libxml-ruby 4.1.2 → 6.0.0

data/ext/libxml/ruby_xml_parser_context.h

@@ -1,10 +1,11 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#ifndef __RXML_PARSER_CONTEXT__
-#define __RXML_PARSER_CONTEXT__
-
-extern VALUE cXMLParserContext;
-
-void rxml_init_parser_context(void);
-
-#endif
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_PARSER_CONTEXT__
+#define __RXML_PARSER_CONTEXT__
+
+extern VALUE cXMLParserContext;
+extern const rb_data_type_t rxml_parser_context_type;
+
+void rxml_init_parser_context(void);
+
+#endif

data/ext/libxml/ruby_xml_parser_options.c

@@ -1,66 +1,74 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#include <stdarg.h>
-#include "ruby_libxml.h"
-
-/* Document-class: LibXML::XML::Parser::Options
- *
- * Options that control the operation of the HTMLParser.  The easiest
- * way to set a parser's options is to use the methods
- * XML::Parser.file, XML::Parser.io or XML::Parser.string.
- * For additional control, see XML::Parser::Context#options=.
-*/
-
-VALUE mXMLParserOptions;
-
-void rxml_init_parser_options(void)
-{
-  mXMLParserOptions = rb_define_module_under(cXMLParser, "Options");
-
-  /* recover on errors */  
-  rb_define_const(mXMLParserOptions, "RECOVER", INT2NUM(XML_PARSE_RECOVER));
-  /* substitute entities */
-  rb_define_const(mXMLParserOptions, "NOENT", INT2NUM(XML_PARSE_NOENT));
-  /* load the external subset */
-  rb_define_const(mXMLParserOptions, "DTDLOAD", INT2NUM(XML_PARSE_DTDLOAD));
-  /* default DTD attributes */
-  rb_define_const(mXMLParserOptions, "DTDATTR", INT2NUM(XML_PARSE_DTDATTR));
-  /* validate with the DTD */
-  rb_define_const(mXMLParserOptions, "DTDVALID", INT2NUM(XML_PARSE_DTDVALID));
-  /* suppress error reports */
-  rb_define_const(mXMLParserOptions, "NOERROR", INT2NUM(XML_PARSE_NOERROR));
-  /* suppress warning reports */
-  rb_define_const(mXMLParserOptions, "NOWARNING", INT2NUM(XML_PARSE_NOWARNING));
-  /* pedantic error reporting */
-  rb_define_const(mXMLParserOptions, "PEDANTIC", INT2NUM(XML_PARSE_PEDANTIC));
-  /* remove blank nodes */
-  rb_define_const(mXMLParserOptions, "NOBLANKS", INT2NUM(XML_PARSE_NOBLANKS));
-  /* use the SAX1 interface internally */
-  rb_define_const(mXMLParserOptions, "SAX1", INT2NUM(XML_PARSE_SAX1));
-  /* Implement XInclude substitition  */
-  rb_define_const(mXMLParserOptions, "XINCLUDE", INT2NUM(XML_PARSE_XINCLUDE));
-  /* Forbid network access */
-  rb_define_const(mXMLParserOptions, "NONET", INT2NUM(XML_PARSE_NONET));
-  /* Do not reuse the context dictionnary */
-  rb_define_const(mXMLParserOptions, "NODICT", INT2NUM(XML_PARSE_NODICT));
-  /* remove redundant namespaces declarations */
-  rb_define_const(mXMLParserOptions, "NSCLEAN", INT2NUM(XML_PARSE_NSCLEAN));
-  /* merge CDATA as text nodes */
-  rb_define_const(mXMLParserOptions, "NOCDATA", INT2NUM(XML_PARSE_NOCDATA));
-#if LIBXML_VERSION >= 20621
-  /* do not generate XINCLUDE START/END nodes */
-  rb_define_const(mXMLParserOptions, "NOXINCNODE", INT2NUM(XML_PARSE_NOXINCNODE));
-#endif
-#if LIBXML_VERSION >= 20700
-  /* compact small text nodes */
-  rb_define_const(mXMLParserOptions, "COMPACT", INT2NUM(XML_PARSE_COMPACT));
-  /* parse using XML-1.0 before update 5 */
-  rb_define_const(mXMLParserOptions, "PARSE_OLD10", INT2NUM(XML_PARSE_OLD10));
-  /* do not fixup XINCLUDE xml:base uris */
-  rb_define_const(mXMLParserOptions, "NOBASEFIX", INT2NUM(XML_PARSE_NOBASEFIX));
-#endif
-#if LIBXML_VERSION >= 20703
-  /* relax any hardcoded limit from the parser */
