nokogiri 1.5.0.beta.4-java → 1.5.0-java

data/ext/java/nokogiri/internals/XmlDomParser.java

@@ -71,7 +71,6 @@ public class XmlDomParser extends DOMParser {
         if (doc == null)
             throw new RuntimeException("null document");
 
-        doc.setUserData(XmlDocument.DTD_RAW_DOCUMENT, dtd.getDocument(),
-                        null);
+        doc.setUserData(XmlDocument.DTD_RAW_DOCUMENT, dtd.getDocument(), null);
     }
 }

data/ext/java/nokogiri/internals/XmlDomParserContext.java

@@ -42,6 +42,7 @@ import java.util.List;
 
 import nokogiri.NokogiriService;
 import nokogiri.XmlDocument;
+import nokogiri.XmlDtd;
 import nokogiri.XmlSyntaxError;
 
 import org.apache.xerces.parsers.DOMParser;
@@ -189,6 +190,11 @@ public class XmlDomParserContext extends ParserContext {
         XmlDocument xmlDocument = (XmlDocument) NokogiriService.XML_DOCUMENT_ALLOCATOR.allocate(context.getRuntime(), klazz);
         xmlDocument.setNode(context, doc);
         xmlDocument.setEncoding(ruby_encoding);
+
+        if (options.dtdLoad) {
+            XmlDtd xmlDtd = (XmlDtd) XmlDtd.newFromExternalSubset(context.getRuntime(), doc);
+            doc.setUserData(XmlDocument.DTD_EXTERNAL_SUBSET, xmlDtd, null);
+        }
         return xmlDocument;
     }
 

data/ext/nokogiri/nokogiri.c

@@ -23,6 +23,29 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
 }
 #endif
 
+#ifdef USING_SYSTEM_ALLOCATOR_LIBRARY /* Ruby Enterprise Edition with tcmalloc */
+void vasprintf_free (void *p)
+{
+  system_free(p);
+}
+#else
+void vasprintf_free (void *p)
+{
+  free(p);
+}
+#endif
+
+#ifndef __MACRUBY__
+/* Allocate strdupped strings with the same memory allocator Ruby uses. */
+static char *ruby_strdup(const char *s)
+{
+  size_t len = strlen(s);
+  char *result = ruby_xmalloc((ssize_t) (len + 1));
+  memcpy(result, s, len + 1);
+  return result;
+}
+#endif
+
 void Init_nokogiri()
 {
 #ifndef __MACRUBY__
@@ -30,7 +53,7 @@ void Init_nokogiri()
       (xmlFreeFunc)ruby_xfree,
       (xmlMallocFunc)ruby_xmalloc,
       (xmlReallocFunc)ruby_xrealloc,
-      strdup
+      ruby_strdup
   );
 #endif
 

data/ext/nokogiri/xml_io.c

@@ -2,22 +2,47 @@
 
 static ID id_read, id_write;
 
+VALUE read_check(VALUE *args) {
+  return rb_funcall(args[0], id_read, 1, args[1]);
+}
+
+VALUE read_failed(void) {
+	return Qnil;
+}
+
 int io_read_callback(void * ctx, char * buffer, int len) {
-  VALUE io = (VALUE)ctx;
-  VALUE string = rb_funcall(io, id_read, 1, INT2NUM(len));
+  VALUE string, args[2];
+  size_t str_len, safe_len;
+
+  args[0] = (VALUE)ctx;
+  args[1] = INT2NUM(len);
+
+  string = rb_rescue(read_check, (VALUE)args, read_failed, 0);
 
   if(NIL_P(string)) return 0;
 
-  memcpy(buffer, StringValuePtr(string), (size_t)RSTRING_LEN(string));
+  str_len = (size_t)RSTRING_LEN(string);
+  safe_len = str_len > (size_t)len ? (size_t)len : str_len;
+  memcpy(buffer, StringValuePtr(string), safe_len);
 
