nokogiri 1.8.5-java → 1.9.0-java

data/ext/nokogiri/extconf.rb

@@ -400,9 +400,11 @@ if openbsd? && !using_system_libraries?
   ENV['CFLAGS'] = "#{ENV['CFLAGS']} -I /usr/local/include"
 end
 
-RbConfig::MAKEFILE_CONFIG['CC'] = ENV['CC'] if ENV['CC']
+if ENV['CC']
+  RbConfig::CONFIG['CC'] = RbConfig::MAKEFILE_CONFIG['CC'] = ENV['CC']
+end
 # use same c compiler for libxml and libxslt
-ENV['CC'] = RbConfig::MAKEFILE_CONFIG['CC']
+ENV['CC'] = RbConfig::CONFIG['CC']
 
 $LIBS << " #{ENV["LIBS"]}"
 
@@ -432,7 +434,7 @@ if RUBY_PLATFORM =~ /mingw/i
   $CPPFLAGS << ' "-Idummypath"'
 end
 
-if RbConfig::MAKEFILE_CONFIG['CC'] =~ /gcc/
+if RbConfig::CONFIG['CC'] =~ /gcc/
   $CFLAGS << " -O3" unless $CFLAGS[/-O\d/]
   $CFLAGS << " -Wall -Wcast-qual -Wwrite-strings -Wmissing-noreturn -Winline"
 end
@@ -460,7 +462,7 @@ else
   # 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 'mini_portile2', '~> 2.3.0'
+  gem 'mini_portile2', '~> 2.4.0'
   require 'mini_portile2'
   message "Using mini_portile version #{MiniPortile::VERSION}\n"
 
@@ -560,7 +562,8 @@ EOM
       *(libiconv_recipe ? "--with-iconv=#{libiconv_recipe.path}" : iconv_configure_flags),
       "--with-c14n",
       "--with-debug",
-      "--with-threads"
+      "--with-threads",
+      *(darwin? ? ["RANLIB=/usr/bin/ranlib", "AR=/usr/bin/ar"] : "")
     ]
   end
 
@@ -573,7 +576,8 @@ EOM
       "--without-python",
       "--without-crypto",
       "--with-debug",
-      "--with-libxml-prefix=#{sh_export_path(libxml2_recipe.path)}"
+      "--with-libxml-prefix=#{sh_export_path(libxml2_recipe.path)}",
+      *(darwin? ? ["RANLIB=/usr/bin/ranlib", "AR=/usr/bin/ar"] : "")
     ]
   end
 

data/ext/nokogiri/xml_attr.c

@@ -4,37 +4,40 @@
  * call-seq:
  *  value=(content)
  *
- * Set the value for this Attr to +content+
+ * Set the value for this Attr to +content+. Use `nil` to remove the value
+ * (e.g., a HTML boolean attribute).
  */
 static VALUE set_value(VALUE self, VALUE content)
 {
   xmlAttrPtr attr;
-  Data_Get_Struct(self, xmlAttr, attr);
+  xmlChar *value;
+  xmlNode *cur;
 
-  if (attr->children) { xmlFreeNodeList(attr->children); }
+  Data_Get_Struct(self, xmlAttr, attr);
 
+  if (attr->children) {
+    xmlFreeNodeList(attr->children);
+  }
   attr->children = attr->last = NULL;
 
-  if (content) {
-    xmlChar *buffer;
-    xmlNode *tmp;
-
-    /* Encode our content */
-    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
+  if (content == Qnil) {
+    return content;
+  }
 
-    attr->children = xmlStringGetNodeList(attr->doc, buffer);
-    attr->last = NULL;
-    tmp = attr->children;
+  value = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
+  if (xmlStrlen(value) == 0) {
+    attr->children = xmlNewDocText(attr->doc, value);
+  } else {
+    attr->children = xmlStringGetNodeList(attr->doc, value);
+  }
+  xmlFree(value);
 
-    /* Loop through the children */
-    for(tmp = attr->children; tmp; tmp = tmp->next) {
-      tmp->parent = (xmlNode *)attr;
-      tmp->doc = attr->doc;
-      if (tmp->next == NULL) { attr->last = tmp; }
+  for (cur = attr->children; cur; cur = cur->next) {
+    cur->parent = (xmlNode *)attr;
+    cur->doc = attr->doc;
+    if (cur->next == NULL) {
+      attr->last = cur;
     }
-
-    /* Free up memory */
-    xmlFree(buffer);
   }
 
   return content;
@@ -74,7 +77,9 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
   rb_obj_call_init(rb_node, argc, argv);
 
