libxml-ruby 0.9.8 → 0.9.9

data/ext/libxml/ruby_xml_error.h

@@ -7,7 +7,7 @@
 
 extern VALUE eXMLError;
 
-void ruby_init_xml_error();
+void rxml_init_error();
 VALUE rxml_error_wrap(xmlErrorPtr xerror);
 void rxml_raise(xmlErrorPtr xerror);
 

data/ext/libxml/ruby_xml_html_parser.c

@@ -1,4 +1,4 @@
-/* $Id: ruby_xml_html_parser.c 737 2009-01-23 01:23:08Z walltndr $ */
+/* $Id: ruby_xml_html_parser.c 758 2009-01-25 20:36:03Z cfis $ */
 
 /* Please see the LICENSE file for copyright and distribution information */
 
@@ -77,13 +77,7 @@ static VALUE rxml_html_parser_parse(VALUE self)
   return rxml_document_wrap(ctxt->myDoc);
 }
 
-// Rdoc needs to know
-#ifdef RDOC_NEVER_DEFINED
-mLibXML = rb_define_module("LibXML");
-mXML = rb_define_module_under(mLibXML, "XML");
-#endif
-
-void ruby_init_html_parser(void)
+void rxml_init_html_parser(void)
 {
   CONTEXT_ATTR = rb_intern("@context");
 

data/ext/libxml/ruby_xml_html_parser.h

@@ -1,4 +1,4 @@
-/* $Id: ruby_xml_html_parser.h 711 2009-01-20 07:17:15Z cfis $ */
+/* $Id: ruby_xml_html_parser.h 758 2009-01-25 20:36:03Z cfis $ */
 
 /* Please see the LICENSE file for copyright and distribution information */
 
@@ -7,6 +7,6 @@
 
 extern VALUE cXMLHtmlParser;
 
-void ruby_init_html_parser(void);
+void rxml_init_html_parser(void);
 
