nokogiri 1.6.8.rc3-java → 1.6.8.1-java

data/ext/java/nokogiri/internals/NokogiriHelpers.java

@@ -119,12 +119,12 @@ public class NokogiriHelpers {
         if (node.getNodeType() == Node.ATTRIBUTE_NODE && isNamespace(node.getNodeName())) {
             XmlDocument xmlDocument = (XmlDocument)node.getOwnerDocument().getUserData(CACHED_NODE);
             if (!(xmlDocument instanceof HtmlDocument)) {
-            String prefix = getLocalNameForNamespace(((Attr)node).getName());
-            prefix = prefix != null ? prefix : "";
-            String href = ((Attr)node).getValue();
-            XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefix, href);
-            if (xmlNamespace != null) return xmlNamespace;
-            else return XmlNamespace.createFromAttr(ruby, (Attr)node);
+                String prefix = getLocalNameForNamespace(((Attr)node).getName());
+                prefix = prefix != null ? prefix : "";
+                String href = ((Attr)node).getValue();
+                XmlNamespace xmlNamespace = xmlDocument.getNamespaceCache().get(prefix, href);
+                if (xmlNamespace != null) return xmlNamespace;
+                else return XmlNamespace.createFromAttr(ruby, (Attr)node);
             }
         }
         XmlNode xmlNode = getCachedNode(node);
@@ -134,7 +134,7 @@ public class NokogiriHelpers {
         }
         return xmlNode;
     }
-    
+
     /**
      * Construct a new XmlNode wrapping <code>node</code>.  The proper
      * subclass of XmlNode is chosen based on the type of

data/ext/java/nokogiri/internals/SaveContextVisitor.java

@@ -741,6 +741,10 @@ public class SaveContextVisitor {
       return htmlDoc && text.getParentNode().getNodeName().equals("script");
     }
 
+    private boolean isHtmlStyle(Text text) {
+      return htmlDoc && text.getParentNode().getNodeName().equals("style");
+    }
+
     private static char lineSeparator = '\n'; // System.getProperty("line.separator"); ?
     public boolean enter(Text text) {
         String textContent = text.getNodeValue();
@@ -752,7 +756,7 @@ public class SaveContextVisitor {
             }
         }
 
-        if (NokogiriHelpers.shouldEncode(text) && !isHtmlScript(text)) {
+        if (NokogiriHelpers.shouldEncode(text) && !isHtmlScript(text) && !isHtmlStyle(text)) {
             textContent = encodeJavaString(textContent);
         }
 

data/ext/nokogiri/extconf.rb

@@ -8,6 +8,42 @@ ROOT = File.expand_path(File.join(File.dirname(__FILE__), '..', '..'))
 #
 # functions
 #
+def windows?
+  RbConfig::CONFIG['target_os'] =~ /mingw32|mswin/
+end
+
+def solaris?
+  RbConfig::CONFIG['target_os'] =~ /solaris/
+end
+
+def darwin?
+  RbConfig::CONFIG['target_os'] =~ /darwin/
+end
+
+def nix?
+  ! (windows? || solaris? || darwin?)
+end
+
+def sh_export_path path
+  # because libxslt 1.1.29 configure.in uses AC_PATH_TOOL which treats ":"
+  # as a $PATH separator, we need to convert windows paths from
+  #
+  #   C:/path/to/foo
+  #
+  # to
+  #
+  #   /C/path/to/foo
+  #
+  # which is sh-compatible, in order to find things properly during
+  # configuration
+  if windows?
+    match = Regexp.new("^([A-Z]):(/.*)").match(path)
+    if match && match.length == 3
+      return File.join("/", match[1], match[2])
+    end
+  end
+  path
+end
 
 def do_help
   print <<HELP
@@ -70,27 +106,28 @@ def do_clean
   exit! 0
 end
 
-# The gem version constraint in the Rakefile is not respected at install time.
-# Keep this version in sync with the one in the Rakefile !
-require 'rubygems'
-gem 'pkg-config', '~> 1.1.7'
-require 'pkg-config'
-message "Using pkg-config version #{PKGConfig::VERSION}\n"
-
 def package_config pkg, options={}
   package = pkg_config(pkg)
   return package if package
 
-  return nil unless PKGConfig.have_package(pkg)
+  begin
+    require 'rubygems'
+    gem 'pkg-config', (gem_ver='~> 1.1.7')
+    require 'pkg-config' and message("Using pkg-config gem version #{PKGConfig::VERSION}\n")
+  rescue LoadError
+    message "pkg-config could not be used to find #{pkg}\nPlease install either `pkg-config` or the pkg-config gem per\n\n    gem install pkg-config -v #{gem_ver.inspect}\n\n"
+  else
+    return nil unless PKGConfig.have_package(pkg)
 