-  return (int)RSTRING_LEN(string);
+  return safe_len;
+}
+
+VALUE write_check(VALUE *args) {
+  return rb_funcall(args[0], id_write, 1, args[1]);
+}
+
+VALUE write_failed(void) {
+	return Qnil;
 }
 
 int io_write_callback(void * ctx, char * buffer, int len) {
-  VALUE io = (VALUE)ctx;
-  VALUE string = rb_str_new(buffer, (long)len);
+  VALUE args[2];
+
+  args[0] = (VALUE)ctx;
+  args[1] = rb_str_new(buffer, (long)len);
 
-  rb_funcall(io, id_write, 1, string);
+  rb_rescue(write_check, (VALUE)args, write_failed, 0);
   return len;
 }
 

data/ext/nokogiri/xml_node.c

@@ -14,10 +14,7 @@ static void debug_node_dealloc(xmlNodePtr x)
 
 static void mark(xmlNodePtr node)
 {
-  /* it's OK if the document isn't fully realized (as in XML::Reader). */
-  /* see http://github.com/tenderlove/nokogiri/issues/closed/#issue/95 */
-  if (DOC_RUBY_OBJECT_TEST(node->doc) && DOC_RUBY_OBJECT(node->doc))
-    rb_gc_mark(DOC_RUBY_OBJECT(node->doc));
+  rb_gc_mark(DOC_RUBY_OBJECT(node->doc));
 }
 
 /* :nodoc: */
@@ -1235,6 +1232,8 @@ VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
   VALUE document = Qnil ;
   VALUE node_cache = Qnil ;
   VALUE rb_node = Qnil ;
+  int node_has_a_document = 0 ;
+  void (*mark_method)(xmlNodePtr) = NULL ;
 
   assert(node);
 
@@ -1243,11 +1242,9 @@ VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
 
   if(NULL != node->_private) return (VALUE)node->_private;
 
-  if(RTEST(klass))
-    rb_node = Data_Wrap_Struct(klass, mark, debug_node_dealloc, node) ;
-
-  else switch(node->type)
-  {
+  if(!RTEST(klass)) {
+    switch(node->type)
+    {
     case XML_ELEMENT_NODE:
       klass = cNokogiriXmlElement;
       break;
@@ -1286,15 +1283,19 @@ VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
       break;
     default:
       klass = cNokogiriXmlNode;
+    }
   }
 
-  rb_node = Data_Wrap_Struct(klass, mark, debug_node_dealloc, node) ;
+  /* It's OK if the node doesn't have a fully-realized document (as in XML::Reader). */
+  /* see https://github.com/tenderlove/nokogiri/issues/95 */
+  /* and https://github.com/tenderlove/nokogiri/issues/439 */
+  node_has_a_document = (DOC_RUBY_OBJECT_TEST(node->doc) && DOC_RUBY_OBJECT(node->doc)) ? 1 : 0 ;
+  mark_method = node_has_a_document ? mark : NULL ;
 
+  rb_node = Data_Wrap_Struct(klass, mark_method, debug_node_dealloc, node) ;
   node->_private = (void *)rb_node;
 
-  if (DOC_RUBY_OBJECT_TEST(node->doc) && DOC_RUBY_OBJECT(node->doc)) {
-    /* it's OK if the document isn't fully realized (as in XML::Reader). */
-    /* see http://github.com/tenderlove/nokogiri/issues/closed/#issue/95 */
+  if (node_has_a_document) {
     document = DOC_RUBY_OBJECT(node->doc);
     node_cache = DOC_NODE_CACHE(node->doc);
     rb_ary_push(node_cache, rb_node);

data/ext/nokogiri/xml_sax_parser.c

@@ -1,6 +1,7 @@
 #include <xml_sax_parser.h>
 
 int vasprintf (char **strp, const char *fmt, va_list ap);
+void vasprintf_free (void *p);
 
 static ID id_start_document, id_end_document, id_start_element, id_end_element;
 static ID id_start_element_namespace, id_end_element_namespace;
@@ -198,14 +199,16 @@ static void warning_func(void * ctx, const char *msg, ...)
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
   VALUE doc = rb_iv_get(self, "@document");
   char * message;
+  VALUE ruby_message;
 
   va_list args;
   va_start(args, msg);
   vasprintf(&message, msg, args);
   va_end(args);
 