-  if (rb_block_given_p()) { rb_yield(rb_node); }
+  if (rb_block_given_p()) {
+    rb_yield(rb_node);
+  }
 
   return rb_node;
 }

data/ext/nokogiri/xml_document.c

@@ -7,7 +7,10 @@ static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
     xmlFreePropList((xmlAttrPtr)node);
     break;
   case XML_NAMESPACE_DECL:
-    xmlFree(node);
+    xmlFreeNs((xmlNsPtr)node);
+    break;
+  case XML_DTD_NODE:
+    xmlFreeDtd((xmlDtdPtr)node);
     break;
   default:
     if(node->parent == NULL) {

data/ext/nokogiri/xml_namespace.c

@@ -63,7 +63,7 @@ static int part_of_an_xpath_node_set_eh(xmlNsPtr node)
 
 VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node)
 {
-  VALUE ns, document, node_cache;
+  VALUE ns = 0, document, node_cache;
 
   assert(doc->type == XML_DOCUMENT_NODE || doc->type == XML_HTML_DOCUMENT_NODE);
 
@@ -71,8 +71,6 @@ VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node)
 
   if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
 
-  ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
-
   if (DOC_RUBY_OBJECT_TEST(doc)) {
     document = DOC_RUBY_OBJECT(doc);
 
@@ -91,6 +89,8 @@ VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node)
     }
 
     rb_iv_set(ns, "@document", document);
+  } else {
+    ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
   }
 
   node->_private = (void *)ns;

data/ext/nokogiri/xml_namespace.h

@@ -7,8 +7,7 @@ void init_xml_namespace();
 
 extern VALUE cNokogiriXmlNamespace ;
 
-VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node) ;
-VALUE Nokogiri_wrap_xml_namespace2(VALUE document, xmlNsPtr node) ;
+VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node);
 
 #define NOKOGIRI_NAMESPACE_EH(node) ((node)->type == XML_NAMESPACE_DECL)
 

data/ext/nokogiri/xml_node.c

@@ -31,13 +31,13 @@ typedef xmlNodePtr (*pivot_reparentee_func)(xmlNodePtr, xmlNodePtr);
 static void relink_namespace(xmlNodePtr reparented)
 {
   xmlNodePtr child;
-  xmlNsPtr ns;
 
   if (reparented->type != XML_ATTRIBUTE_NODE &&
       reparented->type != XML_ELEMENT_NODE) { return; }
 
   if (reparented->ns == NULL || reparented->ns->prefix == NULL) {
-    xmlChar *name = 0, *prefix = 0;
+    xmlNsPtr ns = NULL;
+    xmlChar *name = NULL, *prefix = NULL;
 
     name = xmlSplitQName2(reparented->name, &prefix);
 
@@ -99,6 +99,25 @@ static void relink_namespace(xmlNodePtr reparented)
     }
   }
 
+  /*
+   *  Search our parents for an existing definition of current namespace,
+   *  because the definition it's pointing to may have just been removed nsDef.
+   *
+   *  And although that would technically probably be OK, I'd feel better if we
+   *  referred to a namespace that's still present in a node's nsDef somewhere
+   *  in the doc.
+   */
+  if (reparented->ns) {
+    xmlNsPtr ns = xmlSearchNs(reparented->doc, reparented, reparented->ns->prefix);
+    if (ns
+        && ns != reparented->ns
+        && xmlStrEqual(ns->prefix, reparented->ns->prefix)
+        && xmlStrEqual(ns->href, reparented->ns->href)
+       ) {
+      xmlSetNs(reparented, ns);
+    }
+  }
+
   /* Only walk all children if there actually is a namespace we need to */
   /* reparent. */
   if (NULL == reparented->ns) { return; }
@@ -282,7 +301,11 @@ ok:
     }
 
     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" */
+      /*
+       *  issue #391, where new node's prefix may become the string "default"
+       *  see libxml2 tree.c xmlNewReconciliedNs which implements this behavior.
+       */
+      xmlFree(reparentee->ns->prefix);
       reparentee->ns->prefix = NULL;
     }
   }
@@ -509,22 +532,39 @@ static VALUE internal_subset(VALUE self)
 /*
  * call-seq:
  *  dup
+ *  dup(depth)
+ *  dup(depth, new_parent_doc)
  *
- * Copy this node.  An optional depth may be passed in, but it defaults
- * to a deep copy.  0 is a shallow copy, 1 is a deep copy.
+ * Copy this node.
+ * An optional depth may be passed in. 0 is a shallow copy, 1 (the default) is a deep copy.
+ * An optional new_parent_doc may also be passed in, which will be the new
+ * node's parent document. Defaults to the current node's document.
+ * current document.
  */
 static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
 {
-  VALUE level;
+  VALUE r_level, r_new_parent_doc;
+  int level;
+  int n_args;
+  xmlDocPtr new_parent_doc;
   xmlNodePtr node, dup;
 
