sparklemotion · knu · Aug 8, 2013 · Jul 3, 2013
diff --git a/lib/nokogiri.rb b/lib/nokogiri.rb
@@ -67,7 +67,7 @@ class << self
     def parse string, url = nil, encoding = nil, options = nil
       doc =
         if string.respond_to?(:read) ||
-          string =~ /^\s*<[^Hh>]*html/i # Probably html
+          string =~ /^\s*<(?:!DOCTYPE\s+)?html[\s>]/i # Probably html
           Nokogiri.HTML(
             string,
             url,

diff --git a/test/files/atom.xml b/test/files/atom.xml
@@ -0,0 +1,344 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom" xmlns:media="http://search.yahoo.com/mrss/" xml:lang="en-US">
+  <id>tag:github.com,2008:/sparklemotion/nokogiri/commits/master</id>
+  <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commits/master"/>
+  <link type="application/atom+xml" rel="self" href="https://github.com/sparklemotion/nokogiri/commits/master.atom"/>
+  <title>Recent Commits to nokogiri:master</title>
+  <updated>2013-07-02T08:17:16-07:00</updated>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/233489ddfe4698b24dfcdc1daf3ca15bc880e4b1</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/233489ddfe4698b24dfcdc1daf3ca15bc880e4b1"/>
+    <title>
+        Updated changelog with merged pull requests #887 and #931
+    </title>
+    <updated>2013-07-02T08:17:16-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Updated changelog with merged pull requests #887 and #931&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/15eff8be4539a3461cfc1e3c8a257e7eed134e01</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/15eff8be4539a3461cfc1e3c8a257e7eed134e01"/>
+    <title>
+        Merge pull request #887 from Mange/double-not
+    </title>
+    <updated>2013-07-02T08:13:36-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Merge pull request #887 from Mange/double-not
+
+Add support for bare and multiple :not() functions in selectors&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/7a54b1fd1ba1dbcd50d3566bcc62d554eb0fd30e</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/7a54b1fd1ba1dbcd50d3566bcc62d554eb0fd30e"/>
+    <title>
+        Merge pull request #931 from sorah/dont_call_pkgconfig_when_using_bundled_libraries
+    </title>
+    <updated>2013-07-02T08:11:37-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/74f896b312b786ee75a18073941e2457?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>knu</name>
+      <uri>https://github.com/knu</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Merge pull request #931 from sorah/dont_call_pkgconfig_when_using_bundled_libraries
+
+extconf.rb: Don&#39;t call pkg_config when using bundled libraries&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/e5b93617ba810a34a4fc70dac76a83f9bebd2ea1</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/e5b93617ba810a34a4fc70dac76a83f9bebd2ea1"/>
+    <title>
+        extconf.rb: Don&#39;t call pkg_config when using bundled libraries
+    </title>
+    <updated>2013-07-02T06:30:00-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/c4f076f658dd3464f1d8785ad53a0d99?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>sorah</name>
+      <uri>https://github.com/sorah</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>extconf.rb: Don&#39;t call pkg_config when using bundled libraries&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/64cd3c8f506394b558bda9376844a51db971bd60</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/64cd3c8f506394b558bda9376844a51db971bd60"/>
+    <title>
+        Add support for bare and multiple :not() functions in selectors
+    </title>
+    <updated>2013-06-28T10:19:36-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/1b2792536d87aa21bc739c14980fa103?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>Mange</name>
+      <uri>https://github.com/Mange</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Add support for bare and multiple :not() functions in selectors
+
+This enables parsing of &quot;.flash:not(.error):not(.warning)&quot; and
+&quot;:not(p)&quot;.
+
+This entailed a huge change to the parser, but I think the parser should
+be a bit more robust now by not declaring :not a special case.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/37828ec784faf29d713225fa78585854912356cd</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/37828ec784faf29d713225fa78585854912356cd"/>
+    <title>
+        Report error and stop parsing instead of silently ignoring it.
