nokogiri 1.8.2-java → 1.8.3-java

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

@@ -229,11 +229,8 @@ public class NokogiriHelpers {
     public static String getPrefix(String qName) {
         if (qName == null) return null;
 
-        int pos = qName.indexOf(':');
-        if (pos > 0)
-            return qName.substring(0, pos);
-        else
-            return null;
+        final int pos = qName.indexOf(':');
+        return pos > 0 ? qName.substring(0, pos) : null;
     }
 
     /**
@@ -243,11 +240,8 @@ public class NokogiriHelpers {
     public static String getLocalPart(String qName) {
         if (qName == null) return null;
 
-        int pos = qName.indexOf(':');
-        if (pos > 0)
-            return qName.substring(pos + 1);
-        else
-            return qName;
+        final int pos = qName.indexOf(':');
+        return pos > 0 ? qName.substring(pos + 1) : qName;
     }
 
     public static String getLocalNameForNamespace(String name) {
@@ -593,20 +587,23 @@ public class NokogiriHelpers {
         return "xml:base".equals(attrName) || "xlink:href".equals(attrName);
     }
 
-    public static boolean isWhitespaceText(ThreadContext context, IRubyObject obj) {
-        //if (obj == null || obj.isNil()) return false;
+    public static boolean isBlank(IRubyObject obj) {
         if ( !(obj instanceof XmlText) ) return false;
 
         CharSequence content = ((XmlNode) obj).getContentImpl();
-        return content == null || isWhitespaceText(content);
+        return content == null || isBlank(content);
     }
 
-    public static boolean isWhitespaceText(CharSequence str) {
+    public static boolean isBlank(CharSequence str) {
         int len = str.length(); int beg = 0;
         while ((beg < len) && (str.charAt(beg) <= ' ')) beg++;
         return beg == len;
     }
 
+    public static boolean isBlank(String str) {
+        return str.isEmpty() || isBlank((CharSequence) str);
+    }
+
     public static CharSequence canonicalizeWhitespace(CharSequence str) {
         final int len = str.length();
         StringBuilder sb = new StringBuilder(len);

data/ext/java/nokogiri/internals/ParserContext.java

@@ -43,6 +43,7 @@ import java.io.StringReader;
 import java.net.URI;
 import java.nio.charset.Charset;
 import java.nio.charset.UnsupportedCharsetException;
+import java.util.concurrent.Callable;
 
 import org.jruby.Ruby;
 import org.jruby.RubyClass;
@@ -54,7 +55,6 @@ import org.jruby.runtime.builtin.IRubyObject;
 import org.jruby.util.ByteList;
 import org.jruby.util.TypeConverter;
 import org.xml.sax.InputSource;
-import org.xml.sax.SAXException;
 
 /**
  * Base class for the various parser contexts.  Handles converting
@@ -63,7 +63,7 @@ import org.xml.sax.SAXException;
  * @author Patrick Mahoney <pat@polycrystal.org>
  * @author Yoko Harada <yokolet@gmail.com>
  */
-public class ParserContext extends RubyObject {
+public abstract class ParserContext extends RubyObject {
     protected InputSource source = null;
     protected IRubyObject detected_encoding = null;
     protected int stringDataSize = -1;
@@ -229,23 +229,23 @@ public class ParserContext extends RubyObject {
         protected static final long NOCDATA = 16384;
         protected static final long NOXINCNODE = 32768;
 
-        public boolean strict;
-        public boolean recover;
-        public boolean noEnt;
-        public boolean dtdLoad;
-        public boolean dtdAttr;
-        public boolean dtdValid;
-        public boolean noError;
-        public boolean noWarning;
-        public boolean pedantic;
-        public boolean noBlanks;
-        public boolean sax1;
-        public boolean xInclude;
-        public boolean noNet;
-        public boolean noDict;
-        public boolean nsClean;
-        public boolean noCdata;
-        public boolean noXIncNode;
+        public final boolean strict;
+        public final boolean recover;
+        public final boolean noEnt;
+        public final boolean dtdLoad;
+        public final boolean dtdAttr;
+        public final boolean dtdValid;
+        public final boolean noError;
+        public final boolean noWarning;
+        public final boolean pedantic;
+        public final boolean noBlanks;
+        public final boolean sax1;
+        public final boolean xInclude;
+        public final boolean noNet;
+        public final boolean noDict;
+        public final boolean nsClean;
+        public final boolean noCdata;
+        public final boolean noXIncNode;
 
         protected static boolean test(long options, long mask) {
             return ((options & mask) == mask);
@@ -272,6 +272,7 @@ public class ParserContext extends RubyObject {
         }
     }
 
