nokogiri 1.3.1-java → 1.3.2-java

data/ext/nokogiri/xml_sax_parser.c

@@ -1,5 +1,11 @@
 #include <nokogiri.h>
 
+#define STRING_OR_NULL(str) \
+   (RTEST(str) ? StringValuePtr(str) : NULL)
+
+#define RBSTR_OR_QNIL(_str, rb_enc) \
+  (_str ? NOKOGIRI_STR_NEW2(_str, STRING_OR_NULL(rb_enc)) : Qnil)
+
 /*
  * call-seq:
  *  parse_memory(data)
@@ -89,7 +95,7 @@ static void start_element(void * ctx, const xmlChar *name, const xmlChar **atts)
   if(atts) {
     while((attr = atts[i]) != NULL) {
       rb_funcall(attributes, rb_intern("<<"), 1,
-          NOKOGIRI_STR_NEW2(attr, RTEST(enc) ? StringValuePtr(enc) : NULL)
+          NOKOGIRI_STR_NEW2(attr, STRING_OR_NULL(enc))
       );
       i++;
     }
@@ -98,7 +104,7 @@ static void start_element(void * ctx, const xmlChar *name, const xmlChar **atts)
   rb_funcall( doc,
               rb_intern("start_element"),
               2,
-              NOKOGIRI_STR_NEW2(name, RTEST(enc) ? StringValuePtr(enc) : NULL),
+              NOKOGIRI_STR_NEW2(name, STRING_OR_NULL(enc)),
               attributes
   );
 }
@@ -109,19 +115,51 @@ static void end_element(void * ctx, const xmlChar *name)
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
   rb_funcall(doc, rb_intern("end_element"), 1,
-      NOKOGIRI_STR_NEW2(name, RTEST(enc) ? StringValuePtr(enc) : NULL)
+      NOKOGIRI_STR_NEW2(name, STRING_OR_NULL(enc))
   );
 }
 
-/**
- * start_element_ns was borrowed heavily from libxml-ruby. 
- */
+static VALUE attributes_as_list(
+  VALUE self,
+  int nb_attributes,
+  const xmlChar ** attributes)
+{
+  VALUE list = rb_ary_new2(nb_attributes);
+  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
+
+  VALUE attr_klass = rb_const_get(cNokogiriXmlSaxParser, rb_intern("Attribute"));
+  if (attributes) {
+    /* Each attribute is an array of [localname, prefix, URI, value, end] */
+    int i;
+    for (i = 0; i < nb_attributes * 5; i += 5) {
+      VALUE attribute = rb_funcall(attr_klass, rb_intern("new"), 4,
+        /* localname */
+        RBSTR_OR_QNIL(attributes[i + 0], enc),
+
+        /* prefix */
+        RBSTR_OR_QNIL(attributes[i + 1], enc),
+
+        /* URI */
+        RBSTR_OR_QNIL(attributes[i + 2], enc),
+
+        /* value */
+        NOKOGIRI_STR_NEW((const char*)attributes[i+3],
+          (attributes[i+4] - attributes[i+3]),
+          STRING_OR_NULL(enc))
+      );
+      rb_ary_push(list, attribute);
+    }
+  }
+
+  return list;
+}
+
 static void
 start_element_ns (
   void * ctx,
   const xmlChar * localname,
   const xmlChar * prefix,
-  const xmlChar * URI,
+  const xmlChar * uri,
   int nb_namespaces,
   const xmlChar ** namespaces,
   int nb_attributes,
@@ -132,61 +170,42 @@ start_element_ns (
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
 
-  VALUE attrHash = rb_hash_new();
-  VALUE nsHash = rb_hash_new();
+  VALUE attribute_list = attributes_as_list(self, nb_attributes, attributes);
 
