nokogiri 1.11.0.rc1-x86-linux → 1.11.1-x86-linux

data/ext/nokogiri/nokogiri.c

@@ -1,5 +1,7 @@
 #include <nokogiri.h>
 
+void init_test_global_handlers(); /* in lieu of test_global_handlers.h */
+
 VALUE mNokogiri ;
 VALUE mNokogiriXml ;
 VALUE mNokogiriHtml ;
@@ -7,12 +9,11 @@ VALUE mNokogiriXslt ;
 VALUE mNokogiriXmlSax ;
 VALUE mNokogiriHtmlSax ;
 
-#ifdef USE_INCLUDED_VASPRINTF
+#ifndef HAVE_VASPRINTF
 /*
- * I srsly hate windows.  it doesn't have vasprintf.
  * Thank you Geoffroy Couprie for this implementation of vasprintf!
  */
-int vasprintf (char **strp, const char *fmt, va_list ap)
+int vasprintf(char **strp, const char *fmt, va_list ap)
 {
   /* Mingw32/64 have a broken vsnprintf implementation that fails when
    * using a zero-byte limit in order to retrieve the required size for malloc.
@@ -21,23 +22,15 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
   char tmp[1];
   int len = vsnprintf (tmp, 1, fmt, ap) + 1;
   char *res = (char *)malloc((unsigned int)len);
-  if (res == NULL)
-      return -1;
+  if (res == NULL) {
+    return -1;
+  }
   *strp = res;
   return vsnprintf(res, (unsigned int)len, fmt, ap);
 }
 #endif
 
-void vasprintf_free (void *p)
-{
-  free(p);
-}
-
-#ifdef HAVE_RUBY_UTIL_H
 #include "ruby/util.h"
-#else
-#include "util.h"
-#endif
 
 void nokogiri_root_node(xmlNodePtr node)
 {
@@ -45,7 +38,7 @@ void nokogiri_root_node(xmlNodePtr node)
   nokogiriTuplePtr tuple;
 
   doc = node->doc;
-  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
   tuple = (nokogiriTuplePtr)doc->_private;
   st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
 }
@@ -54,7 +47,7 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
 {
   nokogiriTuplePtr tuple;
 
-  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
   tuple = (nokogiriTuplePtr)doc->_private;
   st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
 }
@@ -62,10 +55,10 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
 void Init_nokogiri()
 {
   xmlMemSetup(
-      (xmlFreeFunc)ruby_xfree,
-      (xmlMallocFunc)ruby_xmalloc,
-      (xmlReallocFunc)ruby_xrealloc,
-      ruby_strdup
+    (xmlFreeFunc)ruby_xfree,
+    (xmlMallocFunc)ruby_xmalloc,
+    (xmlReallocFunc)ruby_xrealloc,
+    ruby_strdup
   );
 
   mNokogiri         = rb_define_module("Nokogiri");
@@ -75,33 +68,26 @@ void Init_nokogiri()
   mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
   mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
 
-  rb_const_set( mNokogiri,
-                rb_intern("LIBXML_COMPILED_VERSION"),
-                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
-              );
-  rb_const_set( mNokogiri,
-                rb_intern("LIBXML_LOADED_VERSION"),
-                NOKOGIRI_STR_NEW2(xmlParserVersion)
-              );
-
-
-  rb_const_set( mNokogiri,
-                rb_intern("LIBXSLT_COMPILED_VERSION"),
-                NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION)
-              );
-  rb_const_set( mNokogiri,
-                rb_intern("LIBXSLT_LOADED_VERSION"),
-                NOKOGIRI_STR_NEW2(xsltEngineVersion)
-              );
-
-#ifdef NOKOGIRI_USE_PACKAGED_LIBRARIES
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qtrue);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
+
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xsltEngineVersion));
+
+#ifdef NOKOGIRI_PACKAGED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qtrue);
+#ifdef NOKOGIRI_PRECOMPILED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qtrue);
+#else
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+#endif
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
 #else
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qfalse);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), Qnil);
-  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), Qnil);
 #endif
 
 #ifdef LIBXML_ICONV_ENABLED
@@ -110,6 +96,10 @@ void Init_nokogiri()
   rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
 #endif
 