-  cflags  = PKGConfig.cflags(pkg)
-  ldflags = PKGConfig.libs_only_L(pkg)
-  libs    = PKGConfig.libs_only_l(pkg)
+    cflags  = PKGConfig.cflags(pkg)
+    ldflags = PKGConfig.libs_only_L(pkg)
+    libs    = PKGConfig.libs_only_l(pkg)
 
-  Logging::message "PKGConfig package configuration for %s\n", pkg
-  Logging::message "cflags: %s\nldflags: %s\nlibs: %s\n\n", cflags, ldflags, libs
+    Logging::message "PKGConfig package configuration for %s\n", pkg
+    Logging::message "cflags: %s\nldflags: %s\nlibs: %s\n\n", cflags, ldflags, libs
 
-  [cflags, ldflags, libs]
+    [cflags, ldflags, libs]
+  end
 end
 
 def nokogiri_try_compile
@@ -364,7 +401,10 @@ when arg_config('--clean')
 end
 
 RbConfig::MAKEFILE_CONFIG['CC'] = ENV['CC'] if ENV['CC']
+# use same c compiler for libxml and libxslt
+ENV['CC'] = RbConfig::MAKEFILE_CONFIG['CC']
 
+# TODO: deprecate MacRuby: https://github.com/sparklemotion/nokogiri/issues/1474
 if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'macruby'
   $LIBRUBYARG_STATIC.gsub!(/-static/, '')
 end
@@ -374,17 +414,20 @@ $LIBS << " #{ENV["LIBS"]}"
 # Read CFLAGS from ENV and make sure compiling works.
 add_cflags(ENV["CFLAGS"])
 
-case RbConfig::CONFIG['target_os']
-when 'mingw32', /mswin/
-  windows_p = true
+if windows?
   $CFLAGS << " -DXP_WIN -DXP_WIN32 -DUSE_INCLUDED_VASPRINTF"
-when /solaris/
+end
+
+if solaris?
   $CFLAGS << " -DUSE_INCLUDED_VASPRINTF"
-when /darwin/
-  darwin_p = true
+end
+
+if darwin?
   # Let Apple LLVM/clang 5.1 ignore unknown compiler flags
   add_cflags("-Wno-error=unused-command-line-argument-hard-error-in-future")
-else
+end
+
+if nix?
   $CFLAGS << " -g -DXP_UNIX"
 end
 
@@ -436,9 +479,12 @@ else
   dependencies = YAML.load_file(File.join(ROOT, "dependencies.yml"))
 
   cross_build_p = enable_config("cross-build")
-  if cross_build_p || windows_p
-    zlib_recipe = process_recipe("zlib", dependencies["zlib"], static_p, cross_build_p) do |recipe|
-      recipe.files = ["http://zlib.net/#{recipe.name}-#{recipe.version}.tar.gz"]
+  if cross_build_p || windows?
+    zlib_recipe = process_recipe("zlib", dependencies["zlib"]["version"], static_p, cross_build_p) do |recipe|
+      recipe.files = [{
+          url: "http://zlib.net/#{recipe.name}-#{recipe.version}.tar.gz",
+          md5: dependencies["zlib"]["md5"]
+        }]
       class << recipe
         attr_accessor :cross_build_p
 
@@ -472,8 +518,11 @@ else
       recipe.cross_build_p = cross_build_p
     end
 
-    libiconv_recipe = process_recipe("libiconv", dependencies["libiconv"], static_p, cross_build_p) do |recipe|
-      recipe.files = ["http://ftp.gnu.org/pub/gnu/libiconv/#{recipe.name}-#{recipe.version}.tar.gz"]
+    libiconv_recipe = process_recipe("libiconv", dependencies["libiconv"]["version"], static_p, cross_build_p) do |recipe|
+      recipe.files = [{
+          url: "http://ftp.gnu.org/pub/gnu/libiconv/#{recipe.name}-#{recipe.version}.tar.gz",
+          md5: dependencies["libiconv"]["md5"]
+        }]
       recipe.configure_options += [
         "CPPFLAGS=-Wall",
         "CFLAGS=-O2 -g",
@@ -482,7 +531,7 @@ else
       ]
     end
   else
-    if darwin_p && !have_header('iconv.h')
+    if darwin? && !have_header('iconv.h')
       abort <<'EOM'.chomp
 -----
 The file "iconv.h" is missing in your build environment,
@@ -498,17 +547,21 @@ EOM
     end
   end
 
-  unless windows_p
+  unless windows?
     preserving_globals {
       have_library('z', 'gzdopen', 'zlib.h')
     } or abort 'zlib is missing; necessary for building libxml2'
   end
 
