improved the pulling out of bids by parsing the html from a snapshot …

…of the table, instead of using a series of webdriver calls

.gitignore

@@ -1,2 +1,4 @@
 csv
-
+*.log
+*.swp
+*~

Gemfile

@@ -1,2 +1,3 @@
 source :rubygems
 gem 'watir-webdriver'
+gem 'hpricot'

Gemfile.lock

@@ -4,6 +4,7 @@ GEM
     childprocess (0.3.7)
       ffi (~> 1.0, >= 1.0.6)
     ffi (1.3.1)
+    hpricot (0.8.6)
     multi_json (1.5.0)
     rubyzip (0.9.9)
     selenium-webdriver (2.29.0)
@@ -19,4 +20,5 @@ PLATFORMS
   ruby
 
 DEPENDENCIES
+  hpricot
   watir-webdriver

site_bindings/quibids.rb

@@ -1,4 +1,5 @@
 require 'watir-webdriver'
+require 'hpricot'
 
 #this module implements
 module QUIBIDS
@@ -98,12 +99,17 @@ def seconds_left
   end
 
 
-   #improve this, have it parse static html to improve the data capture and eliminate live updating problem
   def get_new_bids
-    cur_amt = @last_amt
     bids = []
-    @auction_els[:history].rows.to_a.reverse.each do |bid_row|
-      x, bidder, amt, type = bid_row.cells.to_a.map {|c| c.text}
+
+    #to do this reliably is tricky, the table moves, 
+    #so going row-by-row / cell-by-cell with webdriver calls will cause accuracy issues
+    # instead, just pull out a snapshot of the html, and parse that out
+    # its actually faster this way too, and very accurate
+    hdoc = Hpricot(@auction_els[:history].html)
+
+    hdoc.search('//tr').reverse.each do |tr|
+      bidder, amt, type = tr.search('//td').map {|tr| tr.inner_html}[1..3]
       amt = amt.pa_calc_amt
       if amt > @last_amt
         bids << {