jobparser 0.0.2 → 0.1.0

data/lib/jobparser/cleaner.rb

@@ -7,7 +7,6 @@ module JobParser
 
     SALARY_GROUP_REGEX = /£([\d,]*)(?:.+)£([\d,]*)/
     CLEAN_SALARY_REGEX = /,|\s/
-    NBSP = Nokogiri::HTML(" ").text
 
     def initialize(ary, opts = {})
       @subject = ary
@@ -28,6 +27,10 @@ module JobParser
       str.gsub('/n', '').gsub(NBSP, '').strip
     end
 
+    def self.clean_plain_text(str)
+      str.gsub(/\r|\t/, "").gsub(NBSP, " ")
+    end
+
     def self.make_link_absolute(url, href)
       if href.include?("http")
         href

data/lib/jobparser/parsehtml.rb

@@ -27,7 +27,10 @@ module JobParser
     private
 
     def job_location
-      LOCATION_REGEX.match(@plain_text.gsub(/\r|\t/, "")) { |m|
+      special_case_result = use_special_case(:location)
+      return special_case_result unless special_case_result.nil?
+
+      LOCATION_REGEX.match(Cleaner.clean_plain_text(@plain_text)) { |m|
         Cleaner.strip_string(m[1].to_s)
       } || ""
     end
@@ -54,7 +57,16 @@ module JobParser
       Cleaner.new(ary, :type => type).clean
     end
 
+    def use_special_case(name)
+      if special_case = SpecialCases.case_for_url(@url)
+        special_case[name].call(@doc)
+      end
+    end
+
     def job_salary_string
+      special_case_result = use_special_case(:salary_string)
+      return special_case_result unless special_case_result.nil?
+
       salary = ""
       loop_over_elements do |name, elem|
         SALARY_STRING_REGEX.match(@plain_text) { |m|
@@ -65,6 +77,9 @@ module JobParser
     end
 
     def job_salary
+      special_case_result = use_special_case(:salary)
+      return special_case_result unless special_case_result.nil?
+
       salary = ""
       loop_over_elements do |name, elem|
         SALARY_REGEX.match(@plain_text) { |m|
@@ -75,6 +90,9 @@ module JobParser
     end
 
     def job_title
+      special_case_result = use_special_case(:title)
+      return special_case_result unless special_case_result.nil?
+
       title_scorer = Scorer.new
       page_title = @doc.at_css("title").content
       title_scorer.store(page_title, 20).and_score_now

data/lib/jobparser/specialcases.rb

@@ -0,0 +1,56 @@
+# encoding: utf-8
+module JobParser
+  class SpecialCases
+    def self.dictionary
+      {
+        "jobsearch.direct.gov.uk" => {
+          :title => Proc.new { |doc| doc.css(".jobViewContent h2")[1].content },
+          :location => Proc.new { |doc|
+            location = ""
+            listings = doc.css(".jobViewSummary dl dt")
+            listings.each do |dt|
+              if dt.content == "Location"
+                location = dt.next_element.content
+                break
+              end
+            end
+            location
+          },
+          :salary_string => Proc.new { |doc|
+            salary = ""
+            listings = doc.css(".jobViewSummary dl dt")
+            listings.each do |dt|
+              if dt.content == "Salary"
+                salary = Cleaner.strip_string(dt.next_element.content)
+                break
+              end
+            end
+            salary
+          },
+          :salary => Proc.new { |doc|
+            # get string by calling salary_string special case
+            salary = nil
+            salary_string = SpecialCases.call_special_case("jobsearch.direct.gov.uk", :salary_string, doc)
+            /£?([0-9,\.]+)\D*£?([0-9,\.]+)/.match(salary_string) { |m|
+              low = m[1].gsub(",", "").to_i
+              high = m[2].gsub(",", "").to_i
+              salary = [low, high]
+            }
+            salary
+          }
+        }
+      }
+    end
+
+    def self.call_special_case(key, method, doc)
+      self.dictionary[key][method].call(doc)
+    end
+
+    def self.case_for_url(url)
+      self.dictionary.keys.each do |key|
+        return self.dictionary[key] if url.include?(key)
+      end
+      false
+    end
+  end
+end

data/lib/jobparser/version.rb

@@ -1,3 +1,3 @@
 module JobParser
-  VERSION = "0.0.2"
+  VERSION = "0.1.0"
 end

data/lib/jobparser.rb

@@ -3,6 +3,7 @@ require "jobparser/parsehtml"
 require "jobparser/parseurl"
 require "jobparser/cleaner"
 require "jobparser/scorer"
+require "jobparser/specialcases"
 
 module JobParser
   # Your code goes here...

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jobparser
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.1.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-07-22 00:00:00.000000000 Z
+date: 2013-07-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -103,6 +103,7 @@ files:
 - lib/jobparser/parseurl.rb
 - lib/jobparser/regex.rb
 - lib/jobparser/scorer.rb
+- lib/jobparser/specialcases.rb
 - lib/jobparser/version.rb
 - lib/jobparser.rb
 homepage: ''