got data hash working

tools/mozilla/security_advisory_scraper/mozilla_security_advisory_scraper.rb

@@ -25,6 +25,11 @@ class MozillaSecurityAdvisoryScraper
     Nokogiri::HTML(response_body)
   end
 
+  def get_and_parse_advisory(advisory_url)
+    html = get(advisory_url)
+    parse_html(html)
+  end
+
   def advisory_urls_in_html(html_doc)
     html_doc.xpath('//li[@class="level-item"]/a').map {|link| relative_url = link['href']; "https://www.mozilla.org#{relative_url}"}
   end
@@ -48,14 +53,58 @@ class MozillaSecurityAdvisoryScraper
     end
   end
 
-  # we know its the very first dl tag w class name summary
   def parse_advisory_summary(advisory_doc)
-    nodes = advisory_doc.xpath('//dl[@class="summary"]').first.children
-    # dd tags and dt tags have the info we care about. dt tag is the key, dd
-    # tag is the value
+    advisory_doc.xpath('//dl[@class="summary"]').first.css('dd')
   end
 
-  def parse_advisory_cve(advisory_doc)
-    advisory_doc.xpath('//dl[@class="cve"]')
+
+  def parse_advisory_cves(advisory_doc)
+    advisory_doc.xpath('//section[@class="cve"]')
   end
+
+  def data_hash(advisory_doc)
+    title = advisory_doc.title
+    header = advisory_doc.css('h2').text
+    summary = parse_advisory_summary(advisory_doc)
+    announced = summary[0].text
+    impact = summary[1].text
+    products = summary[2].text
+    fixed_in = summary[3].text
+    cve_nodes = parse_advisory_cves(advisory_doc)
+    cves = cve_nodes.map { | cve_node | cve_data_hash(cve_node) }
+
+    {
+      :title => title,
+      :header => header,
+      :announced => announced,
+      :impact => impact,
+      :products => products,
+      :fixed_in => fixed_in,
+      :cves => cves
+    }
+  end
+
+  def cve_data_hash(advisory_cve_doc)
+    cve_id_and_title = advisory_cve_doc.css('h4').text.gsub("\n", "").gsub("#", "").strip.split(": ")
+    cve_id = cve_id_and_title.first 
+    cve_title = cve_id_and_title.last
+    
+    dl_and_dd_tags = advisory_cve_doc.css('dl').css('dd')
+    cve_reporter = dl_and_dd_tags.first.text
+    cve_impact = dl_and_dd_tags.last.text
+
+    cve_description = advisory_cve_doc.css('p').text
+    cve_references = advisory_cve_doc.css('a').last.values.first
+
+    {
+      :cve_id => cve_id,
+      :title => cve_title,
+      :reporter => cve_reporter,
+      :impact => cve_impact,
+      :description => cve_description,
+      :references => cve_references,
+    }
+ 
+  end
+
 end