+#ifdef NOKOGIRI_OTHER_LIBRARY_VERSIONS
+  rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
+#endif
+
   xmlInitParser();
 
   init_xml_document();
@@ -144,4 +134,5 @@ void Init_nokogiri()
   init_xml_relax_ng();
   init_nokogiri_io();
   init_xml_encoding_handler();
+  init_test_global_handlers();
 }

data/ext/nokogiri/nokogiri.h

@@ -1,20 +1,25 @@
 #ifndef NOKOGIRI_NATIVE
 #define NOKOGIRI_NATIVE
 
+#if _MSC_VER
+#  ifndef WIN32_LEAN_AND_MEAN
+#    define WIN32_LEAN_AND_MEAN
+#  endif /* WIN32_LEAN_AND_MEAN */
+
+#  ifndef WIN32
+#    define WIN32
+#  endif /* WIN32 */
+
+#  include <winsock2.h>
+#  include <ws2tcpip.h>
+#  include <windows.h>
+#endif
+
 #include <stdlib.h>
 #include <string.h>
 #include <assert.h>
 #include <stdarg.h>
-
-#ifdef USE_INCLUDED_VASPRINTF
-int vasprintf (char **strp, const char *fmt, va_list ap);
-#else
-
-#define _GNU_SOURCE
-#  include <stdio.h>
-#undef _GNU_SOURCE
-
-#endif
+#include <stdio.h>
 
 #include <libxml/parser.h>
 #include <libxml/entities.h>
@@ -31,6 +36,7 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
 #include <libxslt/extensions.h>
 #include <libxslt/xsltconfig.h>
 #include <libxml/c14n.h>
+
 #include <ruby.h>
 #include <ruby/st.h>
 #include <ruby/encoding.h>
@@ -94,25 +100,27 @@ extern VALUE mNokogiriHtml ;
 extern VALUE mNokogiriHtmlSax ;
 extern VALUE mNokogiriXslt ;
 
+int vasprintf(char **strp, const char *fmt, va_list ap);
+
 void nokogiri_root_node(xmlNodePtr);
 void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
 
 #ifdef DEBUG
 
-#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
-#define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
+#  define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
+#  define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
 
 #else
 
-#define NOKOGIRI_DEBUG_START(p)
-#define NOKOGIRI_DEBUG_END(p)
+#  define NOKOGIRI_DEBUG_START(p)
+#  define NOKOGIRI_DEBUG_END(p)
 
 #endif
 
 #ifndef __builtin_expect
-# if defined(__GNUC__)
-#  define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
-# endif
+#  if defined(__GNUC__)
+#    define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
+#  endif
 #endif
 
 #define XMLNS_PREFIX "xmlns"

data/ext/nokogiri/test_global_handlers.c

@@ -0,0 +1,41 @@
+#include <nokogiri.h>
+#include "libxml/xmlerror.h"
+
+static VALUE foreign_error_handler_block = Qnil;
+
+static void foreign_error_handler(void* user_data, xmlErrorPtr c_error)
+{
+  rb_funcall(foreign_error_handler_block, rb_intern("call"), 0);
+}
+
+/*
+ * call-seq:
+ *   __foreign_error_handler { ... } -> nil
+ *
+ * Override libxml2's global error handlers to call the block. This method thus has very little
+ * value except to test that Nokogiri is properly setting error handlers elsewhere in the code. See
+ * test/helper.rb for how this is being used.
+ */
+static VALUE
+rb_foreign_error_handler(VALUE klass)
+{
+  rb_need_block();
+  foreign_error_handler_block = rb_block_proc();
+  xmlSetStructuredErrorFunc(NULL, foreign_error_handler);
+  return Qnil;
+}
+
+/*
+ *  Document-module: Nokogiri::Test
+ *
+ *  The Nokogiri::Test module should only be used for testing Nokogiri.
+ *  Do NOT use this outside of the Nokogiri test suite.
+ */
+void
+init_test_global_handlers()
+{
+  VALUE mNokogiri = rb_define_module("Nokogiri");
+  VALUE mNokogiriTest = rb_define_module_under(mNokogiri, "Test");
+
+  rb_define_singleton_method(mNokogiriTest, "__foreign_error_handler", rb_foreign_error_handler, 0);
+}

data/ext/nokogiri/xml_document.c

@@ -1,6 +1,6 @@
 #include <xml_document.h>
 
-static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
+static int dealloc_node_i2(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
 {
   switch(node->type) {
   case XML_ATTRIBUTE_NODE:
@@ -20,6 +20,11 @@ static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
   return ST_CONTINUE;
 }
 