-  if (attributes)
-  {
-    /* Each attribute is an array of [localname, prefix, URI, value, end] */
-    int i;
-    for (i = 0; i < nb_attributes * 5; i += 5)
-    {
-      rb_hash_aset( attrHash,
-                    NOKOGIRI_STR_NEW2((const char*)attributes[i+0], RTEST(enc) ? StringValuePtr(enc) : NULL),
-                    NOKOGIRI_STR_NEW((const char*)attributes[i+3], (attributes[i+4] - attributes[i+3]), RTEST(enc) ? StringValuePtr(enc) : NULL));
-    }
-  }
+  VALUE ns_list = rb_ary_new2(nb_namespaces);
 
-  if (namespaces)
-  {
+  if (namespaces) {
     int i;
     for (i = 0; i < nb_namespaces * 2; i += 2)
     {
-      rb_hash_aset( nsHash,
-                    namespaces[i+0] ? NOKOGIRI_STR_NEW2((const char*)namespaces[i+0], RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil,
-                    namespaces[i+1] ? NOKOGIRI_STR_NEW2((const char*)namespaces[i+1], RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil);
+      rb_ary_push(ns_list,
+        rb_ary_new3(2,
+          RBSTR_OR_QNIL(namespaces[i + 0], enc),
+          RBSTR_OR_QNIL(namespaces[i + 1], enc)
+        )
+      );
     }
   }
 
   rb_funcall( doc,
-              rb_intern("start_element_ns"),
+              rb_intern("start_element_namespace"),
               5,
-              NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL),
-              attrHash,
-              prefix ? NOKOGIRI_STR_NEW2(prefix, RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil,
-              URI ? NOKOGIRI_STR_NEW2(URI, RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil,
-              nsHash
+              NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
+              attribute_list,
+              RBSTR_OR_QNIL(prefix, enc),
+              RBSTR_OR_QNIL(uri, enc),
+              ns_list
   );
 
-  /* Call start element if it's there' */
-  if (rb_respond_to(doc, rb_intern("start_element")))
-  {
-    VALUE name;
-    if (prefix)
-    {
-      name = NOKOGIRI_STR_NEW2(prefix, RTEST(enc) ? StringValuePtr(enc) : NULL);
-      rb_funcall(name, rb_intern("<<"), 1, NOKOGIRI_STR_NEW2(":", RTEST(enc) ? StringValuePtr(enc) : NULL));
-      rb_funcall(name, rb_intern("<<"), 1, NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL));
-    }
-    else
-    {
-      name = NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL);
-    }
-    VALUE attrArray = rb_funcall(attrHash, rb_intern("to_a"), 0);
-    attrArray = rb_funcall(attrArray, rb_intern("flatten"), 0);
-    rb_funcall(doc, rb_intern("start_element"), 2, name, attrArray);
-  }
-
+  rb_funcall( self,
+              rb_intern("start_element_namespace"),
+              5,
+              NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
+              attribute_list,
+              RBSTR_OR_QNIL(prefix, enc),
+              RBSTR_OR_QNIL(uri, enc),
+              ns_list
+  );
 }
 
 /**
@@ -197,35 +216,23 @@ end_element_ns (
   void * ctx,
   const xmlChar * localname,
   const xmlChar * prefix,
-  const xmlChar * URI)
+  const xmlChar * uri)
 {
   VALUE self = (VALUE)ctx;
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
 
-  rb_funcall(doc, rb_intern("end_element_ns"), 3, 
-             NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL),
-             prefix ? NOKOGIRI_STR_NEW2(prefix, RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil,
-             URI ? NOKOGIRI_STR_NEW2(URI, RTEST(enc) ? StringValuePtr(enc) : NULL) : Qnil
+  rb_funcall(doc, rb_intern("end_element_namespace"), 3, 
+    NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
+    RBSTR_OR_QNIL(prefix, enc),
+    RBSTR_OR_QNIL(uri, enc)
   );
 