-  if(rb_scan_args(argc, argv, "01", &level) == 0) {
-    level = INT2NUM((long)1);
+  Data_Get_Struct(self, xmlNode, node);
+
+  n_args = rb_scan_args(argc, argv, "02", &r_level, &r_new_parent_doc);
+
+  if (n_args < 1) {
+    r_level = INT2NUM((long)1);
   }
+  level = (int)NUM2INT(r_level);
 
-  Data_Get_Struct(self, xmlNode, node);
+  if (n_args < 2) {
+    new_parent_doc = node->doc;
+  } else {
+    Data_Get_Struct(r_new_parent_doc, xmlDoc, new_parent_doc);
+  }
 
-  dup = xmlDocCopyNode(node, node->doc, (int)NUM2INT(level));
+  dup = xmlDocCopyNode(node, new_parent_doc, level);
   if(dup == NULL) { return Qnil; }
 
   nokogiri_root_node(dup);
@@ -1308,11 +1348,11 @@ static VALUE line(VALUE self)
  */
 static VALUE add_namespace_definition(VALUE self, VALUE prefix, VALUE href)
 {
-  xmlNodePtr node, namespacee;
+  xmlNodePtr node, namespace;
   xmlNsPtr ns;
 
   Data_Get_Struct(self, xmlNode, node);
-  namespacee = node ;
+  namespace = node ;
 
   ns = xmlSearchNs(
          node->doc,
@@ -1322,10 +1362,10 @@ static VALUE add_namespace_definition(VALUE self, VALUE prefix, VALUE href)
 
   if(!ns) {
     if (node->type != XML_ELEMENT_NODE) {
-      namespacee = node->parent;
+      namespace = node->parent;
     }
     ns = xmlNewNs(
-           namespacee,
+           namespace,
            (const xmlChar *)StringValueCStr(href),
            (const xmlChar *)(NIL_P(prefix) ? NULL : StringValueCStr(prefix))
          );
@@ -1333,7 +1373,7 @@ static VALUE add_namespace_definition(VALUE self, VALUE prefix, VALUE href)
 
   if (!ns) { return Qnil ; }
 
-  if(NIL_P(prefix) || node != namespacee) { xmlSetNs(node, ns); }
+  if(NIL_P(prefix) || node != namespace) { xmlSetNs(node, ns); }
 
   return Nokogiri_wrap_xml_namespace(node->doc, ns);
 }

data/lib/nokogiri/css/parser.rb

@@ -31,64 +31,64 @@ end
 
 racc_action_table = [
     24,    93,    56,    57,    33,    55,    94,    23,    24,    22,
-    12,    93,    33,    27,    89,    52,    92,    22,   -23,    25,
-   109,    98,    23,    33,    26,    18,    20,    25,    27,   -23,
+    12,    93,    33,    27,    35,    52,    44,    22,   -23,    25,
+    45,    98,    23,    33,    26,    18,    20,    25,    27,   -23,
     23,    24,    26,    18,    20,    33,    27,    11,    39,    24,
-    22,    23,    95,    33,    18,    91,    90,    27,    22,    12,
-    25,    90,    96,    23,    33,    26,    18,    20,    25,    27,
-   -23,    23,    24,    26,    18,    20,    33,    27,    74,    39,
-    99,    22,    23,    45,    24,    18,    33,   103,    27,    56,
-    87,    25,    60,    46,    23,    49,    26,    18,    20,   104,
-    27,    39,    24,    51,    23,    93,    44,    18,    26,    33,
-    27,    66,   106,    56,    58,   110,    60,    33,    85,    33,
-    86,    51,    56,    87,    39,    60,    26,    23,    88,    33,
-    18,    20,    39,    27,    39,    23,    35,    23,    18,    33,
-    18,    27,   nil,    27,    39,   nil,    33,    23,   nil,   nil,
-    18,    22,   nil,    27,    39,   101,   100,    23,   102,   nil,
-    18,    39,   nil,    27,    23,    82,    83,    18,    20,   nil,
-    27,   nil,   nil,   nil,    82,    83,    78,    79,    80,   nil,
-    81,   nil,   nil,   nil,    77,    78,    79,    80,   nil,    81,
-     4,     5,    43,    77,     4,     5,    10,   nil,    56,    87,
-     6,    60,     8,     7,     6,   nil,     8,     7,     4,     5,
-    10,    56,    87,   nil,    60,   nil,   nil,   nil,     6,   nil,
-     8,     7 ]
+    22,    23,    74,    33,    18,    91,    90,    27,    22,    12,
+    25,    24,   -23,    23,    85,    26,    18,    20,    25,    27,
+    66,    23,    24,    26,    18,    20,    33,    27,    86,    88,
+    51,    22,    89,    92,    24,    26,    56,    87,    95,    60,
+    90,    25,    96,    46,    23,    49,    26,    18,    20,    99,
+    27,    33,    33,    51,   103,   104,    56,    58,    26,    60,
+    93,   106,    33,    33,    33,   109,    39,    39,   110,    23,
+    23,   nil,    18,    18,    20,    27,    27,    39,    39,    39,
+    23,    23,    23,    18,    18,    18,    27,    27,    27,    33,
+    33,   101,   100,   nil,   102,    22,    56,    87,   nil,    60,
+    33,   nil,   nil,   nil,    39,    39,   nil,    23,    23,   nil,
+    18,    18,    20,    27,    27,    39,    82,    83,    23,    56,
+    87,    18,    60,   nil,    27,    82,    83,    78,    79,    80,
+   nil,    81,   nil,   nil,   nil,    77,    78,    79,    80,   nil,
+    81,     4,     5,    10,    77,     4,     5,    43,   nil,    56,
+    87,     6,    60,     8,     7,     6,   nil,     8,     7,     4,
+     5,    10,   nil,   nil,   nil,   nil,   nil,   nil,   nil,     6,
+   nil,     8,     7 ]
 