+    </title>
+    <updated>2013-06-27T18:46:12-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/2736d9750eb13425e9bf70f112753c49?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>jvshahid</name>
+      <uri>https://github.com/jvshahid</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Report error and stop parsing instead of silently ignoring it.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/42da60e8ca58a5acc9e8c94ae5ad490589f146c6</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/42da60e8ca58a5acc9e8c94ae5ad490589f146c6"/>
+    <title>
+        Merge pull request #930 from ctborg/master
+    </title>
+    <updated>2013-06-27T04:23:33-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Merge pull request #930 from ctborg/master
+
+Adds support for Solaris, OpenSolaris, or illumos.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/7382a6e6a66d3f80da955ebbcd82250f99321900</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/7382a6e6a66d3f80da955ebbcd82250f99321900"/>
+    <title>
+        Updated CHANGELOG with latest merged pulls
+    </title>
+    <updated>2013-06-27T04:04:17-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Updated CHANGELOG with latest merged pulls&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/3382ebc90e17f2503f4acbbdf53a931d38f3322a</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/3382ebc90e17f2503f4acbbdf53a931d38f3322a"/>
+    <title>
+        Remove REE also.
+    </title>
+    <updated>2013-06-19T20:57:48-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/74f896b312b786ee75a18073941e2457?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>knu</name>
+      <uri>https://github.com/knu</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Remove REE also.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/0c83f4027e295435680cda89a44b5e039e0e6cbb</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/0c83f4027e295435680cda89a44b5e039e0e6cbb"/>
+    <title>
+        Added support for Solaris, OpenSolaris, or illumos
+    </title>
+    <updated>2013-06-19T07:12:50-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/aff9165c0ea741d027799d5694d0ceb8?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name></name>
+      <email>[email protected]</email>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Added support for Solaris, OpenSolaris, or illumos&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/fb000c51e3ba4e8730a535d38d617d34111c21ef</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/fb000c51e3ba4e8730a535d38d617d34111c21ef"/>
+    <title>
+        Ensure meta_encoding checks meta charset tag (closes #919)
+    </title>
+    <updated>2013-06-14T06:06:26-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Ensure meta_encoding checks meta charset tag (closes #919)&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/14b7dc21bbeddec1990726760e6f6cc11e71542e</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/14b7dc21bbeddec1990726760e6f6cc11e71542e"/>
+    <title>
+        Updated ROADMAP
+    </title>
+    <updated>2013-06-14T06:00:57-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Updated ROADMAP&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/d1b0afd02f8a2b783380bdfde3acfabf021a0c5a</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/d1b0afd02f8a2b783380bdfde3acfabf021a0c5a"/>
+    <title>
+        Merge pull request #858 from ykzts/feature/not-only-child
+    </title>
+    <updated>2013-06-14T05:53:28-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Merge pull request #858 from ykzts/feature/not-only-child
+
+Add feature that the &#39;:only-child&#39; pseudo class should work even though it&#39;s in &#39;:not&#39; pseudo class.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/d24eb8508c69f2cf336f1ff1b6887fcc3b317fbc</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/d24eb8508c69f2cf336f1ff1b6887fcc3b317fbc"/>
+    <title>
+        Merge pull request #886 from Mange/negative-nth
+    </title>
+    <updated>2013-06-14T05:50:13-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/9077af710bdadc972a7898e457bf6ec1?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>injekt</name>
+      <uri>https://github.com/injekt</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Merge pull request #886 from Mange/negative-nth
+
+Add support for an-b in nth selectors&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/11eaf3d3f1b84adf5225b6686119dbc6714b4009</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/11eaf3d3f1b84adf5225b6686119dbc6714b4009"/>
+    <title>
+        Note 1.8 deprecation in README
+    </title>
+    <updated>2013-06-11T06:48:53-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/bd6e7860b5a891bff077aeaeb5434e60?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>flavorjones</name>
+      <uri>https://github.com/flavorjones</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Note 1.8 deprecation in README&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/82a936a0ceba0508d29cc9f6da84c5a98ae7bede</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/82a936a0ceba0508d29cc9f6da84c5a98ae7bede"/>
+    <title>
+        Fix a typo and DRY with dir_config() calls.