-  rb_funcall(doc, id_warning, 1, NOKOGIRI_STR_NEW2(message));
-  free(message);
+  ruby_message = NOKOGIRI_STR_NEW2(message);
+  vasprintf_free(message);
+  rb_funcall(doc, id_warning, 1, ruby_message);
 }
 
 static void error_func(void * ctx, const char *msg, ...)
@@ -213,14 +216,16 @@ static void error_func(void * ctx, const char *msg, ...)
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
   VALUE doc = rb_iv_get(self, "@document");
   char * message;
+  VALUE ruby_message;
 
   va_list args;
   va_start(args, msg);
   vasprintf(&message, msg, args);
   va_end(args);
 
-  rb_funcall(doc, id_error, 1, NOKOGIRI_STR_NEW2(message));
-  free(message);
+  ruby_message = NOKOGIRI_STR_NEW2(message);
+  vasprintf_free(message);
+  rb_funcall(doc, id_error, 1, ruby_message);
 }
 
 static void cdata_block(void * ctx, const xmlChar * value, int len)

data/ext/nokogiri/xslt_stylesheet.c

@@ -8,6 +8,7 @@
 VALUE xslt;
 
 int vasprintf (char **strp, const char *fmt, va_list ap);
+void vasprintf_free (void *p);
 
 static void dealloc(xsltStylesheetPtr doc)
 {
@@ -20,13 +21,16 @@ NORETURN(static void xslt_generic_error_handler(void * ctx, const char *msg, ...
 static void xslt_generic_error_handler(void * ctx, const char *msg, ...)
 {
   char * message;
+  VALUE exception;
 
   va_list args;
   va_start(args, msg);
   vasprintf(&message, msg, args);
   va_end(args);
 
-  rb_exc_raise(rb_exc_new2(rb_eRuntimeError, message));
+  exception = rb_exc_new2(rb_eRuntimeError, message);
+  vasprintf_free(message);
+  rb_exc_raise(exception);
 }
 
 /*
@@ -100,6 +104,8 @@ static VALUE transform(int argc, VALUE* argv, VALUE self)
 
     rb_scan_args(argc, argv, "11", &xmldoc, &paramobj);
     if (NIL_P(paramobj)) { paramobj = rb_ary_new2(0L) ; }
+    if (!rb_obj_is_kind_of(xmldoc, cNokogiriXmlDocument))
+      rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::Document");
 
     /* handle hashes as arguments. */
     if(T_HASH == TYPE(paramobj)) {

data/lib/nokogiri.rb

@@ -22,28 +22,9 @@ if defined?(RUBY_ENGINE) && RUBY_ENGINE == "jruby"
     require 'nekodtd.jar'
     require 'xercesImpl.jar'
   end
-  require 'nokogiri/nokogiri'
-else
-  require 'nokogiri/nokogiri'
 end
 
-warn(<<-eowarn) if '1.8.6' == RUBY_VERSION && $VERBOSE
-!!! DEPRECATION WARNING !!!
-Hey Champ!  I see you're using Ruby 1.8.6!  While I applaud you for sticking to
-your guns and using The One True Ruby, I have to let you know that we're going
-to stop supporting 1.8.6.  I know, it's sad.  But, we just don't have time to
-support every version of Ruby out there.  Whether we like it or not, time moves
-forward and so does our software.
-
-On August 1, 2010, we will no longer support Ruby 1.8.6.  If nokogiri happens to
-work on 1.8.6 after that date, then great!  We will hownever, no longer test,
-use, or endorse 1.8.6 as a supported platform.
-
-Thanks,
-
-  Team Nokogiri
-eowarn
-
+require 'nokogiri/nokogiri'
 require 'nokogiri/version'
 require 'nokogiri/syntax_error'
 require 'nokogiri/xml'
@@ -86,13 +67,13 @@ module Nokogiri
       doc =
         if string.respond_to?(:read) ||
           string =~ /^\s*<[^Hh>]*html/i # Probably html
-          Nokogiri::HTML(
+          Nokogiri.HTML(
             string,
             url,
             encoding, options || XML::ParseOptions::DEFAULT_HTML
           )
         else
-          Nokogiri::XML(string, url, encoding,
+          Nokogiri.XML(string, url, encoding,
                         options || XML::ParseOptions::DEFAULT_XML)
         end
       yield doc if block_given?

