Bot/Captcha-OCR for facebook-clone StudiVZ

Bot/Captcha-ORC, either Windows based or for linux webserver (GoDaddy hosting): Create a bot for the german facebook clone "StudiVZ" that extracts profile IDs from group member lists, adds them to an invite-link, calls up one of these invite-links after another and resolves the upcoming captcha. ## Deliverables **The software has to run** a) either on a windows xp machine (user pc) or b) on a webserver. The requirements therefore would be: * *has to run using web hosting packages as they can be ordered at hosters like [login to view URL] for less than 10 USD per month (php, mysql available)* * *has to make use of some kind of IP anonymizer (proxy) so the script can not be blocked, even if it's running from a server with a static ip address.* With your bidding please let me know which solution you want to deliver. **What the script will have to do is the following:** A. The script has to ask the user for the following data: * login name and password for StudiVZ account that shall be used * ID of the group that the script will have to invite people to ("targetgroup") * ID of the groups that the script will have to spider for user IDs ("sourcegroups") B. It then has to do the following on its own: 1. Spider for user IDs in the defined groups * call up the group by using a template link ([login to view URL][sourcegroupidhere]) * flipping through the user list pages * extracting the user IDs from links with the pattern [login to view URL][useridhere] 2. Use one of the collected user IDs after another in order to call up an invitation to the defined group: * call up link with template [login to view URL][targetgroupidhere]/[useridehere] * solve the upcoming captcha * repeat this until all collected IDs have been processed. Ideas for resolving the captcha: * there are only 3 attempts for resolving a captcha before the user will be logged out automatically * unreadable captchas can be replaced by a new one as often as you wish (button "Neuen Code generieren") * therefore it might be useful if the captcha-ocr-script would determine a degree of readability of the shown captcha first. if the letters are readably with accuracy, it could read them. if not, it could call up a new captcha image and start again. * the balance between recognition accuracy (script calls up many captchas in order to make sure it uses the easiest one) and speed (script tries to solve captchas on the first try - and therefore sometimes fails 3 times in a row which results in a log out) should be either automatically optimized or at least user-definable. the goals is that the user has to interfere as little as possible, so the script can run over night on its own. * if the script gets logged out because of failing to read the captcha it should login again and resume its work 3. Show a result screen with the following data: * how many invitations have been sent? * how many times did the script get logged out because of captcha ocr failure? * (more statistics about captcha recognition and reloading only if it's convenient) For testing you can use the following data: * login user: [login to view URL]@[login to view URL] * login pass: password * target group id: 1f65ca2a7c8064a5 * example source group link: [login to view URL] An example captcha image is attached to my bid request. Here is also a direct link to one: [login to view URL]