Hotspots: Admin Pages | Turn-in Site |
Current Links: Cases Final Project Summer 2007
Questions on Sp2001 P6 Milestone
Ask here about P6: Parsing, websites, and more!
So we're turning in a hard-copy of the source code this time, correct?
Did the person who dreamed this up actually implement it? Did it occur to anyone that mp3.com takes measures to prevent people from doing exactly what this project requires? (running bots) -Anonymous
Hey, that's just what I was thinking too. But assuming we do perform a grand hack: should we prompt for a user's age and/or address when MP3.com asks, or just forge a cookie? Should we take into consideration that the chooser URL is designed to change periodically? Expect many more questions later. -Patrick Parker
I did a quickie implementation that searches mp3.com and returns the mp3's found. The only trouble was that the site requires cookies, and Squeak doesn't have them. You can get cookies, though, by loading Missing File (/cs2340/uploads/cookies.3.cs), and you can get some tips about approaching this problem at How to Mine Web Pages. -Lex Spoon
Thanks, Lex, but I still can't get it to work. I can parse the link to the .m3u file. When I try to access it, however, all I get is a bunch of status 302 forwarding me around click.mp3.com that sets a couple of cookies, then puts me back at the same page where I started. (I just tried Scamper and the same thing happened.) So it never gave me the m3u. Here are the contents of HTTPSocket's Cookies
Something strange is going on. When you get the link with the m3u file, try deleting everything past the $? in the URL that comes back. Mysterious, but it seems to work. -Lex Spoon
See also HTML parsing in squeak
Personally, I found it much easier to use HTMLTokenizer directly than to sift through the results of HTMLParser. ::Deposits $.02:: –Patrick Parker
Do we have to include www.mp3.com and www.listen.com in our list or can it be any sites we want to use off the web - the milestone description is a little unclear.
Ernest Kyle Oliver
I take it that the MPEG interface we have doesn't have support for streaming Mp3's over the web. What I've gotten to work is to open a socket, and download the file to my local directory, where I then play it. Is this an okay way to solve this? On a fast connection it takes only 20 or so seconds to download the entire song.
|Ernest, yes, you must include www.mp3.com and www.listen.com. Anonymous poster, yes, that sounds like the right answer. Mark Guzdial |
How do you expect us to get songs off of listen.com? All the files I have seen by searching on listen.com are stored on other web sites (amazon.com, mp3.com, lycos.com, rollingstone.com, and many more.). These links are also not direct links to the mp3 files themselves. Do you expect us to parse all these websites?
Yep. Follow the links. -Lex Spoon
You're overlooking the simple solution: Artificial Intelligence! My team is already well on the way. We call our system "SkyNet." As an added feature, it can monitor national defense systems. Just don't ask me about where I found this robotic arm... -Patrick Parker
Heheh. Seriously, there is of course a big difference in parsing particular web pages and parsing arbitrary web pages. Next people will be complaining that it's stupid to write Java compilers, because you can never write a compiler that handles every language ever thought up. -Lex Spoon
Simple question; can we get the lecture slides for the past few weeks that are not included on the book's cd please? Would be really useful to those of us that tend to be forgetful or don't have everything in their notes.
Ugh, sorry about that. Didn't understand how this whole page works. That last post looked horrible.
And what will it cost me not to do listen.com?
Check out the milestone requirements at http://coweb.cc.gatech.edu/cs2340/1057 Only 10% of the grade is on "making the match and seeing the files." Let's call that 5 points for mp3.com and 5 points for listen.com. If you do something else OTHER THAN one of those, we'll give you points, maybe 2 points. So, the cost for not doing listen.com, looks like 3 points...
does Squeak have a Morph that acts like a JPanel class in Java? See, in Java, you can put different components onto a JPanel and then switch out different panels onto the main interface. I would like to do the same thing in Squeak, but the class I'm using now (AlignmentMorph) does not look like I can arrange components on it the way I'd like to.
Link to this Page