-  /* Call end element for old-times sake */
-  if (rb_respond_to(doc, rb_intern("end_element")))
-  {
-    VALUE name;
-    if (prefix)
-    {
-      name = NOKOGIRI_STR_NEW2(prefix, RTEST(enc) ? StringValuePtr(enc) : NULL);
-      rb_funcall(name, rb_intern("<<"), 1, NOKOGIRI_STR_NEW2(":", RTEST(enc) ? StringValuePtr(enc) : NULL));
-      rb_funcall(name, rb_intern("<<"), 1, NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL));
-    }
-    else
-    {
-      name = NOKOGIRI_STR_NEW2(localname, RTEST(enc) ? StringValuePtr(enc) : NULL);
-    }
-    rb_funcall(doc, rb_intern("end_element"), 1, name);
-  }
-
+  rb_funcall(self, rb_intern("end_element_namespace"), 3, 
+    NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
+    RBSTR_OR_QNIL(prefix, enc),
+    RBSTR_OR_QNIL(uri, enc)
+  );
 }
 
 static void characters_func(void * ctx, const xmlChar * ch, int len)
@@ -233,7 +240,7 @@ static void characters_func(void * ctx, const xmlChar * ch, int len)
   VALUE self = (VALUE)ctx;
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE str = NOKOGIRI_STR_NEW(ch, len, RTEST(enc) ? StringValuePtr(enc):NULL);
+  VALUE str = NOKOGIRI_STR_NEW(ch, len, STRING_OR_NULL(enc));
   rb_funcall(doc, rb_intern("characters"), 1, str);
 }
 
@@ -242,7 +249,7 @@ static void comment_func(void * ctx, const xmlChar * value)
   VALUE self = (VALUE)ctx;
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE str = NOKOGIRI_STR_NEW2(value, RTEST(enc) ? StringValuePtr(enc):NULL);
+  VALUE str = NOKOGIRI_STR_NEW2(value, STRING_OR_NULL(enc));
   rb_funcall(doc, rb_intern("comment"), 1, str);
 }
 
@@ -259,7 +266,7 @@ static void warning_func(void * ctx, const char *msg, ...)
   va_end(args);
 
   rb_funcall(doc, rb_intern("warning"), 1,
-      NOKOGIRI_STR_NEW2(message, RTEST(enc) ? StringValuePtr(enc) : NULL)
+      NOKOGIRI_STR_NEW2(message, STRING_OR_NULL(enc))
   );
   free(message);
 }
@@ -277,7 +284,7 @@ static void error_func(void * ctx, const char *msg, ...)
   va_end(args);
 
   rb_funcall(doc, rb_intern("error"), 1,
-      NOKOGIRI_STR_NEW2(message, RTEST(enc) ? StringValuePtr(enc) : NULL)
+      NOKOGIRI_STR_NEW2(message, STRING_OR_NULL(enc))
   );
   free(message);
 }
@@ -288,7 +295,7 @@ static void cdata_block(void * ctx, const xmlChar * value, int len)
   VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
   VALUE doc = rb_funcall(self, rb_intern("document"), 0);
   VALUE string =
-    NOKOGIRI_STR_NEW(value, len, RTEST(enc) ? StringValuePtr(enc) : NULL);
+    NOKOGIRI_STR_NEW(value, len, STRING_OR_NULL(enc));
   rb_funcall(doc, rb_intern("cdata_block"), 1, string);
 }
 

data/ext/nokogiri/xml_schema.c

@@ -78,9 +78,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
   if(NULL == schema) {
     xmlErrorPtr error = xmlGetLastError();
     if(error)
-      rb_funcall(rb_mKernel, rb_intern("raise"), 1,
-          Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error)
-      );
+      Nokogiri_error_raise(NULL, error);
     else
       rb_raise(rb_eRuntimeError, "Could not parse document");
 
@@ -93,6 +91,56 @@ static VALUE read_memory(VALUE klass, VALUE content)
   return rb_schema;
 }
 
