first, in your message, you either forgot the limits of what you want to retrieve or it did not stay in. It helps to have a complete message if you want some help ;)
second, I had a look at your pipe, and the problem is pretty clear: the regex module tells you that the content of 'content' is too long. You feed it the whole page or so, it's not really happy with it.
Either get a better cut with the fetch page module or use a simpler function first instead of the tank that is regex, like a substring in a loop operator as I did there: http://pipes.yahoo.com/luneart/a3d83624aa205c2177d603b9bfb9764a
Thanks for this trick Lolo !
in fact I succeeded to extract the url between the h3 tags for the first item before yesterday, then I fetched it to extract the xml file in the page of the song (containing only one item describing the song).
But in fact, I couldn't extract the url pointing to this xml because it's between tags, and either Fetch page or XFetch could see it... So unless there is another trick here (1), I will have to build the items one by one for the output RSS I'm trying to build.
And I thought to something else : the page contains only the last 10 favorites of the user, so if the user likes more than 10 songs in a very short time, it's very likely some songs will be missed because in the next page. And I don't know if its possible to build Pipes that would be able to detect if all the song are new, and if so, then open the next page (2). Anyway, I won't be able to create these Pipes in reasonable time, like you might guess ^^
I thought to another solution here, which would be to build something with the soundcloud API, but it's way beyond my capacities !
So unless there is a solution for points (1) and (2), this project of mine is dead :)
I'm going to address your (2) then (1) (so you try to play with it and master it :P) then propose a simpler solution (also, grrr.).
You would be able to check if all/most/a certain number of items are within a time-range, but I don't see how to fetch a "2nd" page (but I just dabble with pipes contrary to other users I'm sure, so maybe someone else has a solution!).
(1) : Now, for the fetching page from with an item, it's harder but I think you can do it with the YQL module which will understand the url even if it's inputed as item.link. However I don't know this module so I con't really help you.
Better solution: you could use the "feed auto discovery" module on your first url, to find an rss feed of the user, solving (2) and probably (1).