Forum: Ruby scRUBYt! 0.2.8

Announcement (2017-05-07): www.ruby-forum.com is now read-only since I unfortunately do not have the time to support and maintain the forum any more. Please see rubyonrails.org/community and ruby-lang.org/en/community for other Rails- und Ruby-related community platforms.
Peter S. (Guest)
on 2007-04-19 23:50
(Received via mailing list)
This is long overdue (0.2.8 is out for about a week already), but
anyway, here we go:


============
What's this?
============

scRUBYt! is a very easy to learn and use, yet powerful Web scraping
framework based on Hpricot and mechanize. It's purpose is to free you
from the drudgery of web page crawling, looking up HTML tags,
attributes, XPaths, form names and other typical low-level web scraping
woes by figuring these out from your examples copy'n'pasted from the Web
page.

=========
CHANGELOG
=========

[NEW] download pattern: download the file pointed to by the
       parent pattern
[NEW] checking checkboxes
[NEW] basic authentication support
[NEW] default values for missing elements (basic version)
[NEW] possibility to resolve relative paths against a custom url
[NEW] first simple version of to_csv and to_hash
[NEW] complete rewrite of the exporting system (Credit: Neelance)
[NEW] first version of smart regular expressions: they are constructed
       from examples, just as regular expressions (Credit: Neelance)
[NEW] Possibility to click the n-th link
[FIX] Clicking on links using scRUBYt's advanced example lookup (i.e.
       you can use :begins_with etc.)
[NEW] Forcing writing text of non-leaf nodes with :write_text => true
[NEW] Possibility to set custom user-agent; Specified default user agent
       as Konqueror
[FIX] Fixed crawling to detail pages in case of leaving the
       original site (Credit: Michael Mazour)
[FIX] fixing the '//' problem - if the relative url contained two
       slashes, the fetching failed
[FIX] scrubyt assumed that documents have a list of nested elements
       (Credit: Rick B.)
[FIX] crawling to detail pages works also if the parent pattern is
       a string pattern
[FIX] shorcut url fixed again
[FIX] regexp pattern fixed in case it's parent was a string
[FIX] refactoring the core classes, lots of bugfixes and stabilization

=============
Misc comments
=============

As of 0.2.8, scRUBYt! depends on ParseTree and Ruby2Ruby - unfortunately
it seems ParseTree is not that trivial to set up on windows. However, we
are currently working on a new project to solve this problem, and we are
making quite good progress so I believe for the next release, 0.3.0,
this obstacle will be blown away. Until then windows users should either
install scRUBYt! on cygwin, install ParseTree somehow or use 0.2.6 until
we are ready with the Ruby bridge to ParseTree which will make the
installation on windows possible without the need to compile C.

Please continue to report problems, discuss things or give any kind of
feedback on the scRUBYt! forum at

http://agora.scrubyt.org

Cheers,
Peter - on the behalf of the scRUBYt! devel team

__
http://www.rubyrailways.com :: Ruby and Web2.0 blog
http://scrubyt.org :: Ruby web scraping framework
http://rubykitchensink.ca/ :: The indexed archive of all things Ruby.
Pit C. (Guest)
on 2007-04-20 01:16
(Received via mailing list)
Peter S. schrieb:
> As of 0.2.8, scRUBYt! depends on ParseTree and Ruby2Ruby - unfortunately
> (...)

Peter, I'm curious: could you tell me (only in a few words) how you are
using ParseTree and Ruby2Ruby?

Regards,
Pit
Peter S. (Guest)
on 2007-04-20 01:31
(Received via mailing list)
Pit C. wrote:
> Peter S. schrieb:
>> As of 0.2.8, scRUBYt! depends on ParseTree and Ruby2Ruby -
>> unfortunately (...)
>
> Peter, I'm curious: could you tell me (only in a few words) how you are
> using ParseTree and Ruby2Ruby?

Well, I guess it's the best to illustrate it with an example:

This is a learning extractor:

=========================================================================
google_data = Scrubyt::Extractor.define do

   fetch 'http://www.google.com/ncr'
   fill_textfield 'q', 'ruby'
   submit

   link "Ruby P.ming Language" do
     url "href", :type => :attribute
   end
   next_page "Next", :limit => 3
end
=========================================================================

i.e. it works for just the first page of google, for the query 'ruby'.
To create a generalized extractor which can be used on any google result
page, we have to export it after it 'learned' how to do this on the
given example. Since I would like this so called production extractor to
resemble the original as much as possible, I am using Ruby2Ruby and
ParseTree. With them I can get this result:

=========================================================================
google_data = Scrubyt::Extractor.define do
   fetch("http://www.google.com/ncr")
   fill_textfield("q", "ruby")
   submit

   link("/html/body/div/div/a") do
     url("href", { :type => :attribute })
   end

   next_page("Next", { :limit => 3 })
end
=========================================================================

If you can tell me any other way to achieve this (originally I took the
source code of the learning extractors and replaced the examples with
XPaths and did any other modifications required - but it was a mess
after some time) I would be really thankful.