-  rb_define_const(mXMLParserOptions, "HUGE", INT2NUM(XML_PARSE_HUGE));
-#endif
-}
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include <stdarg.h>
+#include "ruby_libxml.h"
+
+/* Document-class: LibXML::XML::Parser::Options
+ *
+ * Options that control the operation of the HTMLParser.  The easiest
+ * way to set a parser's options is to use the methods
+ * XML::Parser.file, XML::Parser.io or XML::Parser.string.
+ * For additional control, see XML::Parser::Context#options=.
+*/
+
+VALUE mXMLParserOptions;
+
+void rxml_init_parser_options(void)
+{
+  mXMLParserOptions = rb_define_module_under(cXMLParser, "Options");
+
+  /* recover on errors */  
+  rb_define_const(mXMLParserOptions, "RECOVER", INT2NUM(XML_PARSE_RECOVER));
+  /* substitute entities */
+  rb_define_const(mXMLParserOptions, "NOENT", INT2NUM(XML_PARSE_NOENT));
+  /* load the external subset */
+  rb_define_const(mXMLParserOptions, "DTDLOAD", INT2NUM(XML_PARSE_DTDLOAD));
+  /* default DTD attributes */
+  rb_define_const(mXMLParserOptions, "DTDATTR", INT2NUM(XML_PARSE_DTDATTR));
+  /* validate with the DTD */
+  rb_define_const(mXMLParserOptions, "DTDVALID", INT2NUM(XML_PARSE_DTDVALID));
+  /* suppress error reports */
+  rb_define_const(mXMLParserOptions, "NOERROR", INT2NUM(XML_PARSE_NOERROR));
+  /* suppress warning reports */
+  rb_define_const(mXMLParserOptions, "NOWARNING", INT2NUM(XML_PARSE_NOWARNING));
+  /* pedantic error reporting */
+  rb_define_const(mXMLParserOptions, "PEDANTIC", INT2NUM(XML_PARSE_PEDANTIC));
+  /* remove blank nodes */
+  rb_define_const(mXMLParserOptions, "NOBLANKS", INT2NUM(XML_PARSE_NOBLANKS));
+  /* use the SAX1 interface internally */
+  rb_define_const(mXMLParserOptions, "SAX1", INT2NUM(XML_PARSE_SAX1));
+  /* Implement XInclude substitition  */
+  rb_define_const(mXMLParserOptions, "XINCLUDE", INT2NUM(XML_PARSE_XINCLUDE));
+  /* Forbid network access */
+  rb_define_const(mXMLParserOptions, "NONET", INT2NUM(XML_PARSE_NONET));
+  /* Do not reuse the context dictionnary */
+  rb_define_const(mXMLParserOptions, "NODICT", INT2NUM(XML_PARSE_NODICT));
+  /* remove redundant namespaces declarations */
+  rb_define_const(mXMLParserOptions, "NSCLEAN", INT2NUM(XML_PARSE_NSCLEAN));
+  /* merge CDATA as text nodes */
+  rb_define_const(mXMLParserOptions, "NOCDATA", INT2NUM(XML_PARSE_NOCDATA));
+#if LIBXML_VERSION >= 20621
+  /* do not generate XINCLUDE START/END nodes */
+  rb_define_const(mXMLParserOptions, "NOXINCNODE", INT2NUM(XML_PARSE_NOXINCNODE));
+#endif
+#if LIBXML_VERSION >= 20700
+  /* compact small text nodes */
+  rb_define_const(mXMLParserOptions, "COMPACT", INT2NUM(XML_PARSE_COMPACT));
+  /* parse using XML-1.0 before update 5 */
+  rb_define_const(mXMLParserOptions, "OLD10", INT2NUM(XML_PARSE_OLD10));
+  /* do not fixup XINCLUDE xml:base uris */
+  rb_define_const(mXMLParserOptions, "NOBASEFIX", INT2NUM(XML_PARSE_NOBASEFIX));
+#endif
+#if LIBXML_VERSION >= 20703
+  /* relax any hardcoded limit from the parser */
+  rb_define_const(mXMLParserOptions, "HUGE", INT2NUM(XML_PARSE_HUGE));
+#endif
+#if LIBXML_VERSION >= 21106
+  /* parse using SAX2 interface before 2.7.0 */
+  rb_define_const(mXMLParserOptions, "OLDSAX", INT2NUM(XML_PARSE_OLDSAX));
+  /* ignore internal document encoding hint */
+  rb_define_const(mXMLParserOptions, "IGNORE_ENC", INT2NUM(XML_PARSE_IGNORE_ENC));
+  /* Store big lines numbers in text PSVI field */
+  rb_define_const(mXMLParserOptions, "BIG_LINES", INT2NUM(XML_PARSE_BIG_LINES));
+#endif
+}

