RubyGems - hpricot - Versions diffs - 0.5-mswin32 → 0.6-mswin32 - Mend

hpricot 0.5-mswin32 → 0.6-mswin32

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (31) hide show

data/CHANGELOG +13 -1
data/README +4 -1
data/Rakefile +111 -75
data/ext/hpricot_scan/HpricotScanService.java +1340 -0
data/ext/hpricot_scan/hpricot_common.rl +76 -0
data/ext/hpricot_scan/hpricot_scan.c +2435 -2181
data/ext/hpricot_scan/hpricot_scan.java.rl +363 -0
data/ext/hpricot_scan/hpricot_scan.rl +3 -70
data/lib/hpricot.rb +1 -0
data/lib/hpricot/blankslate.rb +63 -0
data/lib/hpricot/builder.rb +200 -0
data/lib/hpricot/elements.rb +143 -32
data/lib/hpricot/inspect.rb +7 -3
data/lib/hpricot/parse.rb +128 -101
data/lib/hpricot/tag.rb +23 -15
data/lib/hpricot/tags.rb +164 -0
data/lib/hpricot/traverse.rb +93 -33
data/lib/hpricot/xchar.rb +94 -0
data/lib/i686-linux/hpricot_scan.so +0 -0
data/test/files/pace_application.html +1320 -0
data/test/files/tenderlove.html +16 -0
data/test/files/why.xml +19 -0
data/test/load_files.rb +1 -1
data/test/test_alter.rb +65 -0
data/test/test_builder.rb +24 -0
data/test/test_parser.rb +92 -4
data/test/test_preserved.rb +20 -0
data/test/test_xml.rb +13 -0
metadata +34 -14
data/lib/hpricot/text.rb +0 -115
data/lib/hpricot_scan.so +0 -0

data/test/files/tenderlove.html ADDED

@@ -0,0 +1,16 @@
+<html>
+<HEAD>
+  <meta http-equiv="Refresh" content="0; url=http://tenderlovemaking.com">
+  <META http-equiv="Refresh" content="0; url=http://tenderlovemaking.com">
+</head>
+<body>
+  <a href ="http://tenderlovemaking.com/">My Site!</a>
+  <A href ="http://whytheluckystiff.net/">Your Site!</A>
+<MAP>
+    <area HREF="http://whytheluckystiff.net/" COORDS="1,2,3,4"></area>
+    <AREA HREF="http://tenderlovemaking.com/" COORDS="1,2,3,4">
+    </area>
+    <AREA HREF="http://tenderlovemaking.com/" COORDS="5,5,10,10" />
+  </MAP>
+  </body>
+</html>

data/test/files/why.xml ADDED

@@ -0,0 +1,19 @@
+<?xml version='1.0'?><rss xmlns:admin='http://webns.net/mvcb/' version='2.0' xmlns:sy='http://purl.org/rss/1.0/modules/syndication/' xmlns:dc='http://purl.org/dc/elements/1.1/' xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'>
+    <channel>
+    <title>why the lucky stiff</title>
+    <link>http://whytheluckystiff.net</link>
+    <description>hex-editing reality to give us infinite grenades!!</description>
+    <dc:language>en-us</dc:language>
+    <dc:creator/>
+    <dc:date>2007-01-16T22:39:04+00:00</dc:date>
+    <admin:generatorAgent rdf:resource='http://hobix.com/?v=0.4'/>
+    <sy:updatePeriod>hourly</sy:updatePeriod>
+    <sy:updateFrequency>1</sy:updateFrequency>
+    <sy:updateBase>2000-01-01T12:00+00:00</sy:updateBase>
+    <item><title>1.3</title><link>http://whytheluckystiff.net/quatrains/1.3.html</link><guid isPermaLink='false'>quatrains/1.3@http://whytheluckystiff.net</guid><dc:subject>quatrains</dc:subject><dc:subject>quatrains</dc:subject><dc:creator>why the lucky stiff</dc:creator><dc:date>2007-01-14T08:47:05+00:00</dc:date><description>&lt;blockquote&gt;
+	&lt;p&gt;That cadillac of yours and that driver of yours!&lt;br /&gt;You and your teacups rattling away in the back seat!&lt;br /&gt;You always took the mike, oh, and all those cowboys you shot!&lt;br /&gt;I held your hand! And I&amp;#8217;ll shoot a cowboy one day!&lt;/p&gt;
+	&lt;/blockquote&gt;
+	&lt;blockquote&gt;
+	&lt;p&gt;You said, &amp;#8220;Let&amp;#8217;s run into the woods like kids!&amp;#8221; &lt;br /&gt;You said, &amp;#8220;Let&amp;#8217;s rub our hands together super-hot!&amp;#8221; &lt;br /&gt;And we scalded the trees and left octagons, I think that was you and&lt;br /&gt;You threw parties on the roof!&lt;/p&gt;
+	&lt;/blockquote&gt;</description></item></channel>
+</rss>