data/lib/nokogiri/css.rb

@@ -1,6 +1,10 @@
 require 'nokogiri/css/node'
 require 'nokogiri/css/xpath_visitor'
+x = $-w
+$-w = false
 require 'nokogiri/css/parser'
+$-w = x
+
 require 'nokogiri/css/tokenizer'
 require 'nokogiri/css/syntax_error'
 

data/lib/nokogiri/html/document.rb

@@ -58,10 +58,7 @@ module Nokogiri
       #   end
       #
       def serialize options = {}
-        options[:save_with] ||= XML::Node::SaveOptions::FORMAT |
-            XML::Node::SaveOptions::AS_HTML |
-            XML::Node::SaveOptions::NO_DECLARATION |
-            XML::Node::SaveOptions::NO_EMPTY_TAGS
+        options[:save_with] ||= XML::Node::SaveOptions::DEFAULT_HTML
         super
       end
 
@@ -73,7 +70,7 @@ module Nokogiri
 
       class << self
         ###
-        # Parse HTML.  +thing+ may be a String, or any object that
+        # Parse HTML.  +string_or_io+ may be a String, or any object that
         # responds to _read_ and _close_ such as an IO, or StringIO.
         # +url+ is resource where this document is located.  +encoding+ is the
         # encoding that should be used when processing the document. +options+
@@ -95,7 +92,7 @@ module Nokogiri
           if string_or_io.respond_to?(:read)
             url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
             if !encoding
-              # Perform further encoding detection that libxml2 does
+              # Perform advanced encoding detection that libxml2 does
               # not do.
               string_or_io = EncodingReader.new(string_or_io)
               begin
@@ -150,6 +147,8 @@ module Nokogiri
               not_found
             when 'meta'
               attr = Hash[attrs]
+              charset = attr['charset'] and
+                found charset
               http_equiv = attr['http-equiv'] and
                 http_equiv.match(/\AContent-Type\z/i) and
                 content = attr['content'] and
@@ -174,7 +173,7 @@ module Nokogiri
             parser.parse(chunk)
           }
           handler.encoding
-        rescue => e
+        rescue
           nil
         end
 
@@ -189,16 +188,13 @@ module Nokogiri
           if !@firstchunk
             @firstchunk = @io.read(len) or return nil
 
-            # This implementation expects and assumes that the first
-            # call from htmlReadIO() is made with a length long enough
-            # (~1KB) to achieve further encoding detection that
-            # libxml2 does not do.
+            # This implementation expects that the first call from
+            # htmlReadIO() is made with a length long enough (~1KB) to
+            # achieve advanced encoding detection.
             if encoding = EncodingReader.detect_encoding(@firstchunk)
+              # The first chunk is stored for the next read in retry.
               raise EncodingFoundException, encoding
             end
-
-            # This chunk is stored for the next read in retry.
-            return @firstchunk
           end
 
           ret = @firstchunk.slice!(0, len)

data/lib/nokogiri/version.rb

@@ -1,35 +1,88 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.5.0.beta.4'
+  VERSION = '1.5.0'
 