+    </title>
+    <updated>2013-06-10T21:01:03-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/74f896b312b786ee75a18073941e2457?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>knu</name>
+      <uri>https://github.com/knu</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Fix a typo and DRY with dir_config() calls.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/c7ef9b339054f8502e7f526b1b8054e3412cf7ba</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/c7ef9b339054f8502e7f526b1b8054e3412cf7ba"/>
+    <title>
+        Iconv was for building libxml2 and not for nokogiri itself.
+    </title>
+    <updated>2013-06-10T17:55:20-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/74f896b312b786ee75a18073941e2457?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>knu</name>
+      <uri>https://github.com/knu</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Iconv was for building libxml2 and not for nokogiri itself.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/4c1c225b9409571156430aeb5bcf89c4f9dd4812</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/4c1c225b9409571156430aeb5bcf89c4f9dd4812"/>
+    <title>
+        Make sure the MRI gem is built correctly.
+    </title>
+    <updated>2013-06-10T07:40:39-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/bd6e7860b5a891bff077aeaeb5434e60?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>flavorjones</name>
+      <uri>https://github.com/flavorjones</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Make sure the MRI gem is built correctly.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/b6100afd7a0ce17097fb3ca58ae8fe5e63522f6d</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/b6100afd7a0ce17097fb3ca58ae8fe5e63522f6d"/>
+    <title>
+        Fiddling with the windows cross-compile gem logic.
+    </title>
+    <updated>2013-06-10T07:33:59-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/bd6e7860b5a891bff077aeaeb5434e60?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>flavorjones</name>
+      <uri>https://github.com/flavorjones</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>Fiddling with the windows cross-compile gem logic.&lt;/pre>
+    </content>
+  </entry>
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/b1483de363524c882b5e682a3d3a6e6ab5d7511b</id>
+    <link type="text/html" rel="alternate" href="https://github.com/sparklemotion/nokogiri/commit/b1483de363524c882b5e682a3d3a6e6ab5d7511b"/>
+    <title>
+        build_all now uses ruby 1.9.3 to build everything.
+    </title>
+    <updated>2013-06-10T07:33:56-07:00</updated>
+    <media:thumbnail height="30" width="30" url="https://secure.gravatar.com/avatar/bd6e7860b5a891bff077aeaeb5434e60?s=30&amp;d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png"/>
+    <author>
+      <name>flavorjones</name>
+      <uri>https://github.com/flavorjones</uri>
+    </author>
+    <content type="html">
+      &lt;pre style='white-space:pre-wrap;width:81ex'>build_all now uses ruby 1.9.3 to build everything.&lt;/pre>
+    </content>
+  </entry>
+</feed>
diff --git a/test/helper.rb b/test/helper.rb
@@ -30,6 +30,7 @@ class TestCase < MiniTest::Spec
     SNUGGLES_FILE       = File.join(ASSETS_DIR, 'snuggles.xml')
     XML_FILE            = File.join(ASSETS_DIR, 'staff.xml')
     XML_XINCLUDE_FILE   = File.join(ASSETS_DIR, 'xinclude.xml')
+    XML_ATOM_FILE       = File.join(ASSETS_DIR, 'atom.xml')
     XSLT_FILE           = File.join(ASSETS_DIR, 'staff.xslt')
 
     def teardown

diff --git a/test/test_nokogiri.rb b/test/test_nokogiri.rb
@@ -41,6 +41,12 @@ def test_xml?
     assert !doc.html?
   end
 
+  def test_atom_is_xml?
+    doc = Nokogiri.parse(File.read(XML_ATOM_FILE))
+    assert doc.xml?
+    assert !doc.html?
+  end
+
   def test_html?
     doc = Nokogiri.parse(File.read(HTML_FILE))
     assert !doc.xml?