+static int dealloc_node_i(st_data_t key, st_data_t node, st_data_t doc)
+{
+  return dealloc_node_i2((xmlNodePtr)key, (xmlNodePtr)node, (xmlDocPtr)doc);
+}
+
 static void remove_private(xmlNodePtr node)
 {
   xmlNodePtr child;
@@ -501,7 +506,7 @@ static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
  * The block must return a non-nil, non-false value if the +obj+ passed in
  * should be included in the canonicalized document.
  */
-static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
+static VALUE nokogiri_xml_document_canonicalize(int argc, VALUE* argv, VALUE self)
 {
   VALUE mode;
   VALUE incl_ns;
@@ -582,7 +587,7 @@ void init_xml_document()
   rb_define_method(klass, "encoding", encoding, 0);
   rb_define_method(klass, "encoding=", set_encoding, 1);
   rb_define_method(klass, "version", version, 0);
-  rb_define_method(klass, "canonicalize", canonicalize, -1);
+  rb_define_method(klass, "canonicalize", nokogiri_xml_document_canonicalize, -1);
   rb_define_method(klass, "dup", duplicate_document, -1);
   rb_define_method(klass, "url", url, 0);
   rb_define_method(klass, "create_entity", create_entity, -1);

data/ext/nokogiri/xml_io.c

@@ -2,12 +2,13 @@
 
 static ID id_read, id_write;
 
-VALUE read_check(VALUE *args) {
+VALUE read_check(VALUE val) {
+  VALUE *args = (VALUE *)val;
   return rb_funcall(args[0], id_read, 1, args[1]);
 }
 
-VALUE read_failed(void) {
-	return Qundef;
+VALUE read_failed(VALUE arg, VALUE exc) {
+  return Qundef;
 }
 
 int io_read_callback(void * ctx, char * buffer, int len) {
@@ -30,12 +31,13 @@ int io_read_callback(void * ctx, char * buffer, int len) {
   return (int)safe_len;
 }
 
-VALUE write_check(VALUE *args) {
+VALUE write_check(VALUE val) {
+  VALUE *args = (VALUE *)val;
   return rb_funcall(args[0], id_write, 1, args[1]);
 }
 
-VALUE write_failed(void) {
-	return Qundef;
+VALUE write_failed(VALUE arg, VALUE exc) {
+  return Qundef;
 }
 
 int io_write_callback(void * ctx, char * buffer, int len) {

data/ext/nokogiri/xml_node.c

@@ -301,7 +301,7 @@ ok:
        *  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);
+      xmlFree((xmlChar*)reparentee->ns->prefix);
       reparentee->ns->prefix = NULL;
     }
   }

data/ext/nokogiri/xml_node_set.c

@@ -319,7 +319,7 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
  *
  * Return this list as an Array
  */
-static VALUE to_array(VALUE self, VALUE rb_node)
+static VALUE to_array(VALUE self)
 {
   xmlNodeSetPtr node_set ;
   VALUE list;

data/ext/nokogiri/xml_reader.c

@@ -28,35 +28,24 @@ static int has_attributes(xmlTextReaderPtr reader)
 static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
 {
   xmlNsPtr ns;
-  static char buffer[XMLNS_BUFFER_LEN] ;
-  char *key ;
-  size_t keylen ;
+  VALUE key;
 
   if (node->type != XML_ELEMENT_NODE) return ;
 
   ns = node->nsDef;
   while (ns != NULL) {
 
-    keylen = XMLNS_PREFIX_LEN + (ns->prefix ? (strlen((const char*)ns->prefix) + 1) : 0) ;
-    if (keylen > XMLNS_BUFFER_LEN) {
-      key = (char*)malloc(keylen) ;
-    } else {
-      key = buffer ;
-    }
-
+    key = rb_enc_str_new_cstr(XMLNS_PREFIX, rb_utf8_encoding());
     if (ns->prefix) {
-      sprintf(key, "%s:%s", XMLNS_PREFIX, ns->prefix);
-    } else {
-      sprintf(key, "%s", XMLNS_PREFIX);
+      rb_str_cat_cstr(key, ":");
+      rb_str_cat_cstr(key, (const char*)ns->prefix);
     }
 