-  libxml2_recipe = process_recipe("libxml2", dependencies["libxml2"], static_p, cross_build_p) do |recipe|
-    recipe.files = ["http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz"]
+  libxml2_recipe = process_recipe("libxml2", dependencies["libxml2"]["version"], static_p, cross_build_p) do |recipe|
+    recipe.files = [{
+        url: "http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz",
+        md5: dependencies["libxml2"]["md5"]
+      }]
     recipe.configure_options += [
       "--without-python",
       "--without-readline",
+      *(zlib_recipe ? ["--with-zlib=#{zlib_recipe.path}", "CFLAGS=-I#{zlib_recipe.path}/include"] : []),
       *(libiconv_recipe ? "--with-iconv=#{libiconv_recipe.path}" : iconv_configure_flags),
       "--with-c14n",
       "--with-debug",
@@ -516,13 +569,16 @@ EOM
     ]
   end
 
-  libxslt_recipe = process_recipe("libxslt", dependencies["libxslt"], static_p, cross_build_p) do |recipe|
-    recipe.files = ["http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz"]
+  libxslt_recipe = process_recipe("libxslt", dependencies["libxslt"]["version"], static_p, cross_build_p) do |recipe|
+    recipe.files = [{
+        url: "http://xmlsoft.org/sources/#{recipe.name}-#{recipe.version}.tar.gz",
+        md5: dependencies["libxslt"]["md5"]
+      }]
     recipe.configure_options += [
       "--without-python",
       "--without-crypto",
       "--with-debug",
-      "--with-libxml-prefix=#{libxml2_recipe.path}"
+      "--with-libxml-prefix=#{sh_export_path(libxml2_recipe.path)}"
     ]
   end
 

data/ext/nokogiri/xml_node.c

@@ -35,13 +35,13 @@ static void relink_namespace(xmlNodePtr reparented)
   xmlNsPtr ns;
 
   if (reparented->type != XML_ATTRIBUTE_NODE &&
-      reparented->type != XML_ELEMENT_NODE) return;
+      reparented->type != XML_ELEMENT_NODE) { return; }
 
   if (reparented->ns == NULL || reparented->ns->prefix == NULL) {
     name = xmlSplitQName2(reparented->name, &prefix);
 
     if(reparented->type == XML_ATTRIBUTE_NODE) {
-      if (prefix == NULL || strcmp((char*)prefix, XMLNS_PREFIX) == 0) return;
+      if (prefix == NULL || strcmp((char*)prefix, XMLNS_PREFIX) == 0) { return; }
     }
 
     ns = xmlSearchNs(reparented->doc, reparented, prefix);
@@ -57,18 +57,19 @@ static void relink_namespace(xmlNodePtr reparented)
   }
 
   /* Avoid segv when relinking against unlinked nodes. */
-  if (reparented->type != XML_ELEMENT_NODE || !reparented->parent) return;
+  if (reparented->type != XML_ELEMENT_NODE || !reparented->parent) { return; }
 
   /* Make sure that our reparented node has the correct namespaces */
-  if(!reparented->ns && reparented->doc != (xmlDocPtr)reparented->parent)
+  if (!reparented->ns && reparented->doc != (xmlDocPtr)reparented->parent) {
     xmlSetNs(reparented, reparented->parent->ns);
+  }
 
   /* Search our parents for an existing definition */
-  if(reparented->nsDef) {
+  if (reparented->nsDef) {
     xmlNsPtr curr = reparented->nsDef;
     xmlNsPtr prev = NULL;
 
-    while(curr) {
+    while (curr) {
       xmlNsPtr ns = xmlSearchNsByHref(
           reparented->doc,
           reparented->parent,
@@ -76,7 +77,7 @@ static void relink_namespace(xmlNodePtr reparented)
       );
       /* If we find the namespace is already declared, remove it from this
        * definition list. */
-      if(ns && ns != curr) {
+      if (ns && ns != curr && xmlStrEqual(ns->prefix, curr->prefix)) {
         if (prev) {
           prev->next = curr->next;
         } else {
@@ -92,12 +93,12 @@ static void relink_namespace(xmlNodePtr reparented)
 
   /* Only walk all children if there actually is a namespace we need to */
   /* reparent. */
-  if(NULL == reparented->ns) return;
+  if (NULL == reparented->ns) { return; }
 