data/test/load_files.rb CHANGED

@@ -1,6 +1,6 @@
 module TestFiles
     Dir.chdir(File.dirname(__FILE__)) do
-        Dir['files/*.{html,xhtml}'].each do |fname|
+        Dir['files/*.{html,xhtml,xml}'].each do |fname|
             const_set fname[%r!/(\w+)\.\w+$!, 1].upcase, IO.read(fname)
         end
     end

data/test/test_alter.rb ADDED

@@ -0,0 +1,65 @@
+#!/usr/bin/env ruby
+require 'test/unit'
+require 'hpricot'
+require 'load_files'
+class TestAlter < Test::Unit::TestCase
+  def setup
+    @basic = Hpricot.parse(TestFiles::BASIC)
+  end
+  def test_before
+    test0 = "<link rel='stylesheet' href='test0.css' />"
+    @basic.at("link").before(test0)
+    assert_equal 'test0.css', @basic.at("link").attributes['href']
+  end
+  def test_after
+    test_inf = "<link rel='stylesheet' href='test_inf.css' />"
+    @basic.search("link")[-1].after(test_inf)
+    assert_equal 'test_inf.css', @basic.search("link")[-1].attributes['href']
+  end
+  def test_wrap
+    ohmy = (@basic/"p.ohmy").wrap("<div id='wrapper'></div>")
+    assert_equal 'wrapper', ohmy[0].parent['id']
+    assert_equal 'ohmy', Hpricot(@basic.to_html).at("#wrapper").children[0]['class']
+  end
+  def test_add_class
+    first_p = (@basic/"p:first").add_class("testing123")
+    assert first_p[0].get_attribute("class").split(" ").include?("testing123")
+    assert (Hpricot(@basic.to_html)/"p:first")[0].attributes["class"].split(" ").include?("testing123")
+    assert !(Hpricot(@basic.to_html)/"p:gt(0)")[0].attributes["class"].split(" ").include?("testing123")
+  end
+  def test_change_attributes
+    all_ps = (@basic/"p").attr("title", "Some Title")
+    all_as = (@basic/"a").attr("href", "http://my_new_href.com")
+    all_lb = (@basic/"link").attr("href") { |e| e.name }
+    assert_changed(@basic, "p", all_ps) {|p| p.attributes["title"] == "Some Title"}
+    assert_changed(@basic, "a", all_as) {|a| a.attributes["href"] == "http://my_new_href.com"}
+    assert_changed(@basic, "link", all_lb) {|a| a.attributes["href"] == "link" }
+  end
+  def test_remove_attr
+    all_rl = (@basic/"link").remove_attr("href")
+    assert_changed(@basic, "link", all_rl) { |link| link['href'].nil? }
+  end
+  def test_remove_class
+    all_c1 = (@basic/"p[@class*='last']").remove_class("last")
+    assert_changed(@basic, "p[@class*='last']", all_c1) { |p| p['class'] == 'final' }
+  end
+  def test_remove_all_classes
+    all_c2 = (@basic/"p[@class]").remove_class
+    assert_changed(@basic, "p[@class]", all_c2) { |p| p['class'].nil? }
+  end
+  def assert_changed original, selector, set, &block
+    assert set.all?(&block)
+    assert Hpricot(original.to_html).search(selector).all?(&block)
+  end
+end