 #endif

data/ext/libxml/ruby_xml_html_parser_context.c

@@ -1,145 +1,175 @@
-/* $Id: ruby_xml_parser_context.c 710 2009-01-20 05:30:51Z cfis $ */
-
-/* Please see the LICENSE file for copyright and distribution information */
-
-#include "ruby_libxml.h"
-#include "ruby_xml_html_parser_context.h"
-
-/*
- * Document-class: LibXML::XML::HTMLParser::Context
- *
- * The XML::HTMLParser::Context class provides in-depth control over how
- * a document is parsed.
- */
-
-VALUE cXMLHtmlParserContext;
-
-static void rxml_html_parser_context_free(xmlParserCtxtPtr ctxt)
-{
-  xmlFreeParserCtxt(ctxt);
-}
-
-static VALUE rxml_html_parser_context_wrap(xmlParserCtxtPtr ctxt)
-{
-  return Data_Wrap_Struct(cXMLHtmlParserContext, NULL, rxml_html_parser_context_free, ctxt);
-}
-
-/* call-seq:
- *    XML::HTMLParser::Context.file(file) -> XML::HTMLParser::Context
- *
- * Creates a new parser context based on the specified file or uri.
- *
- * Parameters:
- *
- *  file - A filename or uri.
-*/
-static VALUE rxml_html_parser_context_file(VALUE klass, VALUE file)
-{
-  xmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(StringValuePtr(file), NULL);
-  return rxml_html_parser_context_wrap(ctxt);
-}
-
-/* call-seq:
- *    XML::HTMLParser::Context.io(io) -> XML::HTMLParser::Context
- *
- * Creates a new parser context based on the specified io object.
- *
- * Parameters:
- *
- *  io - A ruby IO object.
-*/
-static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
-{
-  htmlParserCtxtPtr ctxt;
-  xmlParserInputBufferPtr input;
-  xmlParserInputPtr stream;
-
-  input = xmlParserInputBufferCreateIO((xmlInputReadCallback) rxml_read_callback, NULL,
-                                     (void*)io, XML_CHAR_ENCODING_NONE);
-
-  ctxt = htmlNewParserCtxt();
-  if (!ctxt)
-  {
-    xmlFreeParserInputBuffer(input);
-    rxml_raise(&xmlLastError);
-  }
-
-  stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
-
-  if (!stream)
-  {
-    xmlFreeParserInputBuffer(input);
-    xmlFreeParserCtxt(ctxt);
-    rxml_raise(&xmlLastError);
-  }
-  inputPush(ctxt, stream);
-
-  return rxml_html_parser_context_wrap(ctxt);
-}
-
-/* call-seq:
- *    XML::HTMLParser::Context.string(string) -> XML::HTMLParser::Context
- *
- * Creates a new parser context based on the specified string.
- *
- * Parameters:
- *
- *  string - A string that contains the data to parse.
-*/
-static VALUE rxml_html_parser_context_string(VALUE klass, VALUE string)
-{
-  xmlParserCtxtPtr ctxt;
-  Check_Type(string, T_STRING);
-
-  if (RSTRING_LEN(string) == 0)
-    rb_raise(rb_eArgError, "Must specify a string with one or more characters");
-
-  ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(string),
-                                   RSTRING_LEN(string));
-  if (!ctxt)
-    rxml_raise(&xmlLastError);
-
-  htmlDefaultSAXHandlerInit();
-  if (ctxt->sax != NULL)
-    memcpy(ctxt->sax, &htmlDefaultSAXHandler, sizeof(xmlSAXHandlerV1));
-  
-  return rxml_html_parser_context_wrap(ctxt);
-}
-
-/*
- * call-seq:
- *    context.options = XML::Parser::Options::NOENT |
-                        XML::Parser::Options::NOCDATA
- *
- * Provides control over the execution of a parser.  Valid values 
- * are the constants defined on XML::Parser::Options.  Multiple
- * options can be combined by using Bitwise OR (|).
- */
-static VALUE rxml_html_parser_context_options_set(VALUE self, VALUE options)
-{
-  int result;
-  xmlParserCtxtPtr ctxt;
-  Check_Type(options, T_FIXNUM);
-
-  Data_Get_Struct(self, xmlParserCtxt, ctxt);
-  result = htmlCtxtUseOptions(ctxt, NUM2INT(options));
-
-  return self;
-}
-
-// Rdoc needs to know 
-#ifdef RDOC_NEVER_DEFINED
-  mLibXML = rb_define_module("LibXML");
-  mXML = rb_define_module_under(mLibXML, "XML");
-  cXMLHtmlParser = rb_define_class_under(mXML, "HTMLParser", rb_cObject);
-#endif
-
-void ruby_init_html_parser_context(void)
-{
-  cXMLHtmlParserContext = rb_define_class_under(cXMLHtmlParser, "Context", cXMLParserContext);
-
-  rb_define_singleton_method(cXMLHtmlParserContext, "file", rxml_html_parser_context_file, 1);
-  rb_define_singleton_method(cXMLHtmlParserContext, "io", rxml_html_parser_context_io, 1);
-  rb_define_singleton_method(cXMLHtmlParserContext, "string", rxml_html_parser_context_string, 1);
-  rb_define_method(cXMLHtmlParserContext, "options=", rxml_html_parser_context_options_set, 1);
-}
+/* $Id: ruby_xml_parser_context.c 710 2009-01-20 05:30:51Z cfis $ */
+
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include "ruby_libxml.h"
+#include "ruby_xml_html_parser_context.h"
+
+/*
+ * Document-class: LibXML::XML::HTMLParser::Context
+ *
+ * The XML::HTMLParser::Context class provides in-depth control over how
+ * a document is parsed.
+ */
+
+VALUE cXMLHtmlParserContext;
+ID IO_ATTR;
+
+static void rxml_html_parser_context_free(xmlParserCtxtPtr ctxt)
+{
+  xmlFreeParserCtxt(ctxt);
+}
+
+static VALUE rxml_html_parser_context_wrap(xmlParserCtxtPtr ctxt)
+{
+  return Data_Wrap_Struct(cXMLHtmlParserContext, NULL, rxml_html_parser_context_free, ctxt);
+}
+
+/* call-seq:
+ *    XML::HTMLParser::Context.file(file) -> XML::HTMLParser::Context
+ *
+ * Creates a new parser context based on the specified file or uri.
+ *
+ * Parameters:
+ *
+ *  file - A filename or uri.
+*/
+static VALUE rxml_html_parser_context_file(VALUE klass, VALUE file)
+{
+  xmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(StringValuePtr(file), NULL);
+  if (!ctxt)
+    rxml_raise(&xmlLastError);
+
+  return rxml_html_parser_context_wrap(ctxt);
+}
+
+/* call-seq:
+ *    XML::HTMLParser::Context.io(io) -> XML::HTMLParser::Context
+ *
+ * Creates a new parser context based on the specified io object.
+ *
+ * Parameters:
+ *
+ *  io - A ruby IO object.
+*/
+static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
+{
+  VALUE result;
+  htmlParserCtxtPtr ctxt;
+  xmlParserInputBufferPtr input;
+  xmlParserInputPtr stream;
+
+  if (NIL_P(io))
+    rb_raise(rb_eTypeError, "Must pass in an IO object");
+
+  input = xmlParserInputBufferCreateIO((xmlInputReadCallback) rxml_read_callback, NULL,
+                                     (void*)io, XML_CHAR_ENCODING_NONE);
+
+  ctxt = htmlNewParserCtxt();
+  if (!ctxt)
+  {
+    xmlFreeParserInputBuffer(input);
+    rxml_raise(&xmlLastError);
+  }
+
+  stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
+
+  if (!stream)
+  {
+    xmlFreeParserInputBuffer(input);
+    xmlFreeParserCtxt(ctxt);
+    rxml_raise(&xmlLastError);
+  }
+  inputPush(ctxt, stream);
+  result = rxml_html_parser_context_wrap(ctxt);
+
+  /* Attach io object to parser so it won't get freed.*/
+  rb_ivar_set(result, IO_ATTR, io);
+
+  return result;
+}
+
+/* call-seq:
+ *    XML::HTMLParser::Context.string(string) -> XML::HTMLParser::Context
+ *
+ * Creates a new parser context based on the specified string.
+ *
+ * Parameters:
+ *
+ *  string - A string that contains the data to parse.
+*/
+static VALUE rxml_html_parser_context_string(VALUE klass, VALUE string)
+{
+  xmlParserCtxtPtr ctxt;
+  Check_Type(string, T_STRING);
+
+  if (RSTRING_LEN(string) == 0)
+    rb_raise(rb_eArgError, "Must specify a string with one or more characters");
+
+  ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(string),
+                                   RSTRING_LEN(string));
+  if (!ctxt)
+    rxml_raise(&xmlLastError);
+
+  htmlDefaultSAXHandlerInit();
+  if (ctxt->sax != NULL)
+    memcpy(ctxt->sax, &htmlDefaultSAXHandler, sizeof(xmlSAXHandlerV1));
+  
+  return rxml_html_parser_context_wrap(ctxt);
+}
+
+/*
+ * call-seq:
+ *    context.disable_cdata = (true|false)
+ *
+ * Control whether the CDATA nodes will be created in this context.
+ */
+static VALUE rxml_html_parser_context_disable_cdata_set(VALUE self, VALUE bool)
+{
+  xmlParserCtxtPtr ctxt;
+  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+
+  if (ctxt->sax == NULL)
+    rb_raise(rb_eRuntimeError, "Sax handler is not yet set");
+
+  /* LibXML controls this internally with the default SAX handler. */ 
+  if (bool)
+    ctxt->sax->cdataBlock = NULL;
+  else
+    ctxt->sax->cdataBlock = htmlDefaultSAXHandler.cdataBlock;
+
+  return bool;
+}
+
+/*
+ * call-seq:
+ *    context.options = XML::Parser::Options::NOENT |
+                        XML::Parser::Options::NOCDATA
+ *
+ * Provides control over the execution of a parser.  Valid values 
+ * are the constants defined on XML::Parser::Options.  Multiple
+ * options can be combined by using Bitwise OR (|).
+ */
+static VALUE rxml_html_parser_context_options_set(VALUE self, VALUE options)
+{
+  int result;
+  xmlParserCtxtPtr ctxt;
+  Check_Type(options, T_FIXNUM);
+
+  Data_Get_Struct(self, xmlParserCtxt, ctxt);
+  result = htmlCtxtUseOptions(ctxt, NUM2INT(options));
+
+  return self;
+}
+
+void rxml_init_html_parser_context(void)
+{
+  IO_ATTR = ID2SYM(rb_intern("@io"));
+  cXMLHtmlParserContext = rb_define_class_under(cXMLHtmlParser, "Context", cXMLParserContext);
+
+  rb_define_singleton_method(cXMLHtmlParserContext, "file", rxml_html_parser_context_file, 1);
+  rb_define_singleton_method(cXMLHtmlParserContext, "io", rxml_html_parser_context_io, 1);
+  rb_define_singleton_method(cXMLHtmlParserContext, "string", rxml_html_parser_context_string, 1);
+  rb_define_method(cXMLHtmlParserContext, "disable_cdata=", rxml_html_parser_context_disable_cdata_set, 1);
+  rb_define_method(cXMLHtmlParserContext, "options=", rxml_html_parser_context_options_set, 1);
+}

