ucnv-chainsaw 0.0.1 → 0.0.2

data/README.rdoc

@@ -12,7 +12,7 @@ A Ruby library for spidering web resources.
 == Synopsis
 
   Chainsaw.launch('http://example.com/').open { |cs|
-    cs.doc.css('#navi a')[2]
+    cs.set_next cs.doc.css('#navi a')[2]
   }.open { |cs|
     form = cs.doc.xpath('//form')[0]
     input = form.xpath('.//input[@type="text"]')[0]
@@ -32,6 +32,7 @@ A Ruby library for spidering web resources.
 
 == Installation
 
+* gem source -a http://gems.github.com
 * gem install ucnv-chainsaw
 
 == Copyright

data/Rakefile

@@ -17,7 +17,7 @@ DESCRIPTION       = "A Ruby library for spidering web resources."
 RUBYFORGE_PROJECT = "chainsaw"
 HOMEPATH          = "http://github.com/ucnv/chainsaw/tree/master"
 BIN_FILES         = %w(  )
-VERS              = "0.0.1"
+VERS              = "0.0.2"
 
 REV = File.read(".svn/entries")[/committed-rev="(d+)"/, 1] rescue nil
 CLEAN.include ['**/.*.sw?', '*.gem', '.config']
@@ -60,7 +60,7 @@ spec = Gem::Specification.new do |s|
 
   s.add_dependency('nokogiri', '>=1.2.1')
   s.add_dependency('httpclient', '>=2.1.4')
-  s.required_ruby_version = '>= 1.8.6'
+  s.required_ruby_version = '>= 1.8.7'
 
   s.files = %w(README.rdoc ChangeLog Rakefile) +
     Dir.glob("{bin,doc,test,lib,templates,generator,extras,website,script}/**/*") + 

data/examples/02_twitter.rb

@@ -1,15 +1,15 @@
-require 'rubygems'
+require 'rubygems'
 require 'chainsaw'
 
 username, password, tweet = ARGV
 
-Chainsaw.launch('https://twitter.com/home').> { |a|
+Chainsaw.launch('https://twitter.com/home').process { |a|
   f = a.doc.css('.signin').first
   f.xpath('id("username_or_email")').first['value'] = username
   f.xpath('id("session[password]")').first['value'] = password
   a.set_next f
-}.> { |a|
+}.process { |a|
   f = a.doc.css('#doingForm').first
   f.xpath('id("status")').first.content = tweet || "(o'-')-o fwip fwip"
   a.set_next f
-}.>
+}.process

data/examples/03_delicious.rb

@@ -0,0 +1,16 @@
+require 'rubygems'
+require 'chainsaw'
+
+# post to delicious.com using API
+username = ARGV.shift
+password = ARGV.shift
+url      = ARGV.shift
+desc     = ARGV.shift || ''
+tags     = ARGV.shift || ''
+
+uri = URI.parse 'https://api.del.icio.us/v1/posts/add'
+uri.query = "url=#{URI.escape url}&description=#{URI.escape desc}&tags=#{URI.escape tags}"
+
+Chainsaw(uri).set_auth(username, password).open {
+  puts doc.xpath '//result/@code'
+}

data/lib/chainsaw/browser.rb

@@ -5,7 +5,7 @@ module Chainsaw
   class Browser
     include Chainsaw::Util
     
-    DEFAULT_USER_AGENT = '' # TODO: set value
+    DEFAULT_USER_AGENT = "Chainsaw/#{VERSION}"
     
     attr_accessor :user_agent, :ignore_redirect, :hide_referer, :max_history_count, :encoding  # configurables
     attr_accessor :request_headers, :url_base, :results
@@ -138,10 +138,16 @@ module Chainsaw
     
     def document
       return @document unless @document.nil?
-      return nil if res.content.nil?
-      return nil unless is_xml_parsable?(res.contenttype)
-      enc = @encoding || Encoding.guess(res.content)
-      @document = Nokogiri.parse(res.content, nil, enc)
+      begin
+        raise 'Require HTML or XML.' unless 
+          is_xml_parsable?(res.contenttype)
+        enc = @encoding || Encoding.guess(res.content)
+        @document = Nokogiri.parse(res.content, nil, enc)
+        raise "Something wrong to parse this: #{res.content.to_s}" if 
+          @document.root.nil?
+      rescue
+        raise Chainsaw::ParseError
+      end
       
       b = @document.xpath('//base')
       base = b.empty? ? '' : b[0]['href']
@@ -168,10 +174,17 @@ module Chainsaw
   private
     
     def process_chain(&block)
-      if block.arity == 0
-        results.push yield
-      else
+      unless block.arity == -1
         results.push yield(self)
+      else
+        instance_exec(eval('self', block)) do |caller_self|
+          mm = lambda do |name, *args|
+            caller_self.__send__(name, *args)
+          end
+          self.class.__send__(:define_method, :method_missing, &mm)
+        end
+        results.push instance_eval &block
+        self.class.__send__(:undef_method, :method_missing)
       end
     end
     
@@ -184,6 +197,8 @@ module Chainsaw
       @request_count += 1
       r = begin
         @engine.send(call, uri, query, @request_headers)
+      rescue HTTPClient::BadResponseError => e
+        e.res
       rescue
         raise(
           Chainsaw::RequestError, 
@@ -256,7 +271,5 @@ module Chainsaw
       @history.slice! @max_history_count, @history.size - @max_history_count
     end
     