data/test/test_builder.rb ADDED

@@ -0,0 +1,24 @@
+#!/usr/bin/env ruby
+require 'test/unit'
+require 'hpricot'
+class TestBuilder < Test::Unit::TestCase
+  def test_escaping_text
+    doc = Hpricot() { b "<a\"b>" }
+    assert_equal "<b>&lt;a&quot;b&gt;</b>", doc.to_html
+    assert_equal %{<a"b>}, doc.at("text()").to_s
+  end
+  def test_no_escaping_text
+    doc = Hpricot() { div.test.me! { text "<a\"b>" } }
+    assert_equal %{<div class="test" id="me"><a"b></div>}, doc.to_html
+    assert_equal %{<a"b>}, doc.at("text()").to_s
+  end
+  def test_latin1_entities
+    doc = Hpricot() { b "\200\225" }
+    assert_equal "<b>&#8364;&#8226;</b>", doc.to_html
+    assert_equal "\342\202\254\342\200\242", doc.at("text()").to_s
+  end
+end

data/test/test_parser.rb CHANGED

@@ -23,6 +23,19 @@ class TestParser < Test::Unit::TestCase
     assert_equal 'FOO', Hpricot.make("FOO").first.content
   end
+  def test_filter_by_attr
+    @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
+    # this link is escaped in the doc
+    link = 'http://www.youtube.com/watch?v=TvSNXyNw26g&search=chris%20ware'
+    assert_equal link, @boingboing.at("a[@href='#{link}']")['href']
+  end
+  def test_filter_contains
+    @basic = Hpricot.parse(TestFiles::BASIC)
+    assert_equal '<title>Sample XHTML</title>', @basic.search("title:contains('Sample')").to_s
+  end
   def test_get_element_by_id
     @basic = Hpricot.parse(TestFiles::BASIC)
     assert_equal 'link1', @basic.get_element_by_id('link1')['id']
@@ -84,6 +97,12 @@ class TestParser < Test::Unit::TestCase
     assert_equal "<p>one</p>", h.search("//div/p:first()").to_s
   end
+  def test_pace
+    doc = Hpricot(TestFiles::PACE_APPLICATION)
+    assert_equal 'get', doc.at('form[@name=frmSect11]')['method']
+    # assert_equal '2', doc.at('#hdnSpouse')['value']
+  end
   def test_scan_boingboing
     @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
     assert_equal 60, (@boingboing/'p.posted').length
@@ -95,7 +114,7 @@ class TestParser < Test::Unit::TestCase
     assert_equal 60, @boingboing.search("h3").length
     assert_equal 59, @boingboing.search("h3[text()!='College kids reportedly taking more smart drugs']").length
     assert_equal 17, @boingboing.search("h3[text()$='s']").length
-    assert_equal 128, @boingboing.search("p[text()]").length
+    assert_equal 129, @boingboing.search("p[text()]").length
     assert_equal 211, @boingboing.search("p").length
   end
@@ -135,8 +154,7 @@ class TestParser < Test::Unit::TestCase
     assert_equal 60, @boingboing.search("/*/body//p[@class='posted']").length
     assert_equal 18, @boingboing.search("//script").length
     divs = @boingboing.search("//script/../div")
-    assert_equal 2,  divs.length
-    assert_equal 1,  divs.search('a').length
+    assert_equal 1,  divs.length
     imgs = @boingboing.search('//div/p/a/img')
     assert_equal 15, imgs.length
     assert_equal 17, @boingboing.search('//div').search('p/a/img').length
@@ -155,9 +173,16 @@ class TestParser < Test::Unit::TestCase
     assert_equal 1, @boingboing.search('//input[@checked]').length
   end
+  def test_tag_case
+    @tenderlove = Hpricot.parse(TestFiles::TENDERLOVE)
+    assert_equal 2, @tenderlove.search('//a').length
+    assert_equal 3, @tenderlove.search('//area').length
+    assert_equal 2, @tenderlove.search('//meta').length
+  end
   def test_alt_predicates
     @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
-    assert_equal 2, @boingboing.search('//table/tr:last').length
+    assert_equal 1, @boingboing.search('//table/tr:last').length
     @basic = Hpricot.parse(TestFiles::BASIC)
     assert_equal "<p>The third paragraph</p>",
@@ -167,6 +192,22 @@ class TestParser < Test::Unit::TestCase
     assert_equal 'last final', @basic.search('//p:last-of-type').first.get_attribute('class')
   end
+  def test_insert_after # ticket #63
+    doc = Hpricot('<html><body><div id="a-div"></div></body></html>')