data/ext/libxml/ruby_xml_parser_options.h

@@ -1,10 +1,10 @@
-/* Please see the LICENSE file for copyright and distribution information */
-
-#ifndef __RXML_PARSER_OPTIONS__
-#define __RXML_PARSER_OPTIONS__
-
-extern VALUE mXMLParserOptions;
-
-void rxml_init_parser_options();
-
-#endif
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_PARSER_OPTIONS__
+#define __RXML_PARSER_OPTIONS__
+
+extern VALUE mXMLParserOptions;
+
+void rxml_init_parser_options(void);
+
+#endif

data/ext/libxml/ruby_xml_reader.c

@@ -4,9 +4,12 @@
 #include "ruby_libxml.h"
 #include "ruby_xml_reader.h"
 
+#include <errno.h>
+
 #include <libxml/xmlreader.h>
 #include <libxml/xmlschemas.h>
 
+
 /*
  * Document-class: LibXML::XML::Reader
  *
@@ -57,28 +60,40 @@ static ID ENCODING_SYMBOL;
 static ID IO_ATTR;
 static ID OPTIONS_SYMBOL;
 
-static void rxml_reader_free(xmlTextReaderPtr xreader)
+static void rxml_reader_free(void* data)
 {
+  xmlTextReaderPtr xreader = (xmlTextReaderPtr)data;
   xmlFreeTextReader(xreader);
 }
 
-static void rxml_reader_mark(xmlTextReaderPtr xreader)
+static void rxml_reader_mark(void* data)
 {
+  xmlTextReaderPtr xreader = (xmlTextReaderPtr)data;
   xmlDocPtr xdoc = xmlTextReaderCurrentDoc(xreader);
-  VALUE doc = (VALUE)xdoc->_private;
-  rb_gc_mark(doc);
+  if (xdoc)
+  {
+    VALUE doc = rxml_registry_lookup(xdoc);
+    if (!NIL_P(doc))
+      rb_gc_mark(doc);
+  }
 }
 
+static const rb_data_type_t rxml_reader_data_type = {
+  .wrap_struct_name = "LibXML::XML::Reader",
+  .function = { .dmark = rxml_reader_mark, .dfree = rxml_reader_free },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY,
+};
+
 static VALUE rxml_reader_wrap(xmlTextReaderPtr xreader)
 {
-  return Data_Wrap_Struct(cXMLReader, NULL, rxml_reader_free, xreader);
+  return TypedData_Wrap_Struct(cXMLReader, &rxml_reader_data_type, xreader);
 }
 
 
 static xmlTextReaderPtr rxml_text_reader_get(VALUE obj)
 {
   xmlTextReaderPtr xreader;
-  Data_Get_Struct(obj, xmlTextReader, xreader);
+  TypedData_Get_Struct(obj, xmlTextReader, &rxml_reader_data_type, xreader);
   return xreader;
 }
 
@@ -93,12 +108,12 @@ VALUE rxml_reader_document(VALUE klass, VALUE doc)
   xmlDocPtr xdoc;
   xmlTextReaderPtr xreader;
 
-  Data_Get_Struct(doc, xmlDoc, xdoc);
+  TypedData_Get_Struct(doc, xmlDoc, &rxml_document_data_type, xdoc);
 
   xreader = xmlReaderWalker(xdoc);
 
   if (xreader == NULL)
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
 
   return rxml_reader_wrap(xreader);
 }
@@ -122,34 +137,31 @@ VALUE rxml_reader_document(VALUE klass, VALUE doc)
  */
 static VALUE rxml_reader_file(int argc, VALUE *argv, VALUE klass)
 {
-  xmlTextReaderPtr xreader;
   VALUE path;
   VALUE options;
 
-  const char *xencoding = NULL;
-  int xoptions = 0;
-
   rb_scan_args(argc, argv, "11", &path, &options);
   Check_Type(path, T_STRING);
 
+  const char* xencoding = NULL;
+  int xoptions = 0;
+
   if (!NIL_P(options))
   {
-    VALUE encoding = Qnil;
-    VALUE parserOptions = Qnil;
-
     Check_Type(options, T_HASH);
 
-    encoding = rb_hash_aref(options, BASE_URI_SYMBOL);
+    VALUE encoding = rb_hash_aref(options, BASE_URI_SYMBOL);
     xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
 
-    parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
+    VALUE parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
     xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
   }
 
-  xreader = xmlReaderForFile(StringValueCStr(path), xencoding, xoptions);
+  xmlTextReaderPtr xreader = xmlReaderForFile(StringValueCStr(path), xencoding, xoptions);
 
+  // Unfortunately libxml2 does not set xmlLastError and just returns a null reader
   if (xreader == NULL)
-    rxml_raise(&xmlLastError);
+    rb_syserr_fail(ENOENT, StringValueCStr(path));
 
   return rxml_reader_wrap(xreader);
 }
@@ -207,7 +219,7 @@ static VALUE rxml_reader_io(int argc, VALUE *argv, VALUE klass)
                            xbaseurl, xencoding, xoptions);
 
   if (xreader == NULL)
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
 
   result = rxml_reader_wrap(xreader);
 
@@ -269,7 +281,7 @@ static VALUE rxml_reader_string(int argc, VALUE *argv, VALUE klass)
                                xbaseurl, xencoding, xoptions);
 
   if (xreader == NULL)
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
 
   return rxml_reader_wrap(xreader);
 }
@@ -472,7 +484,7 @@ static VALUE rxml_reader_read(VALUE self)
   switch(result)
   {
     case -1:
-      rxml_raise(&xmlLastError);
+      rxml_raise(xmlGetLastError());
       return Qnil;
       break;
     case 0:
@@ -607,7 +619,7 @@ static VALUE rxml_reader_relax_ng_validate(VALUE self, VALUE rng)
   xmlTextReaderPtr xreader = rxml_text_reader_get(self);
   xmlRelaxNGPtr xrelax;
   int status;
-  Data_Get_Struct(rng, xmlRelaxNG, xrelax);
+  TypedData_Get_Struct(rng, xmlRelaxNG, &rxml_relaxng_data_type, xrelax);
   
   status = xmlTextReaderRelaxNGSetSchema(xreader, xrelax);
   return (status == 0 ? Qtrue : Qfalse);
@@ -632,7 +644,7 @@ rxml_reader_schema_validate(VALUE self, VALUE xsd)
   xmlSchemaPtr xschema;
   int status;
 
-  Data_Get_Struct(xsd, xmlSchema, xschema);
+  TypedData_Get_Struct(xsd, xmlSchema, &rxml_schema_type, xschema);
   status = xmlTextReaderSetSchema(xreader, xschema);
   return (status == 0 ? Qtrue : Qfalse);
 }
@@ -998,12 +1010,12 @@ static VALUE rxml_reader_expand(VALUE self)
   }
   else
   {
-	  /* We cannot call rxml_node_wrap here because its sets up a mark function
+	  /* We cannot call rxml_node_wrap here because it sets up a mark function
 	   for the node. But according to the libxml docs (http://xmlsoft.org/html/libxml-xmlreader.html#xmlTextReaderExpand)
 	   this is only valid until the next xmlTextReaderRead call.  At that point the node is freed (from reading
-	   the libxml2 source code.  So don't set a mark or free function, because they will get called in the next
+	   the libxml2 source code).  So don't set a mark or free function, because they will get called in the next
 	   garbage collection run and cause a segfault.*/
-	return Data_Wrap_Struct(cXMLNode, NULL, NULL, xnode);
+	return TypedData_Wrap_Struct(cXMLNode, &rxml_node_unmanaged_data_type, xnode);
   }
 }
 
@@ -1028,9 +1040,6 @@ static VALUE rxml_reader_doc(VALUE self)
 
   result = rxml_document_wrap(xdoc);
 
-  // And now hook in a mark function to keep the document alive as long as the reader is valid
-  RDATA(self)->dmark = (RUBY_DATA_FUNC)rxml_reader_mark;
-
   return result;
 }
 
@@ -1048,7 +1057,7 @@ static VALUE
 rxml_reader_byte_consumed(VALUE self)
 {
   xmlTextReaderPtr xreader = rxml_text_reader_get(self);
-  return INT2NUM(xmlTextReaderByteConsumed(xreader));
+  return LONG2NUM(xmlTextReaderByteConsumed(xreader));
 }
 #endif
 

data/ext/libxml/ruby_xml_reader.h

@@ -1,14 +1,14 @@
-/* Copyright (c) 2006 Apple Computer Inc.
- * Please see the LICENSE file for copyright and distribution information. */
-
-#ifndef __RXML_READER__
-#define __RXML_READER__
-
-extern VALUE cXMLReader;
-
-void rxml_init_reader(void);
-
-/* Exported to be used by XML::Document#reader */
-VALUE rxml_reader_new_walker(VALUE self, VALUE doc);
-
-#endif /* __rxml_READER__ */
+/* Copyright (c) 2006 Apple Computer Inc.
+ * Please see the LICENSE file for copyright and distribution information. */
+
+#ifndef __RXML_READER__
+#define __RXML_READER__
+
+extern VALUE cXMLReader;
+
+void rxml_init_reader(void);
+
+/* Exported to be used by XML::Document#reader */
+VALUE rxml_reader_new_walker(VALUE self, VALUE doc);
+
+#endif /* __rxml_READER__ */

data/ext/libxml/ruby_xml_registry.c

@@ -0,0 +1,31 @@
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include "ruby_libxml.h"
+#include "ruby_xml_registry.h"
+
+#include <ruby/st.h>
+
+static st_table *rxml_registry;
+
+void rxml_init_registry(void)
+{
+  rxml_registry = st_init_numtable();
+}
+
+void rxml_registry_register(void *ptr, VALUE obj)
+{
+  st_insert(rxml_registry, (st_data_t)ptr, (st_data_t)obj);
+}
+
+void rxml_registry_unregister(void *ptr)
+{
+  st_delete(rxml_registry, (st_data_t *)&ptr, NULL);
+}
+
+VALUE rxml_registry_lookup(void *ptr)
+{
+  st_data_t val;
+  if (st_lookup(rxml_registry, (st_data_t)ptr, &val))
+    return (VALUE)val;
+  return Qnil;
+}

data/ext/libxml/ruby_xml_registry.h

@@ -0,0 +1,22 @@
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_REGISTRY__
+#define __RXML_REGISTRY__
+
+/* Pointer-to-VALUE registry for mapping C structs (xmlDocPtr, xmlNodePtr,
+   etc.) back to their Ruby wrappers.
+
+   The registry is a global st_table keyed by C pointer address.  It is NOT a
+   GC root -- entries do not prevent garbage collection.  Objects stay alive
+   through the normal mark chains (mark functions look up the registry instead
+   of reading _private).
+
+   Registered pointers MUST be unregistered before the underlying C struct is
+   freed, typically in the wrapper's dfree function. */
+
+void  rxml_init_registry(void);
+void  rxml_registry_register(void *ptr, VALUE obj);
+void  rxml_registry_unregister(void *ptr);
+VALUE rxml_registry_lookup(void *ptr);   /* Qnil on miss */
+
+#endif

data/ext/libxml/ruby_xml_relaxng.c

@@ -34,11 +34,18 @@
 
 VALUE cXMLRelaxNG;
 
-static void rxml_relaxng_free(xmlRelaxNGPtr xrelaxng)
+static void rxml_relaxng_free(void* data)
 {
+  xmlRelaxNGPtr xrelaxng = (xmlRelaxNGPtr)data;
   xmlRelaxNGFree(xrelaxng);
 }
 
+const rb_data_type_t rxml_relaxng_data_type = {
+  .wrap_struct_name = "LibXML::XML::RelaxNG",
+  .function = { .dmark = NULL, .dfree = rxml_relaxng_free },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY,
+};
+
 /*
  * call-seq:
  *    XML::Relaxng.new(relaxng_uri) -> relaxng
@@ -56,7 +63,10 @@ static VALUE rxml_relaxng_init_from_uri(VALUE class, VALUE uri)
   xrelaxng = xmlRelaxNGParse(xparser);
   xmlRelaxNGFreeParserCtxt(xparser);
 
-  return Data_Wrap_Struct(cXMLRelaxNG, NULL, rxml_relaxng_free, xrelaxng);
+  if (!xrelaxng)
+    rxml_raise(xmlGetLastError());
+
+  return TypedData_Wrap_Struct(cXMLRelaxNG, &rxml_relaxng_data_type, xrelaxng);
 }
 
 /*
@@ -71,13 +81,16 @@ static VALUE rxml_relaxng_init_from_document(VALUE class, VALUE document)
   xmlRelaxNGPtr xrelaxng;
   xmlRelaxNGParserCtxtPtr xparser;
 
-  Data_Get_Struct(document, xmlDoc, xdoc);
+  TypedData_Get_Struct(document, xmlDoc, &rxml_document_data_type, xdoc);
 
   xparser = xmlRelaxNGNewDocParserCtxt(xdoc);
   xrelaxng = xmlRelaxNGParse(xparser);
   xmlRelaxNGFreeParserCtxt(xparser);
 
-  return Data_Wrap_Struct(cXMLRelaxNG, NULL, rxml_relaxng_free, xrelaxng);
+  if (!xrelaxng)
+    rxml_raise(xmlGetLastError());
+
+  return TypedData_Wrap_Struct(cXMLRelaxNG, &rxml_relaxng_data_type, xrelaxng);
 }
 
 /*
@@ -97,7 +110,10 @@ static VALUE rxml_relaxng_init_from_string(VALUE self, VALUE relaxng_str)
   xrelaxng = xmlRelaxNGParse(xparser);
   xmlRelaxNGFreeParserCtxt(xparser);
 
-  return Data_Wrap_Struct(cXMLRelaxNG, NULL, rxml_relaxng_free, xrelaxng);
+  if (!xrelaxng)
+    rxml_raise(xmlGetLastError());
+
+  return TypedData_Wrap_Struct(cXMLRelaxNG, &rxml_relaxng_data_type, xrelaxng);
 }
 
 void rxml_init_relaxng(void)

data/ext/libxml/ruby_xml_relaxng.h

@@ -1,8 +1,9 @@
-#ifndef __RXML_RELAXNG__
-#define __RXML_RELAXNG__
-
-extern VALUE cXMLRelaxNG;
-
-void  rxml_init_relaxng(void);
-#endif
-
+#ifndef __RXML_RELAXNG__
+#define __RXML_RELAXNG__
+
+extern VALUE cXMLRelaxNG;
+extern const rb_data_type_t rxml_relaxng_data_type;
+
+void  rxml_init_relaxng(void);
+#endif
+