+    key = rb_str_conv_enc(key, rb_utf8_encoding(), rb_default_internal_encoding());
     rb_hash_aset(attr_hash,
-        NOKOGIRI_STR_NEW2(key),
+        key,
         (ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
     );
-    if (key != buffer) {
-      free(key);
-    }
     ns = ns->next ;
   }
 }

data/ext/nokogiri/xml_relax_ng.c

@@ -53,16 +53,24 @@ static VALUE validate_document(VALUE self, VALUE document)
  *
  * Create a new RelaxNG from the contents of +string+
  */
-static VALUE read_memory(VALUE klass, VALUE content)
+static VALUE read_memory(int argc, VALUE *argv, VALUE klass)
 {
-  xmlRelaxNGParserCtxtPtr ctx = xmlRelaxNGNewMemParserCtxt(
-      (const char *)StringValuePtr(content),
-      (int)RSTRING_LEN(content)
-  );
+  VALUE content;
+  VALUE parse_options;
+  xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
-  VALUE errors = rb_ary_new();
+  VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &content, &parse_options);
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
+  ctx = xmlRelaxNGNewMemParserCtxt((const char *)StringValuePtr(content), (int)RSTRING_LEN(content));
+
+  errors = rb_ary_new();
   xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
 
 #ifdef HAVE_XMLRELAXNGSETPARSERSTRUCTUREDERRORS
@@ -90,6 +98,7 @@ static VALUE read_memory(VALUE klass, VALUE content)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -100,18 +109,25 @@ static VALUE read_memory(VALUE klass, VALUE content)
  *
  * Create a new RelaxNG schema from the Nokogiri::XML::Document +doc+
  */
-static VALUE from_document(VALUE klass, VALUE document)
+static VALUE from_document(int argc, VALUE *argv, VALUE klass)
 {
+  VALUE document;
+  VALUE parse_options;
   xmlDocPtr doc;
   xmlRelaxNGParserCtxtPtr ctx;
   xmlRelaxNGPtr schema;
   VALUE errors;
   VALUE rb_schema;
+  int scanned_args = 0;
+
+  scanned_args = rb_scan_args(argc, argv, "11", &document, &parse_options);
 
   Data_Get_Struct(document, xmlDoc, doc);
+  doc = doc->doc; /* In case someone passes us a node. ugh. */
 
-  /* In case someone passes us a node. ugh. */
-  doc = doc->doc;
+  if (scanned_args == 1) {
+    parse_options = rb_const_get(rb_const_get(mNokogiriXml, rb_intern("ParseOptions")), rb_intern("DEFAULT_SCHEMA"));
+  }
 
   ctx = xmlRelaxNGNewDocParserCtxt(doc);
 
@@ -129,6 +145,7 @@ static VALUE from_document(VALUE klass, VALUE document)
   schema = xmlRelaxNGParse(ctx);
 
   xmlSetStructuredErrorFunc(NULL, NULL);
+  xmlRelaxNGFreeParserCtxt(ctx);
 
   if(NULL == schema) {
     xmlErrorPtr error = xmlGetLastError();
@@ -142,6 +159,7 @@ static VALUE from_document(VALUE klass, VALUE document)
 
   rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
   rb_iv_set(rb_schema, "@errors", errors);
+  rb_iv_set(rb_schema, "@parse_options", parse_options);
 
   return rb_schema;
 }
@@ -155,7 +173,7 @@ void init_xml_relax_ng()
 
   cNokogiriXmlRelaxNG = klass;
 
-  rb_define_singleton_method(klass, "read_memory", read_memory, 1);
-  rb_define_singleton_method(klass, "from_document", from_document, 1);
+  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);
 }