+    (doc/'div').each do |element|
+      element.after('<p>Paragraph 1</p><p>Paragraph 2</p>')
+    end
+    assert_equal doc.to_html, '<html><body><div id="a-div"></div><p>Paragraph 1</p><p>Paragraph 2</p></body></html>'
+  end
+  def test_insert_before # ticket #61
+    doc = Hpricot('<html><body><div id="a-div"></div></body></html>')
+    (doc/'div').each do |element|
+      element.before('<p>Paragraph 1</p><p>Paragraph 2</p>')
+    end
+    assert_equal doc.to_html, '<html><body><p>Paragraph 1</p><p>Paragraph 2</p><div id="a-div"></div></body></html>'
+  end
   def test_many_paths
     @boingboing = Hpricot.parse(TestFiles::BOINGBOING)
     assert_equal 62, @boingboing.search('p.posted, link[@rel="alternate"]').length
@@ -264,6 +305,53 @@ class TestParser < Test::Unit::TestCase
     end
   end
+  def test_youtube_attr
+    str = <<-edoc
+    <html><body>
+    Lorem ipsum. Jolly roger, ding-dong sing-a-long
+    <object width="425" height="350">
+      <param name="movie" value="http://www.youtube.com/v/NbDQ4M_cuwA"></param>
+      <param name="wmode" value="transparent"></param>
+        <embed src="http://www.youtube.com/v/NbDQ4M_cuwA"
+          type="application/x-shockwave-flash" wmode="transparent" width="425" height="350">
+        </embed>
+    </object>
+    Check out my posting, I have bright mice in large clown cars.
+    <object width="425" height="350">
+      <param name="movie" value="http://www.youtube.com/v/foobar"></param>
+      <param name="wmode" value="transparent"></param>
+        <embed src="http://www.youtube.com/v/foobar"
+          type="application/x-shockwave-flash" wmode="transparent" width="425" height="350">
+        </embed>
+    </object>
+    </body></html?
+    edoc
+    doc = Hpricot(str)
+    assert_equal "http://www.youtube.com/v/NbDQ4M_cuwA",
+      doc.at("//object/param[@value='http://www.youtube.com/v/NbDQ4M_cuwA']")['value']
+  end
+  # ticket #84 by jamezilla
+  def test_screwed_xmlns
+    doc = Hpricot(<<-edoc)
+      <?xml:namespace prefix = cwi />
+      <html><body>HAI</body></html>
+    edoc
+    assert_equal "HAI", doc.at("body").inner_text
+  end
+  # Reported by Jonathan Nichols on the Hpricot list (24 May 2007)
+  def test_self_closed_form
+    doc = Hpricot(<<-edoc)
+      <body>
+      <form action="/loginRegForm" name="regForm" method="POST" />
+      <input type="button">
+      </form>
+      </body>
+    edoc
+    assert_equal "button", doc.at("//form/input")['type']
+  end
   def test_filters
     @basic = Hpricot.parse(TestFiles::BASIC)
     assert_equal 0, (@basic/"title:parent").size

data/test/test_preserved.rb CHANGED

@@ -38,9 +38,29 @@ class TestPreserved < Test::Unit::TestCase
     end
   end
+  def test_escaping_of_contents
+    doc = Hpricot(TestFiles::BOINGBOING)
+    assert_equal "Fukuda\342\200\231s Automatic Door opens around your body as you pass through it. The idea is to save energy and keep the room clean.", doc.at("img[@alt='200606131240']").next.to_s.strip
+  end
   def test_files
     assert_roundtrip TestFiles::BASIC
     assert_roundtrip TestFiles::BOINGBOING
     assert_roundtrip TestFiles::CY0
   end
+  def test_escaping_of_attrs
+    # ampersands in URLs
+    str = %{<a href="http://google.com/search?q=hpricot&amp;l=en">Google</a>}
+    link = (doc = Hpricot(str)).at(:a)
+    assert_equal "http://google.com/search?q=hpricot&l=en", link['href']
+    assert_equal "http://google.com/search?q=hpricot&l=en", link.attributes['href']
+    assert_equal "http://google.com/search?q=hpricot&l=en", link.get_attribute('href')
+    assert_equal "http://google.com/search?q=hpricot&amp;l=en", link.raw_attributes['href']
+    assert_equal str, doc.to_html
+    # alter the url
+    link['href'] = "javascript:alert(\"AGGA-KA-BOO!\")"
+    assert_equal %{<a href="javascript:alert(&quot;AGGA-KA-BOO!&quot;)">Google</a>}, doc.to_html
+  end
 end