There is a disagreement in the development team about this, too - one
viewpoint is that a dependency on parsetree and ruby2ruby costs us too
much (mainly on windows), and the other is that this problem  has to be
solved then we can depend on these packages....

What would you suggest?

Cheers,
Peter
__
http://www.rubyrailways.com :: Ruby and Web2.0 blog
http://scrubyt.org :: Ruby web scraping framework
http://rubykitchensink.ca/ :: The indexed archive of all things Ruby.
Pit C. (Guest)
on 2007-04-20 02:15
(Received via mailing list)
Peter S. schrieb:
> Pit C. wrote:
>> Peter, I'm curious: could you tell me (only in a few words) how you
>> are using ParseTree and Ruby2Ruby?
>
> Well, I guess it's the best to illustrate it with an example:
> (...)

Thank you for the example. Lets see if I understand what you're doing:
you have a given Ruby code (the learning extractor) and need to
manipulate it according to certain rules to create a modified version of
the code (the generalized extractor).

> If you can tell me any other way to achieve this (originally I took the
> source code of the learning extractors and replaced the examples with
> XPaths and did any other modifications required - but it was a mess
> after some time) I would be really thankful.

Well, in your example it looks like you use nothing but the Scrubyt DSL.
In this case you could capture the calling sequence and the method
arguments when you execute the DSL. But if you want to allow normal Ruby
code in the #define block (loops, conditionals, etc), then I can't think
of other solutions.

> There is a disagreement in the development team about this, too - one
> viewpoint is that a dependency on parsetree and ruby2ruby costs us too
> much (mainly on windows), and the other is that this problem  has to be
> solved then we can depend on these packages....
>
> What would you suggest?

If I really wanted to support Windows users, I'd try to compile
ParseTree with MinGW and/or convince the maintainers to provide a binary
version of the gem :-)

But my question wasn't meant as a recommendation not to use ParseTree.
I'm simply interested in use cases for working with the Ruby AST.

Regards,
Pit
Peter S. (Guest)
on 2007-04-20 02:51
(Received via mailing list)
Hallo Pit,

> Thank you for the example. Lets see if I understand what you're doing:
> you have a given Ruby code (the learning extractor) and need to
> manipulate it according to certain rules to create a modified version of
> the code (the generalized extractor).

Exactly!

> Well, in your example it looks like you use nothing but the Scrubyt DSL.
> In this case you could capture the calling sequence and the method
> arguments when you execute the DSL. But if you want to allow normal Ruby
> code in the #define block (loops, conditionals, etc), then I can't think
> of other solutions.

Yes, we are just adding this possibility (so you can do branching etc.
with native Ruby code - a DSL is fine but it always has its limits,
whereas Ruby doesn't :-) - so in the long run we have to do this (or
something equally powerful).

> If I really wanted to support Windows users,
Yeah, we surely do! A lot of them are using scRUBYt!, too...

> I'd try to compile
> ParseTree with MinGW and/or convince the maintainers to provide a binary
> version of the gem :-)
Well, I don't really understand why aren't they doing this anyway, since
their windows users have the same trouble as we do (actually we have it
just because of ParseTree). This is a viable alternative if others do
not work out (mainly the next paragraph).

ATM one team member is working on a solution called ParseTreeReloaded
which will wrap pure Ruby code around ParseTree so no C compiling will
be needed. ATM ParseTreeReloaded can already parse its own source code
so I guess he's making some great progress... Let's see.

> But my question wasn't meant as a recommendation not to use ParseTree.
> I'm simply interested in use cases for working with the Ruby AST.
Yeah sure, I am also in the don't-drop-parsetree camp... however if we
can not solve this problem permanently under windows (which I am 99%
positive we can) I'll have to look for a different solutions because of
the win32 ppl...

Cheers,
Peter
__
http://www.rubyrailways.com :: Ruby and Web2.0 blog
http://scrubyt.org :: Ruby web scraping framework
http://rubykitchensink.ca/ :: The indexed archive of all things Ruby.
This topic is locked and can not be replied to.