceHolder1_ucSearchResultsHeader_lblResultSummary"]'
     SEARCH_RESULTS_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_ucSeachResults_lblQuizzes"]/table'
 
+    COLLECTION_ROOT_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_ucCollection_pnlSearchFormPanel"]'
     COLLECTION_PAGE_COUNT_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_ucCollection_ucSeachResults_lblResultsSummaryTop"]'
+    COLLECTION_RESULTS_COUNT = '//*[@id="ctl00_ContentPlaceHolder1_ucCollection_ucSearchResultsHeader_lblResultSummary"]'
     COLLECTION_RESULTS_XPATH = '//*[@id="ctl00_ContentPlaceHolder1_ucCollection_ucSeachResults_lblQuizzes"]/table'
 
     BOOK_XPATH = 'tbody/tr/td[2]'
@@ -15,28 +19,34 @@ class SearchResultsParser
     def initialize(html, collection = false)
       @doc = Nokogiri::HTML.parse(html)
       @xpath_const = collection ? :COLLECTION : :SEARCH
+      @root = @doc.xpath(self.class.const_get(:"#{@xpath_const}_ROOT_XPATH"))
       @books = []
     end
 
     def parse
       @current_page = parse_current_page.to_i
       @page_count = parse_page_count.to_i
+      @total_books = parse_total_books.to_i
       @books = parse_results
       self
     end
 
     private
     def parse_current_page
-      @doc.xpath(self.class.const_get(:"#{@xpath_const}_PAGE_COUNT_XPATH")).text.gsub(/Page /, '').gsub(/ of \d+/, '')
+      @root.xpath(self.class.const_get(:"#{@xpath_const}_PAGE_COUNT_XPATH")).text.gsub(/Page /, '').gsub(/ of \d+/, '')
     end
 
     def parse_page_count
-      @doc.xpath(self.class.const_get(:"#{@xpath_const}_PAGE_COUNT_XPATH")).text.gsub(/Page \d+ of /, '')
+      @root.xpath(self.class.const_get(:"#{@xpath_const}_PAGE_COUNT_XPATH")).text.gsub(/Page \d+ of /, '')
+    end
+
+    def parse_total_books
+      @root.xpath(self.class.const_get(:"#{@xpath_const}_RESULTS_COUNT")).text.gsub(/Titles \d+ - \d+ of /, '')
     end
 
     def parse_results
       books = []
-      @doc.xpath(self.class.const_get(:"#{@xpath_const}_RESULTS_XPATH")).each_with_index do |result, i|
+      @root.xpath(self.class.const_get(:"#{@xpath_const}_RESULTS_XPATH")).each_with_index do |result, i|
         next if i.odd?
         book = result.xpath(BOOK_XPATH)
         book_detail = book.xpath(BOOK_DETAIL_XPATH)

diff --git a/spec/ar_book_finder/collection_spec.rb b/spec/ar_book_finder/collection_spec.rb
@@ -16,6 +16,11 @@
     results.current_page.should == 2
   end
 
+  it 'should return total book count' do
+    results = ARBookFinder.collection({ 'Awards' => 'ALA Notable/Best Books' })
+    results.total_books.should > 0
+  end
+
   it 'should fetch book data' do
     results = ARBookFinder.collection('Awards' => 'ALA Notable/Best Books')
     book = results.books[0]

diff --git a/spec/ar_book_finder/search_spec.rb b/spec/ar_book_finder/search_spec.rb
@@ -11,6 +11,11 @@
     results.current_page.should == 2
   end
 
+  it 'should return total book count' do
+    results = ARBookFinder.search('harry potter')
+    results.total_books.should > 0
+  end
+
   it 'should fetch book data' do
     results = ARBookFinder.search('harry potter')
     book = results.books[0]