data/test/test_xml.rb CHANGED

@@ -12,4 +12,17 @@ class TestParser < Test::Unit::TestCase
     assert_equal "this is title", (doc/:rss/:channel/:title).text
     assert_equal "http://fake.com", (doc/:rss/:channel/:link).text
   end
+  # make sure XML doesn't get downcased
+  def test_casing
+    doc = Hpricot::XML(TestFiles::WHY)
+    assert_equal "hourly", (doc.at "sy:updatePeriod").inner_html
+    assert_equal 1, (doc/"guid[@isPermaLink]").length
+  end
+  # be sure tags named "text" are ok
+  def test_text_tags
+    doc = Hpricot::XML("<feed><title>City Poisoned</title><text>Rita Lee has poisoned Brazil.</text></feed>")
+    assert_equal "City Poisoned", (doc/"title").text
+  end
 end

metadata CHANGED

@@ -3,10 +3,11 @@ rubygems_version: 0.9.0
 specification_version: 1
 name: hpricot
 version: !ruby/object:Gem::Version
-  version: "0.5"
-date: 2007-01-31 00:00:00 -08:00
+  version: "0.6"
+date: 2007-06-15 00:00:00 -07:00
 summary: a swift, liberal HTML parser with a fantastic library
 require_paths:
+- lib/i686-linux
 - lib
 email: why@ruby-lang.org
 homepage: http://code.whytheluckystiff.net/hpricot/
@@ -15,7 +16,7 @@ description: a swift, liberal HTML parser with a fantastic library
 autorequire:
 default_executable:
 bindir: bin
-has_rdoc: false
+has_rdoc: true
 required_ruby_version: !ruby/object:Gem::Version::Requirement
   requirements:
   - - ">"
@@ -29,41 +30,60 @@ post_install_message:
 authors:
 - why the lucky stiff
 files:
+- CHANGELOG
+- COPYING
+- README
+- Rakefile
+- test/files
 - test/test_preserved.rb
 - test/test_paths.rb
 - test/load_files.rb
 - test/test_xml.rb
 - test/test_parser.rb
+- test/test_alter.rb
+- test/test_builder.rb
+- test/files/why.xml
 - test/files/boingboing.html
 - test/files/uswebgen.html
 - test/files/immob.html
 - test/files/week9.html
 - test/files/utf8.html
-- test/files/cy0.html
 - test/files/basic.xhtml
+- test/files/cy0.html
+- test/files/tenderlove.html
+- test/files/pace_application.html
+- lib/hpricot
 - lib/hpricot.rb
+- lib/i686-linux
+- lib/hpricot/builder.rb
 - lib/hpricot/htmlinfo.rb
-- lib/hpricot/text.rb
+- lib/hpricot/xchar.rb
 - lib/hpricot/inspect.rb
 - lib/hpricot/modules.rb
 - lib/hpricot/parse.rb
 - lib/hpricot/tag.rb
 - lib/hpricot/traverse.rb
 - lib/hpricot/elements.rb
+- lib/hpricot/tags.rb
+- lib/hpricot/blankslate.rb
+- extras/mingw-rbconfig.rb
+- ext/hpricot_scan/hpricot_scan.h
+- ext/hpricot_scan/HpricotScanService.java
 - ext/hpricot_scan/hpricot_scan.c
 - ext/hpricot_scan/extconf.rb
-- ext/hpricot_scan/hpricot_scan.h
+- ext/hpricot_scan/hpricot_common.rl
 - ext/hpricot_scan/hpricot_scan.rl
-- CHANGELOG
-- README
-- Rakefile
-- COPYING
-- extras/mingw-rbconfig.rb
-- lib/hpricot_scan.so
+- ext/hpricot_scan/hpricot_scan.java.rl
+- lib/i686-linux/hpricot_scan.so
 test_files: []
-rdoc_options: []
+rdoc_options:
+- --quiet
+- --title
+- The Hpricot Reference
+- --main
+- README
+- --inline-source
 extra_rdoc_files:
 - README
 - CHANGELOG

