Web Crawler/Robot to collect files/data

I am looking for a research application/robot to login, search using the website's built in search engine, collect information and download files on a daily bases. This application must mimic human behavior (keystokes, timing, etc). It must be easily updatable in the future to account for changes in the website's functionality. It must also avoid downloading duplicate files and information. The application will run on a Windows XP Server from our office as should allow for statistics reporting, easy Boolean logic search string updates, multiple searches, chron start and stops and web based remote access/management. I am open to suggestions on the application language, but would probably prefer Java.