-  # More complete version information about libxml
-  VERSION_INFO = {}
-  VERSION_INFO['warnings']              = []
-  VERSION_INFO['nokogiri']              = Nokogiri::VERSION
-  VERSION_INFO['ruby']                  = {}
-  VERSION_INFO['ruby']['version']       = ::RUBY_VERSION
-  VERSION_INFO['ruby']['platform']      = ::RUBY_PLATFORM
-  VERSION_INFO['ruby']['engine']        = defined?(RUBY_ENGINE) ? RUBY_ENGINE : 'mri'
-
-  VERSION_INFO['ruby']['jruby']         = ::JRUBY_VERSION if RUBY_PLATFORM == "java"
-  if defined?(LIBXML_VERSION)
-    VERSION_INFO['libxml']              = {}
-    VERSION_INFO['libxml']['binding']   = 'extension'
-    VERSION_INFO['libxml']['compiled']  = LIBXML_VERSION
-    VERSION_INFO['libxml']['loaded']    = LIBXML_PARSER_VERSION.scan(/^(.*)(..)(..)$/).first.collect{|j|j.to_i}.join(".")
-
-    if VERSION_INFO['libxml']['compiled'] != VERSION_INFO['libxml']['loaded']
-      warning = "Nokogiri was built against LibXML version #{VERSION_INFO['libxml']['compiled']}, but has dynamically loaded #{VERSION_INFO['libxml']['loaded']}"
-      VERSION_INFO['warnings'] << warning
+  class VersionInfo # :nodoc:
+    def jruby?
+      ::JRUBY_VERSION if RUBY_PLATFORM == "java"
+    end
+
+    def engine
+      defined?(RUBY_ENGINE) ? RUBY_ENGINE : 'mri'
+    end
+
+    def loaded_parser_version
+      LIBXML_PARSER_VERSION.scan(/^(.*)(..)(..)$/).first.collect{ |j|
+        j.to_i
+      }.join(".")
+    end
+
+    def compiled_parser_version
+      LIBXML_VERSION
+    end
+
+    def libxml2?
+      defined?(LIBXML_VERSION)
+    end
+
+    def warnings
+      return [] unless libxml2?
+
+      if compiled_parser_version != loaded_parser_version
+        ["Nokogiri was built against LibXML version #{compiled_parser_version}, but has dynamically loaded #{loaded_parser_version}"]
+      else
+        []
+      end
+    end
+
+    def to_hash
+      hash_info = {}
+      hash_info['warnings']              = []
+      hash_info['nokogiri']              = Nokogiri::VERSION
+      hash_info['ruby']                  = {}
+      hash_info['ruby']['version']       = ::RUBY_VERSION
+      hash_info['ruby']['platform']      = ::RUBY_PLATFORM
+      hash_info['ruby']['description']   = ::RUBY_DESCRIPTION
+      hash_info['ruby']['engine']        = engine
+      hash_info['ruby']['jruby']         = jruby? if jruby?
+
+      if libxml2?
+        hash_info['libxml']              = {}
+        hash_info['libxml']['binding']   = 'extension'
+        hash_info['libxml']['compiled']  = compiled_parser_version
+        hash_info['libxml']['loaded']    = loaded_parser_version
+        hash_info['warnings']            = warnings
+      end
+
+      hash_info
+    end
+
+    def to_markdown
+      begin
+        require 'psych'
+      rescue LoadError
+      end
+      require 'yaml'
+      "# Nokogiri (#{Nokogiri::VERSION})\n" +
+      YAML.dump(to_hash).each_line.map { |line| "    #{line}" }.join
+    end
+
+    # FIXME: maybe switch to singleton?
+    @@instance = new
+    @@instance.warnings.each do |warning|
       warn "WARNING: #{warning}"
     end
+    def self.instance; @@instance; end
   end
 
+  # More complete version information about libxml
+  VERSION_INFO = VersionInfo.instance.to_hash
+
   def self.uses_libxml? # :nodoc:
-    !Nokogiri::VERSION_INFO['libxml'].nil?
+    VersionInfo.instance.libxml2?
   end
 
-  def self.jruby?
-    !Nokogiri::VERSION_INFO['ruby']['jruby'].nil?
+  def self.jruby? # :nodoc:
+    VersionInfo.instance.jruby?
   end
 end