+    /*
     public static class NokogiriXInlcudeEntityResolver implements org.xml.sax.EntityResolver {
         InputSource source;
         public NokogiriXInlcudeEntityResolver(InputSource source) {
@@ -285,6 +286,20 @@ public class ParserContext extends RubyObject {
             if (publicId != null) source.setPublicId(publicId);
             return source;
         }
+    } */
+
+    public static abstract class ParserTask<T extends ParserContext> implements Callable<T> {
+
+        protected final ThreadContext context; // TODO does not seem like a good idea!?
+        protected final IRubyObject handler;
+        protected final T parser;
+
+        protected ParserTask(ThreadContext context, IRubyObject handler, T parser) {
+            this.context = context;
+            this.handler = handler;
+            this.parser = parser;
+        }
+
     }
 
 }

data/ext/java/nokogiri/internals/ReaderNode.java

@@ -32,13 +32,6 @@
 
 package nokogiri.internals;
 
-import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
-import static nokogiri.internals.NokogiriHelpers.isNamespace;
-import static nokogiri.internals.NokogiriHelpers.isXmlBase;
-import static nokogiri.internals.NokogiriHelpers.rubyStringToString;
-import static nokogiri.internals.NokogiriHelpers.stringOrBlank;
-import static nokogiri.internals.NokogiriHelpers.stringOrNil;
-
 import java.util.ArrayList;
 import java.util.HashMap;
 import java.util.List;
@@ -61,6 +54,8 @@ import org.jruby.runtime.builtin.IRubyObject;
 import org.w3c.dom.Attr;
 import org.w3c.dom.Document;
 
+import static nokogiri.internals.NokogiriHelpers.*;
+
 /**
  * Abstract class of Node for XmlReader.
  *
@@ -263,7 +258,7 @@ public abstract class ReaderNode {
             super(runtime);
             nodeType = ReaderNodeType.END_ELEMENT.getValue();
             this.uri = "".equals(uri) ? null : uri;
-            this.localName = localName.trim().length() > 0 ? localName : qName;
+            this.localName = ! isBlank(localName) ? localName : qName;
             this.name = qName;
             parsePrefix(qName);
             this.depth = depth;
@@ -299,7 +294,7 @@ public abstract class ReaderNode {
             super(runtime);
             this.nodeType = ReaderNodeType.ELEMENT.getValue();
             this.uri = "".equals(uri) ? null : uri;
-            this.localName = localName.trim().length() > 0 ? localName : qName;
+            this.localName = ! isBlank(localName) ? localName : qName;
             this.name = qName;
             parsePrefix(qName);
             this.depth = depth;
@@ -475,7 +470,7 @@ public abstract class ReaderNode {
             this.localName = "#text";
             this.name = "#text";
             this.depth = depth;
-            if (content.trim().length() > 0) nodeType = ReaderNodeType.TEXT.getValue();
+            if (!isBlank(content)) nodeType = ReaderNodeType.TEXT.getValue();
             else nodeType = ReaderNodeType.SIGNIFICANT_WHITESPACE.getValue();
             if (!langStack.isEmpty()) this.lang = langStack.peek();
             if (!xmlBaseStack.isEmpty()) this.xmlBase = xmlBaseStack.peek();
@@ -491,4 +486,5 @@ public abstract class ReaderNode {
             return value;
         }
     }
+
 }