 racc_action_check = [
-     3,    58,    24,    24,     3,    24,    57,    15,    42,     3,
-    64,    57,    42,    15,    54,    24,    56,    42,    58,     3,
-    94,    64,     3,    31,     3,     3,     3,    42,     3,    22,
-    42,     9,    42,    42,    42,     9,    42,     1,    31,    43,
-     9,    31,    59,    43,    31,    55,    55,    31,    43,     1,
-     9,    60,    61,     9,    30,     9,     9,     9,    43,     9,
-    46,    43,    12,    43,    43,    43,    12,    43,    45,    30,
-    75,    12,    30,    21,    23,    30,    29,    84,    30,    93,
-    93,    12,    93,    23,    12,    23,    12,    12,    12,    86,
-    12,    29,    27,    23,    29,    87,    18,    29,    23,    25,
-    29,    27,    91,    25,    25,   106,    25,    28,    49,    62,
-    50,    27,    51,    51,    25,    51,    27,    25,    53,    14,
-    25,    25,    28,    25,    62,    28,    11,    62,    28,    32,
-    62,    28,   nil,    62,    14,   nil,    39,    14,   nil,   nil,
-    14,    39,   nil,    14,    32,    76,    76,    32,    76,   nil,
-    32,    39,   nil,    32,    39,    47,    47,    39,    39,   nil,
-    39,   nil,   nil,   nil,    48,    48,    47,    47,    47,   nil,
-    47,   nil,   nil,   nil,    47,    48,    48,    48,   nil,    48,
-    17,    17,    17,    48,     0,     0,     0,   nil,    90,    90,
-    17,    90,    17,    17,     0,   nil,     0,     0,    26,    26,
-    26,    92,    92,   nil,    92,   nil,   nil,   nil,    26,   nil,
-    26,    26 ]
+     3,    58,    24,    24,     3,    24,    57,    15,     9,     3,
+    64,    57,     9,    15,    11,    24,    18,     9,    58,     3,
+    21,    64,     3,    14,     3,     3,     3,     9,     3,    22,
+     9,    12,     9,     9,     9,    12,     9,     1,    14,    42,
+    12,    14,    45,    42,    14,    55,    55,    14,    42,     1,
+    12,    27,    46,    12,    49,    12,    12,    12,    42,    12,
+    27,    42,    43,    42,    42,    42,    43,    42,    50,    53,
+    27,    43,    54,    56,    23,    27,    51,    51,    59,    51,
+    60,    43,    61,    23,    43,    23,    43,    43,    43,    75,
+    43,    28,    25,    23,    84,    86,    25,    25,    23,    25,
+    87,    91,    29,    30,    31,    94,    28,    25,   106,    28,
+    25,   nil,    28,    25,    25,    28,    25,    29,    30,    31,
+    29,    30,    31,    29,    30,    31,    29,    30,    31,    32,
+    39,    76,    76,   nil,    76,    39,    90,    90,   nil,    90,
+    62,   nil,   nil,   nil,    32,    39,   nil,    32,    39,   nil,
+    32,    39,    39,    32,    39,    62,    47,    47,    62,    92,
+    92,    62,    92,   nil,    62,    48,    48,    47,    47,    47,
+   nil,    47,   nil,   nil,   nil,    47,    48,    48,    48,   nil,
+    48,     0,     0,     0,    48,    17,    17,    17,   nil,    93,
+    93,     0,    93,     0,     0,    17,   nil,    17,    17,    26,
+    26,    26,   nil,   nil,   nil,   nil,   nil,   nil,   nil,    26,
+   nil,    26,    26 ]
 
