What TokersBall_CDXX has presented is a parser pretty much. It will parse the html of the document and then using Regex he found a pattern of urls which contain the "http://" beginning and fished them out. (They were in between < and > probably). He compiled them in an array and then you can get the urls from that array .