data/lib/hpricot/text.rb DELETED

@@ -1,115 +0,0 @@
-require 'hpricot/modules'
-require 'hpricot/raw_string'
-require 'hpricot/htmlinfo'
-require 'hpricot/encoder'
-require 'hpricot/fstr'
-require 'iconv'
-module Hpricot
-  class Text
-    # :stopdoc:
-    class << self
-      alias new_internal new
-    end
-    # :startdoc:
-    def Text.new(arg)
-      arg = arg.to_node if Hpricot::Location === arg
-      if Text === arg
-        new_internal arg.rcdata, arg.normalized_rcdata
-      elsif String === arg
-        arg2 = arg.gsub(/&/, '&amp;')
-        arg = arg2.freeze if arg != arg2
-        new_internal arg
-      else
-        raise TypeError, "cannot initialize Text with #{arg.inspect}"
-      end
-    end
-    def initialize(rcdata, normalized_rcdata=internal_normalize(rcdata)) # :notnew:
-      init_raw_string
-      @rcdata = rcdata && Hpricot.frozen_string(rcdata)
-      @normalized_rcdata = @rcdata == normalized_rcdata ? @rcdata : normalized_rcdata
-    end
-    attr_reader :rcdata, :normalized_rcdata
-    def internal_normalize(rcdata)
-      # - character references are decoded as much as possible.
-      # - undecodable character references are converted to decimal numeric character refereces.
-      result = rcdata.gsub(/&(?:#([0-9]+)|#x([0-9a-fA-F]+)|([A-Za-z][A-Za-z0-9]*));/o) {|s|
-        u = nil
-        if $1
-          u = $1.to_i
-        elsif $2
-          u = $2.hex
-        elsif $3
-          u = NamedCharacters[$3]
-        end
-        if !u || u < 0 || 0x7fffffff < u
-          '?'
-        elsif u == 38 # '&' character.
-          '&#38;'
-        elsif u <= 0x7f
-          [u].pack("C")
-        else
-          begin
-            Iconv.conv(Encoder.internal_charset, 'UTF-8', [u].pack("U"))
-          rescue Iconv::Failure
-            "&##{u};"
-          end
-        end
-      }
-      Hpricot.frozen_string(result)
-    end
-    private :internal_normalize
-    # Hpricot::Text#to_s converts the text to a string.
-    # - character references are decoded as much as possible.
-    # - undecodable character reference are converted to `?' character.
-    def to_s
-      @normalized_rcdata.gsub(/&(?:#([0-9]+));/o) {|s|
-        u = $1.to_i
-        if 0 <= u && u <= 0x7f
-          [u].pack("C")
-        else
-          '?'
-        end
-      }
-    end
-    def empty?
-      @normalized_rcdata.empty?
-    end
-    def strip
-      rcdata = @normalized_rcdata.dup
-      rcdata.sub!(/\A(?:\s|&nbsp;)+/, '')
-      rcdata.sub!(/(?:\s|&nbsp;)+\z/, '')
-      if rcdata == @normalized_rcdata
-        self
-      else
-        rcdata.freeze
-        Text.new_internal(rcdata, rcdata)
-      end
-    end
-    # Hpricot::Text.concat returns a text which is concatenation of arguments.
-    #
-    # An argument should be one of follows.
-    # - String
-    # - Hpricot::Text
-    # - Hpricot::Location which points Hpricot::Text
-    def Text.concat(*args)
-      rcdata = ''
-      args.each {|arg|
-        arg = arg.to_node if Hpricot::Location === arg
-        if Text === arg
-          rcdata << arg.rcdata
-        else
-          rcdata << arg.gsub(/&/, '&amp;')
-        end
-      }
-      new_internal rcdata
-    end
-  end
-end