 racc_action_pointer = [
-   177,    37,   nil,    -2,   nil,   nil,   nil,   nil,   nil,    29,
-   nil,   126,    60,   nil,   113,   -17,   nil,   173,    85,   nil,
-   nil,    44,     0,    72,    -8,    93,   191,    90,   101,    70,
-    48,    17,   123,   nil,   nil,   nil,   nil,   nil,   nil,   130,
-   nil,   nil,     6,    37,   nil,    57,    31,   152,   161,    83,
-    81,   102,   nil,    95,    -9,    34,     4,    -1,   -11,    19,
-    39,    29,   103,   nil,    -2,   nil,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,    45,   135,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,    52,   nil,    78,    83,   nil,   nil,
-   178,    95,   191,    69,     7,   nil,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,   nil,    92,   nil,   nil,   nil,
+   174,    37,   nil,    -2,   nil,   nil,   nil,   nil,   nil,     6,
+   nil,    14,    29,   nil,    17,   -17,   nil,   178,     5,   nil,
+   nil,    -9,     0,    72,    -8,    86,   192,    49,    85,    96,
+    97,    98,   123,   nil,   nil,   nil,   nil,   nil,   nil,   124,
+   nil,   nil,    37,    60,   nil,    31,    23,   153,   162,    29,
+    39,    66,   nil,    46,    49,    34,    61,    -1,   -11,    55,
+    68,    59,   134,   nil,    -2,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,   nil,   nil,   nil,    64,   121,   nil,   nil,   nil,
+   nil,   nil,   nil,   nil,    69,   nil,    84,    88,   nil,   nil,
+   126,    94,   149,   179,    92,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,   nil,   nil,   nil,   nil,    95,   nil,   nil,   nil,
    nil ]
 
 racc_action_default = [
@@ -106,26 +106,26 @@ racc_action_default = [
    -41 ]
 
 racc_goto_table = [
-    53,    38,    13,     1,    54,    48,    62,    42,    34,    65,
-    37,    36,    63,    75,    84,    67,    68,    69,    70,    71,
-    62,    40,    41,    50,    47,   nil,    63,   nil,   nil,    64,
+    53,    38,    13,     1,    42,    48,    62,    37,    34,    65,
+    40,    36,    63,    75,    84,    67,    68,    69,    70,    71,
+    62,    41,    50,    47,    54,   nil,    63,   nil,   nil,    64,
    nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
    nil,    72,    73,   nil,   nil,   nil,   nil,   nil,   nil,    97,
    nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
    nil,   nil,   nil,   nil,   nil,   nil,   105,   nil,   107,   108 ]
 
 racc_goto_check = [
-    18,    12,     2,     1,    19,     9,     7,     5,     2,     9,
-     8,     2,    12,    17,    17,    12,    12,    12,    12,    12,
-     7,    10,    11,    15,    16,   nil,    12,   nil,   nil,     1,
+    18,    12,     2,     1,     5,     9,     7,     8,     2,     9,
+    10,     2,    12,    17,    17,    12,    12,    12,    12,    12,
+     7,    11,    15,    16,    19,   nil,    12,   nil,   nil,     1,
    nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
    nil,     2,     2,   nil,   nil,   nil,   nil,   nil,   nil,    12,
    nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,
    nil,   nil,   nil,   nil,   nil,   nil,    18,   nil,    18,    18 ]
 
 racc_goto_pointer = [
-   nil,     3,    -1,   nil,   nil,   -10,   nil,   -19,    -4,   -18,
-     6,     7,   -13,   nil,   nil,     0,     1,   -34,   -24,   -20,
+   nil,     3,    -1,   nil,   nil,   -13,   nil,   -19,    -7,   -18,
+    -5,     6,   -13,   nil,   nil,    -1,     0,   -34,   -24,     0,
    nil,   nil,   nil,   nil ]
 
 racc_goto_default = [