+/*
+ * call-seq:
+ *  from_document(doc)
+ *
+ * Create a new Schema from the Nokogiri::XML::Document +doc+
+ */
+static VALUE from_document(VALUE klass, VALUE document)
+{
+  xmlDocPtr doc;
+  Data_Get_Struct(document, xmlDoc, doc);
+
+  // In case someone passes us a node. ugh.
+  doc = doc->doc;
+
+  xmlSchemaParserCtxtPtr ctx = xmlSchemaNewDocParserCtxt(doc);
+
+  VALUE errors = rb_ary_new();
+  xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
+
+#ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
+  xmlSchemaSetParserStructuredErrors(
+    ctx,
+    Nokogiri_error_array_pusher,
+    (void *)errors
+  );
+#endif
+
+  xmlSchemaPtr schema = xmlSchemaParse(ctx);
+
+  xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlSchemaFreeParserCtxt(ctx);
+
+  if(NULL == schema) {
+    xmlErrorPtr error = xmlGetLastError();
+    if(error)
+      Nokogiri_error_raise(NULL, error);
+    else
+      rb_raise(rb_eRuntimeError, "Could not parse document");
+
+    return Qnil;
+  }
+
+  VALUE rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
+  rb_iv_set(rb_schema, "@errors", errors);
+
+  return rb_schema;
+
+  return Qnil;
+}
+
 VALUE cNokogiriXmlSchema;
 void init_xml_schema()
 {
@@ -103,5 +151,6 @@ void init_xml_schema()
   cNokogiriXmlSchema = klass;
 
   rb_define_singleton_method(klass, "read_memory", read_memory, 1);
+  rb_define_singleton_method(klass, "from_document", from_document, 1);
   rb_define_private_method(klass, "validate_document", validate_document, 1);
 }

data/ext/nokogiri/xml_syntax_error.c

@@ -166,6 +166,13 @@ void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error)
   rb_ary_push(list,  Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
 }
 
+void Nokogiri_error_raise(void * ctx, xmlErrorPtr error)
+{
+  rb_funcall(rb_mKernel, rb_intern("raise"), 1,
+      Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error)
+  );
+}
+
 VALUE Nokogiri_wrap_xml_syntax_error(VALUE klass, xmlErrorPtr error)
 {
   if(!klass) klass = cNokogiriXmlSyntaxError;

data/ext/nokogiri/xml_syntax_error.h

@@ -6,6 +6,7 @@
 void init_xml_syntax_error();
 VALUE Nokogiri_wrap_xml_syntax_error(VALUE klass, xmlErrorPtr error);
 void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error);
+void Nokogiri_error_raise(void * ctx, xmlErrorPtr error);
 
 extern VALUE cNokogiriXmlSyntaxError;
 #endif

data/lib/nokogiri.rb

@@ -55,9 +55,9 @@ module Nokogiri
     def parse string, url = nil, encoding = nil, options = nil
       doc =
         if string =~ /^\s*<[^Hh>]*html/i # Probably html
-          Nokogiri::HTML::Document.parse(string, url, encoding, options || 2145)
+          Nokogiri::HTML::Document.parse(string, url, encoding, options || XML::ParseOptions::DEFAULT_HTML)
         else
-          Nokogiri::XML::Document.parse(string, url, encoding, options || 2159)
+          Nokogiri::XML::Document.parse(string, url, encoding, options || XML::ParseOptions::DEFAULT_XML)
         end
       yield doc if block_given?
       doc

data/lib/nokogiri/css/parser.rb

@@ -43,7 +43,7 @@ module Nokogiri
         def parse selector
           @warned ||= false
           unless @warned
-            $stderr.puts('Nokogiri::CSS::Parser.parse is deprecated, call Nokogiri::CSS.parse()')
+            $stderr.puts('Nokogiri::CSS::Parser.parse is deprecated, call Nokogiri::CSS.parse(), this will be removed August 1st or version 1.4.0 (whichever is first)')
             @warned = true
           end
           new.parse selector
@@ -59,7 +59,7 @@ module Nokogiri
 