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

@@ -35,7 +35,8 @@ package nokogiri.internals;
 import static nokogiri.internals.NokogiriHelpers.canonicalizeWhitespace;
 import static nokogiri.internals.NokogiriHelpers.encodeJavaString;
 import static nokogiri.internals.NokogiriHelpers.isNamespace;
-import static nokogiri.internals.NokogiriHelpers.isWhitespaceText;
+import static nokogiri.internals.NokogiriHelpers.isBlank;
+import static nokogiri.internals.NokogiriHelpers.shouldEncode;
 
 import java.nio.charset.Charset;
 import java.nio.charset.CharsetEncoder;
@@ -737,13 +738,13 @@ public class SaveContextVisitor {
         CharSequence textContent = text.getNodeValue();
         if (canonical) {
             c14nNodeList.add(text);
-            if (isWhitespaceText(textContent)) {
+            if (isBlank(textContent)) {
                 buffer.append(canonicalizeWhitespace(textContent));
                 return true;
             }
         }
 
-        if (NokogiriHelpers.shouldEncode(text) && !isHtmlScript(text) && !isHtmlStyle(text)) {
+        if (shouldEncode(text) && !isHtmlScript(text) && !isHtmlStyle(text)) {
             textContent = encodeJavaString(textContent);
         }
 

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

@@ -33,6 +33,7 @@
 package nokogiri.internals;
 
 import static nokogiri.internals.NokogiriHelpers.getNokogiriClass;
+import static nokogiri.internals.NokogiriHelpers.isBlank;
 
 import java.io.IOException;
 import java.util.ArrayList;
@@ -47,6 +48,7 @@ import org.apache.xerces.parsers.DOMParser;
 import org.jruby.Ruby;
 import org.jruby.RubyArray;
 import org.jruby.RubyClass;
+import org.jruby.RubyFixnum;
 import org.jruby.exceptions.RaiseException;
 import org.jruby.runtime.ThreadContext;
 import org.jruby.runtime.builtin.IRubyObject;
@@ -64,6 +66,7 @@ import org.xml.sax.SAXException;
  * @author Yoko Harada <yokolet@gmail.com>
  */
 public class XmlDomParserContext extends ParserContext {
+
     protected static final String FEATURE_LOAD_EXTERNAL_DTD =
         "http://apache.org/xml/features/nonvalidating/load-external-dtd";
     protected static final String FEATURE_LOAD_DTD_GRAMMAR =
@@ -90,13 +93,13 @@ public class XmlDomParserContext extends ParserContext {
     
     public XmlDomParserContext(Ruby runtime, IRubyObject encoding, IRubyObject options) {
         super(runtime);
-        this.options = new ParserContext.Options((Long)options.toJava(Long.class));
+        this.options = new ParserContext.Options(RubyFixnum.fix2long(options));
         java_encoding = NokogiriHelpers.getValidEncoding(runtime, encoding);
         ruby_encoding = encoding;
         initErrorHandler();
         initParser(runtime);
     }
-    
+
     protected void initErrorHandler() {
         if (options.recover) {
             errorHandler = new NokogiriNonStrictErrorHandler(options.noError, options.noWarning);
@@ -274,7 +277,7 @@ public class XmlDomParserContext extends ParserContext {
     }
     
     private static void findEmptyTexts(Node node, List<Node> emptyNodes) {
-        if (node.getNodeType() == Node.TEXT_NODE && "".equals(node.getTextContent().trim())) {
+        if (node.getNodeType() == Node.TEXT_NODE && isBlank(node.getTextContent())) {
             emptyNodes.add(node);
         } else {
             NodeList children = node.getChildNodes();

data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java

@@ -75,7 +75,7 @@ public class InclusiveNamespaces extends ElementProxy {
             prefixList = new TreeSet<String>(prefixes);
         }
 
-        StringBuilder sb = new StringBuilder();
+        StringBuilder sb = new StringBuilder(prefixList.size() * 8);
         for (String prefix : prefixList) {
             if (prefix.equals("xmlns")) {
                 sb.append("#default ");
@@ -83,9 +83,10 @@ public class InclusiveNamespaces extends ElementProxy {
                 sb.append(prefix).append(' ');
             }
         }
+        int last = sb.length() - 1;
+        while (last >= 0 && sb.charAt(last) == ' ') sb.setLength(last--); // trim
         
