AddressBookImporter 0.0.4 → 0.0.5

data/lib/address_book_importer.rb

@@ -1,8 +1,9 @@
 require 'rubygems'
 require 'mechanize'
+require 'cgi'
 
 module AddressBookImporter
-  VERSION = '0.0.4'
+  VERSION = '0.0.5'
 
   class EmptyEmailException < Exception ; end
   class LoginErrorException < Exception ; end
@@ -14,7 +15,8 @@ module AddressBookImporter
     def initialize(login, password)
       begin
         @agent = ::WWW::Mechanize.new {|a| a.log = nil }
-        @agent.user_agent = "Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)"
+        #@agent.user_agent = "Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)"
+        @agent.user_agent = "Mozilla/5.0 (X11; U; Linux i686; en; rv:1.8.1.1) Gecko/20060601 Epiphany/2.16 Firefox/2.0.0.1 (Ubuntu-edgy)"
         p = login(login, password)
         @contacts = fetch_contacts(p)
       rescue Exception => e
@@ -32,28 +34,63 @@ module AddressBookImporter
   
   class Hotmail < Importer
 
-    attr_accessor :curr_page
+    attr_accessor :curr_page, :mode
 
     def login(login, password)
-      page = @agent.get('http://www.hotmail.com')
+      page = @agent.get('http://login.live.com/login.srf?id=2')
       form = page.forms.first
       form.fields.find {|f| f.name == 'login'}.value = login
       form.fields.find {|f| f.name == 'passwd'}.value = password
       page = @agent.submit(form, form.buttons.first)
       raise LoginErrorException if page.body.match(/icon_err\.gif/)
+      
       url = page.root.at('script').inner_html.match(%r{(http:[^"]+)})[0]
-      @agent.get(url)
+      p = @agent.get(url) # http://by132w.bay132.mail.live.com/mail/mail.aspx
+      if (match = p.body.match(/(http:\/\/[^"'\/\\]*\.mail\.live\.com)/))
+        @mode = :live
+        p = @agent.get(match[1])
+      else
+        @mode = :normal
+      end
+      p
     end 
 
     def fetch_contacts(page)
-      params = page.body.match(/(curmbox=[^&]+)/)[1]
-      start = page.body.match(/(a=[^"]+)/)[1]
-      contact_page = @agent.get("/cgi-bin/AddressPicker?" + start+params + "&Context=InsertAddress&_HMaction=Edit&qF=to")
       rval = []
-      match_string = contact_page.body
-      while !(match = match_string.match(/'([^@\s]+@[-a-z0-9]+\.+[a-z]{2,})'/)).nil?
-        rval << match[1]
-        match_string = match.post_match
+      if @mode == :normal
+        params = page.body.match(/(curmbox=[^&]+)/)[1]
+        start = page.body.match(/(a=[^"]+)/)[1]
+        contact_page = @agent.get("/cgi-bin/AddressPicker?" + start+params + "&Context=InsertAddress&_HMaction=Edit&qF=to")
+        rval = get_email(contact_page.body)
+      else
+        link = page.links.select{|l|l.href.match(/Inbox/)}.first
+        p = @agent.click(link)
+        link = page.links.select{|l|l.href.match(/EditMessage/)}.first
+        if p.body.match(/NewMessageGo/)
+          if (match = p.body.match(/(\/mail\/ApplicationMainReach\.aspx\?Control=EditMessage[^"]+)/))
+            p = @agent.get(match[1])
+          end
+          form = p.forms.first
+          form.add_field!('ToContact.x', 9)
+          form.add_field!('ToContact.y', 11)   
+          p2 = form.submit
+          rval = get_email(p2.body)
+        else
+          p = @agent.click(link)
+          rval = get_email(p.body, /([_\-a-z0-9.A-Z]+%40((?:[-a-z0-9]+\.)+[a-z]{2,}))/)
+          rval.collect!{|a|a.gsub(/%40/, '@')}
+        end
+        
+        rval
+      end
+    end
+
+    protected
+    def get_email(body, match_regex = /([_\-a-z0-9.A-Z]+@((?:[-a-z0-9]+\.)+[a-z]{2,}))/)
+      rval = []
+      while !(match = body.match(match_regex)).nil?
+        rval << match[0]
+        body = match.post_match
       end
       rval
     end

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.9.0
 specification_version: 1
 name: AddressBookImporter
 version: !ruby/object:Gem::Version 
-  version: 0.0.4
-date: 2007-01-22 00:00:00 +01:00
+  version: 0.0.5
+date: 2007-02-05 00:00:00 +01:00
 summary: Mechanize scraper for address books
 require_paths: 
 - lib