   /* When a node gets reparented, walk it's children to make sure that */
   /* their namespaces are reparented as well. */
   child = reparented->children;
-  while(NULL != child) {
+  while (NULL != child) {
     relink_namespace(child);
     child = child->next;
   }
@@ -140,6 +141,7 @@ static VALUE reparent_node_with(VALUE pivot_obj, VALUE reparentee_obj, pivot_rep
 {
   VALUE reparented_obj ;
   xmlNodePtr reparentee, pivot, reparented, next_text, new_next_text, parent ;
+  int original_ns_prefix_is_default = 0 ;
 
   if(!rb_obj_is_kind_of(reparentee_obj, cNokogiriXmlNode))
     rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
@@ -250,10 +252,21 @@ ok:
        */
       reparentee->_private = NULL ;
     }
+
+    if (reparentee->ns != NULL && reparentee->ns->prefix == NULL) {
+      original_ns_prefix_is_default = 1;
+    }
+
     nokogiri_root_node(reparentee);
+
     if (!(reparentee = xmlDocCopyNode(reparentee, pivot->doc, 1))) {
       rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
     }
+
+    if (original_ns_prefix_is_default && reparentee->ns != NULL && reparentee->ns->prefix != NULL) {
+      /* issue #391, where new node's prefix may become the string "default" */
+      reparentee->ns->prefix = NULL;
+    }
   }
 
   if (prf != xmlAddPrevSibling && prf != xmlAddNextSibling

data/ext/nokogiri/xml_reader.c

@@ -219,19 +219,6 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   name = StringValue(name) ;
 
   value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValueCStr(name));
-  if(value == NULL) {
-    /* this section is an attempt to workaround older versions of libxml that
-       don't handle namespaces properly in all attribute-and-friends functions */
-    xmlChar *prefix = NULL ;
-    xmlChar *localname = xmlSplitQName2((xmlChar*)StringValueCStr(name), &prefix);
-    if (localname != NULL) {
-      value = xmlTextReaderLookupNamespace(reader, localname);
-      xmlFree(localname) ;
-    } else {
-      value = xmlTextReaderLookupNamespace(reader, prefix);
-    }
-    xmlFree(prefix);
-  }
   if(value == NULL) return Qnil;
 
   rb_value = NOKOGIRI_STR_NEW2(value);

data/ext/nokogiri/xml_sax_parser.c

@@ -18,16 +18,16 @@ static void start_document(void * ctx)
 
   if(NULL != ctxt && ctxt->html != 1) {
     if(ctxt->standalone != -1) {  /* -1 means there was no declaration */
-      VALUE encoding = ctxt->encoding ?
-        NOKOGIRI_STR_NEW2(ctxt->encoding) :
-        Qnil;
+      VALUE encoding = Qnil ;
+      if (ctxt->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->encoding) ;
+      } else if (ctxt->input && ctxt->input->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->input->encoding) ;
+      }
 
-      VALUE version = ctxt->version ?
-        NOKOGIRI_STR_NEW2(ctxt->version) :
-        Qnil;
+      VALUE version = ctxt->version ? NOKOGIRI_STR_NEW2(ctxt->version) : Qnil;
 
       VALUE standalone = Qnil;
-
       switch(ctxt->standalone)
       {
         case 0:

data/lib/nokogiri.rb

@@ -114,15 +114,19 @@ module Nokogiri
     def Slop(*args, &block)
       Nokogiri(*args, &block).slop!
     end
+
+    def install_default_aliases
+      # Make sure to support some popular encoding aliases not known by
+      # all iconv implementations.
+      {
+        'Windows-31J' => 'CP932',	# Windows-31J is the IANA registered name of CP932.
+      }.each { |alias_name, name|
+        EncodingHandler.alias(name, alias_name) if EncodingHandler[alias_name].nil?
+      }
+    end
   end
 
-  # Make sure to support some popular encoding aliases not known by
-  # all iconv implementations.
-  {
-    'Windows-31J' => 'CP932',	# Windows-31J is the IANA registered name of CP932.
-  }.each { |alias_name, name|
-    EncodingHandler.alias(name, alias_name) if EncodingHandler[alias_name].nil?
-  }
+  Nokogiri.install_default_aliases
 end
 
 ###

data/lib/nokogiri/html/document.rb

@@ -173,7 +173,7 @@ module Nokogiri
 
           if string_or_io.respond_to?(:read)
             url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
-            if !encoding
+            unless encoding
               # Libxml2's parser has poor support for encoding
               # detection.  First, it does not recognize the HTML5
               # style meta charset declaration.  Secondly, even if it
@@ -196,7 +196,9 @@ module Nokogiri
           end
 
           # read_memory pukes on empty docs
-          return new if string_or_io.nil? or string_or_io.empty?
+          if string_or_io.nil? or string_or_io.empty?
+            return encoding ? new.tap { |i| i.encoding = encoding } : new
+          end
 
           encoding ||= EncodingReader.detect_encoding(string_or_io)
 

data/lib/nokogiri/version.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.6.8.rc3'
+  VERSION = '1.6.8.1'
 
   class VersionInfo # :nodoc:
     def jruby?