       # Get the xpath for +string+ using +options+
       def xpath_for string, options={}
-        key = string + options[:ns].to_s
+        key = "#{string}#{options[:ns]}#{options[:prefix]}"
         v = self.class[key]
         return v if v
 

data/lib/nokogiri/ffi/io_callbacks.rb

@@ -2,22 +2,30 @@ module Nokogiri
   module IoCallbacks # :nodoc:
     
     class << self
+
+      def plain_old_reader(io)
+        lambda do |ctx, buffer, len|
+          string = io.read(len)
+          return 0 if string.nil?
+          buffer.put_bytes(0, string, 0, string.length)
+          string.length
+        end
+      end
+
       if defined?(FFI::IO.native_read)
-        def reader(io)
-          lambda do |ctx, buffer, len|
-            rcode = FFI::IO.native_read(io, buffer, len)
-            (rcode < 0) ? 0 : rcode
+        def ffi_io_native_reader(io)
+          if io.is_a?(StringIO)
+            plain_old_reader(io)
+          else
+            lambda do |ctx, buffer, len|
+              rcode = FFI::IO.native_read(io, buffer, len)
+              (rcode < 0) ? 0 : rcode
+            end
           end
         end
+        alias :reader :ffi_io_native_reader
       else
-        def reader(io) # TODO: this can be removed once JRuby 1.3.0RC2 and ruby-ffi 0.4.0 are both released
-          lambda do |ctx, buffer, len|
-            string = io.read(len)
-            return 0 if string.nil?
-            buffer.put_bytes(0, string, 0, string.length)
-            string.length
-          end
-        end
+        alias :reader :plain_old_reader
       end
 
       def writer(io)

data/lib/nokogiri/ffi/libxml.rb

@@ -118,6 +118,7 @@ module Nokogiri
     attach_function :xmlStringGetNodeList, [:pointer, :pointer], :pointer # second arg should be a :string, but we only ship the results of xmlEncodeEntitiesReentrant, so let's optimize.
     attach_function :xmlNewNs, [:pointer, :string, :string], :pointer
     attach_function :xmlNewNsProp, [:pointer, :pointer, :string, :string], :pointer
+    attach_function :xmlSearchNs, [:pointer, :pointer, :string], :pointer
     attach_function :xmlSearchNsByHref, [:pointer, :pointer, :string], :pointer
     attach_function :xmlGetIntSubset, [:pointer], :pointer
     attach_function :xmlBufferCreate, [], :pointer
@@ -126,6 +127,11 @@ module Nokogiri
     attach_function :xmlNewDocProp, [:pointer, :string, :string], :pointer
     attach_function :xmlFreePropList, [:pointer], :void
 
+    # valid.c
+    attach_function :xmlNewValidCtxt, [], :pointer
+    attach_function :xmlValidateDtd, [:pointer, :pointer, :pointer], :int
+    attach_function :xmlFreeValidCtxt, [:pointer], :void
+
     # xmlsave.c
     attach_function :xmlDocDumpMemory, [:pointer, :pointer, :pointer], :void
     attach_function :xmlNodeDump, [:pointer, :pointer, :pointer, :int, :int], :int
@@ -240,6 +246,7 @@ module Nokogiri
     attach_function :xmlSchemaSetParserStructuredErrors, [:pointer, :syntax_error_handler, :pointer], :void unless Nokogiri.is_2_6_16?
     attach_function :xmlSchemaParse, [:pointer], :pointer
     attach_function :xmlSchemaFreeParserCtxt, [:pointer], :void
+    attach_function :xmlSchemaNewDocParserCtxt, [:pointer], :pointer
 
     # relaxng.c
     attach_function :xmlRelaxNGNewValidCtxt, [:pointer], :pointer
@@ -250,6 +257,7 @@ module Nokogiri
     attach_function :xmlRelaxNGSetParserStructuredErrors, [:pointer, :syntax_error_handler, :pointer], :void unless Nokogiri.is_2_6_16?
     attach_function :xmlRelaxNGParse, [:pointer], :pointer
     attach_function :xmlRelaxNGFreeParserCtxt, [:pointer], :void
+    attach_function :xmlRelaxNGNewDocParserCtxt, [:pointer], :pointer
 
     # helpers
     def self.pointer_offset(n)