-        this.constructionElement.setAttributeNS(
-            null, InclusiveNamespaces._ATT_EC_PREFIXLIST, sb.toString().trim());
+        this.constructionElement.setAttributeNS(null, InclusiveNamespaces._ATT_EC_PREFIXLIST, sb.toString());
     }
 
     /**

data/ext/nokogiri/extconf.rb

@@ -434,7 +434,7 @@ end
 
 if RbConfig::MAKEFILE_CONFIG['CC'] =~ /gcc/
   $CFLAGS << " -O3" unless $CFLAGS[/-O\d/]
-  $CFLAGS << " -Wall -Wcast-qual -Wwrite-strings -Wconversion -Wmissing-noreturn -Winline"
+  $CFLAGS << " -Wall -Wcast-qual -Wwrite-strings -Wextra -Wmissing-noreturn -Winline"
 end
 
 case

data/ext/nokogiri/html_element_description.c

@@ -8,7 +8,7 @@
  */
 static VALUE required_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -33,7 +33,7 @@ static VALUE required_attributes(VALUE self)
  */
 static VALUE deprecated_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -58,7 +58,7 @@ static VALUE deprecated_attributes(VALUE self)
  */
 static VALUE optional_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -83,7 +83,7 @@ static VALUE optional_attributes(VALUE self)
  */
 static VALUE default_sub_element(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if (description->defaultsubelt)
@@ -100,7 +100,7 @@ static VALUE default_sub_element(VALUE self)
  */
 static VALUE sub_elements(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -125,7 +125,7 @@ static VALUE sub_elements(VALUE self)
  */
 static VALUE description(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   return NOKOGIRI_STR_NEW2(description->desc);
@@ -139,7 +139,7 @@ static VALUE description(VALUE self)
  */
 static VALUE inline_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->isinline) return Qtrue;
@@ -154,7 +154,7 @@ static VALUE inline_eh(VALUE self)
  */
 static VALUE deprecated_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->depr) return Qtrue;
@@ -169,7 +169,7 @@ static VALUE deprecated_eh(VALUE self)
  */
 static VALUE empty_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->empty) return Qtrue;
@@ -184,7 +184,7 @@ static VALUE empty_eh(VALUE self)
  */
 static VALUE save_end_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->saveEndTag) return Qtrue;
@@ -199,7 +199,7 @@ static VALUE save_end_tag_eh(VALUE self)
  */
 static VALUE implied_end_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->endTag) return Qtrue;
@@ -214,7 +214,7 @@ static VALUE implied_end_tag_eh(VALUE self)
  */
 static VALUE implied_start_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->startTag) return Qtrue;
@@ -229,7 +229,7 @@ static VALUE implied_start_tag_eh(VALUE self)
  */
 static VALUE name(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(NULL == description->name) return Qnil;
@@ -249,7 +249,7 @@ static VALUE get_description(VALUE klass, VALUE tag_name)
   );
 
   if(NULL == description) return Qnil;
-  return Data_Wrap_Struct(klass, 0, 0, (void *)description);
+  return Data_Wrap_Struct(klass, 0, 0, (void *)(uintptr_t)description);
 }
 
 VALUE cNokogiriHtmlElementDescription ;

data/ext/nokogiri/xml_cdata.c

@@ -17,15 +17,17 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   VALUE content;
   VALUE rest;
   VALUE rb_node;
-  const xmlChar *content_str;
-  int content_str_len;
+  xmlChar *content_str = NULL;
+  int content_str_len = 0;
 
   rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
 
   Data_Get_Struct(doc, xmlDoc, xml_doc);
 
-  content_str = NIL_P(content) ? NULL : (const xmlChar *)StringValueCStr(content);
-  content_str_len = (content_str == NULL) ? 0 : strlen(content_str);
+  if (!NIL_P(content)) {
+    content_str = (xmlChar *)StringValuePtr(content);
+    content_str_len = RSTRING_LEN(content);
+  }
 
   node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);