-    
-    
   end
 end

data/lib/chainsaw/common.rb

@@ -83,5 +83,9 @@ module Chainsaw
     include Chainsaw::ErrorWrapper
   end
   
+  class ParseError < StandardError
+    include Chainsaw::ErrorWrapper
+  end
+  
 end
 

data/lib/chainsaw/ext/httpclient.rb

@@ -22,6 +22,5 @@ class HTTPClient
   end
 end
 
-HTTP::Status::SUCCESSFUL_STATUS.push 404
 
 

data/lib/chainsaw.rb

@@ -17,21 +17,23 @@ Nokogiri::XML::Element.module_eval do
 end
 
 module Chainsaw
-  VERSION = '0.0.1'
+  VERSION = '0.0.2'
   
   #
   # Return a instance of the Chainsaw::Browser class.
   def self.launch(*args)
-    args.pop! if args.last.is_a? Proc
     cs = Chainsaw::Browser.new *args
-    yield cs if block_given?
+    if block_given?
+      block = Proc.new
+      cs.instance_eval { process_chain &block }
+    end
     cs
   end
 end
 
 module Kernel
   #
-  # alias for Chainsaw.launch
+  # alias for Chainsaw#launch
   def Chainsaw(*args)
     Chainsaw.launch *args
   end

data/test/htdocs/06.html

@@ -1,6 +1,6 @@
 <html>
   <head>
-    <title>test sjis</title>
+    <title>test euc</title>
   </head>
   <body>
     <div id="links">

data/test/htdocs/cgi.rb

@@ -32,6 +32,16 @@ if cgi.has_key? 'auth'
   end
 end
 
+if cgi.keys.find {|k| k =~ /^(\d{3})$/}
+  status = $1
+  cgi.print [
+    "Status: #{status}", 
+    "\n", 
+    "Status #{status}"
+  ].join("\n")
+  exit 0
+end
+
 if env['CONTENT_TYPE'] =~ %r{^multipart/form-data;}
   upload = cgi.params['f'][0]
   res['upload'] = {

data/test/test_browser.rb

@@ -174,6 +174,7 @@ class TestBrowser < Test::Unit::TestCase
   end
   
 =begin
+  ## this test works fine but very slow
   def test_auth
     user_pass = 'testuser:testpass'
     Chainsaw.launch(TEST_URL + 'cgi.rb?auth').
@@ -240,7 +241,7 @@ class TestBrowser < Test::Unit::TestCase
   end
   
   def test_ignore_redirect
-    
+    # TODO
   end
   
   def test_result
@@ -256,6 +257,15 @@ class TestBrowser < Test::Unit::TestCase
     
   end
   
+  def test_bad_response
+    assert_nothing_raised do 
+      Chainsaw.launch(TEST_URL + 'cgi.rb?500').open { |cs|
+        assert_equal 500, cs.res.status
+        assert_equal 'Status 500', cs.res.content
+      }
+    end
+  end
+  
   def test_aliases
     Chainsaw.launch(TEST_URL + '03.html').> { |cs|
       assert_instance_of Nokogiri::HTML::Document, cs.doc
@@ -267,7 +277,22 @@ class TestBrowser < Test::Unit::TestCase
       assert_equal @text_val, x['params']['t'][0]
       assert_equal 'go', x['params']['s'][0]
     }
+    
+  end
   
+  def test_mixed_instance
+    cs = Chainsaw.launch(TEST_URL + '01.html').open {
+      assert_instance_of Nokogiri::HTML::Document, doc
+      assert_equal 200, res.status
+      links = doc.search('//a')
+      assert_equal links.length, 5
+      set_next links[1]
+      'result1'
+    }.open {
+      assert_equal res.uri.to_s, TEST_URL + '02.html'
+      'result2'
+    }
+    assert_equal ['result1', 'result2'], cs.results
   end
   
 end

data/test/test_chainsaw.rb

@@ -12,9 +12,30 @@ class TestChainsaw < Test::Unit::TestCase
     agent2 = Chainsaw.launch('http://example.com/', {:user_agent => 'Chainsaw XXX'})
     
     assert_instance_of Chainsaw::Browser, agent1
-    assert_equal agent1.to_yaml, agent2.to_yaml
+    #assert_equal agent1.to_yaml, agent2.to_yaml
+    assert_equal agent1.user_agent, agent2.user_agent
     
   end
+  
+  def test_launch_more
+    assert_nothing_raised do
+      Chainsaw {
+        set_next 'http://example.com/'
+      }
+    end
+
+    assert_nothing_raised do
+      Chainsaw('http://example.com/')
+    end
+
+    assert_nothing_raised do
+      Chainsaw.launch('http://example.com/some') {
+        set_next 'http://example.com/'
+      }
+    end
+  end
+
+
 
 end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: ucnv-chainsaw
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors: 
 - ucnv
@@ -9,7 +9,7 @@ autorequire: ""
 bindir: bin
 cert_chain: []
 
-date: 2009-03-02 00:00:00 -08:00
+date: 2009-03-07 00:00:00 -08:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -71,6 +71,7 @@ files:
 - lib/chainsaw.rb
 - examples/01_google.rb
 - examples/02_twitter.rb
+- examples/03_delicious.rb
 has_rdoc: true
 homepage: http://github.com/ucnv/chainsaw/tree/master
 post_install_message: 
@@ -93,7 +94,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      version: 1.8.6
+      version: 1.8.7
   version: 
 required_rubygems_version: !ruby/object:Gem::Requirement 
   requirements: