Yay! Hpricot totally rocks the Windows. Thanks for the speedy work on that!

VM/370 ONLINE
            VV        VV    MM        MM
            VV        VV    MMM      MMM
            VV        VV    MMMM    MMMM
            VV        VV    MM MM  MM MM
     3333333333     777777777777MMMM  00000000
   333333333333    77777777777  MM  0000000000
    33      VV33    77VV    77      00MM      00
             V33     VV    77M      00MM      00
              33    VV    77MM      00MM      00
           3333VV  VV    77 MM      00MM      00
           3333 VVVV     77 MM      00MM      00
              33 VV      77 MM      00MM      00
              33         77         00        00
    33        33         77         00        00
    333333333333         77          0000000000
     3333333333          77           00000000

I mean that’s alot better than putting together a VM and trying to track down a decent free Microsoft compiler. A couple weeks ago, I spent six hours on it and made no progress.

Is there something wrong with Microsoft’s own C++ compiler (on Windows, that is, of course)? The toolkit is free as in beer.

These win32 binaries provided by the puny MinGW toolchain, which doesn’t stand comparison with the almighty MS VC2005 Express, but empowers non-win32 developers to build binaries for their win32 users.

More details about the cross-compilation magic in the Rakefile, and here .

Asztal: First of all, it runs on win32 only :-) Second, and most importantly: it’s not binary-compatible with the VC6 ruby build (I read now that MinGW isn’t fully either, but it works for all but a few extensions).

Does Hpricot not support [] in an Xpath? I know I can index the elements from the resultant ruby array following a search, but I have to believe it could be faster using an internal lookup to an internal C data structure via an Xpath string.

irb(main):008:0> Hpricot(URI.parse("http://feeds.feedburner.com/vedana").read)
/usr/lib/ruby/gems/1.8/gems/hpricot-0.3/lib/hpricot/parse.rb:164:in `build_node': 
[bug] unknown structure: 
[:xmlprocins, "href=\"http://feeds.feedburner.com/~d/styles/itemcontent.css\" 
type=\"text/css\" media=\"screen", nil, nil] (Exception)
 from /usr/lib/ruby/gems/1.8/gems/hpricot-0.3/lib/hpricot/parse.rb:59:in `make' 
 from /usr/lib/ruby/gems/1.8/gems/hpricot-0.3/lib/hpricot/parse.rb:59:in `make'
 from /usr/lib/ruby/gems/1.8/gems/hpricot-0.3/lib/hpricot/parse.rb:11:in `parse'
 from /usr/lib/ruby/gems/1.8/gems/hpricot-0.3/lib/hpricot/parse.rb:4:in `Hpricot'
 from (irb):8:in `irb_binding'
 from /usr/lib/ruby/1.8/irb/workspace.rb:52:in `irb_binding' 
 from /usr/lib/ruby/1.8/irb/workspace.rb:52

hoyhoy: Thankyou for that. Is fixed in SVN , the feed parses fine.

The thing is: I’m not keeping any internal C data structures here. It’s usually the parsing that really slows you down, not the data structures. It would be faster to do it all in C, but then it would be something that I couldn’t possibly finish.

That makes supporting [] in XPath kind of moot. I’ve often considered making my own whizbang XML parser gem using flex/bison or spirit. Unfortunately, after having that kind of thought, I usually sit down and eat a sandwich and it goes away.

Haha, oh yeah dreaming about scanners and parsers is really fun, it’s too bad the reality is so vomitously bleak.

Well, maybe we should make that not so. :) (HAH)

I am the proud owner of a 2006 Win32 Sandbox, Unsafe Edition, with leather seats.

irb(main):035:0> Hpricot(URI.parse("http://rss.slashdot.org/Slashdot/slashdot").read).search("item")[1].to_html.each_line { |l| puts l if l.match(/dc:date/) }; nil<dc:date>2006-07-22T19:29:00+00:00</dc:date>=> nilirb(main):036:0> puts Hpricot(URI.parse("http://rss.slashdot.org/Slashdot/slashdot").read).search("dc:date")[1]nil=> nil

irb(main):035:0> Hpricot(URI.parse("http://rss.slashdot.org/Slashdot/slashdot").read).search("item")[1].to_html.each_line
 { |l| puts l if l.match(/dc:date/) }; nil
<dc:date>2006-07-22T19:29:00+00:00</dc:date>
=> nil
irb(main):036:0> puts Hpricot(URI.parse("http://rss.slashdot.org/Slashdot/slashdot").read).search("dc:date")[1]
nil
=> nil

m = expr.match %r!^([#.]?)([a-z0-9\\*_:-]*)!i

Line 102 in traverse.rb, but you knew that.

... it’s not binary-compatible with the VC6 ruby build…

Dear God, why do people still use VC6 ?

Capn: Because it’s become a de facto standard, and tends to have good windows compatibility.

hoyhoy: You have a sneaky smiley man in your regexp! :-]

this is great! but i have a little problem: is v0.3 missing something? Container::Trav.filter calls to_node.subst_subnode on line 402, but that doesn’t seem to exist?

FlashHater: However, VC7 has the same perfect level of windows compatibility AND is one of the most standards compliant compilers on the market. Essentially, with the availability of VC7 , there is no reason to use VC6 .

So am I using hpricot wrong or is this a bug?

doc = Hpricot(open(“http://usgenweb.org/”)) puts (doc/:a).length # => 15 but it should be 66

also the follow errors out

doc.to_s

Hpricot doesn’t support this:

doc.search("//table/tr/td[3]")

... does it?

I’m trying to fetch the 3rd td in each tr.

bearik, this should work

 doc.search("table tr td:nth(3)")

There’s still some unimplemented XPath. Refer to the supported CSS selectors for other ideas.