data/ext/libxml/ruby_xml_html_parser_context.h

@@ -7,6 +7,6 @@
 
 extern VALUE cXMLHtmlParserContext;
 
-void ruby_init_html_parser_context(void);
+void rxml_init_html_parser_context(void);
 
 #endif

data/ext/libxml/ruby_xml_html_parser_options.c

@@ -4,45 +4,37 @@
 
 #include "ruby_libxml.h"
 
-
-/* Document-class: LibXML::XML::HTMLParserOptions
+/* Document-class: LibXML::XML::HTMLParser::Options
  *
- * Options that control the operation of the HTMLParser.  The easiest
- * way to set a parser's options is to use the methods
+ * Options to control the operation of the HTMLParser.  The easiest
+ * way to set a parser's options is via the methods
  * XML::HTMLParser.file, XML::HTMLParser.io or XML::HTMLParser.string.
  * For additional control, see XML::HTMLParser::Context#options=.
 */
 
 VALUE mXMLHtmlParserOptions;
 
-
-// Rdoc needs to know
-#ifdef RDOC_NEVER_DEFINED
-mLibXML = rb_define_module("LibXML");
-mXML = rb_define_module_under(mLibXML, "XML");
-cXMLHtmlParser = rb_define_class_under(mXML, "HTMLParser", rb_cObject);
-#endif
-
-void ruby_init_html_parser_options(void)
+void rxml_init_html_parser_options(void)
 {
   mXMLHtmlParserOptions = rb_define_module_under(cXMLHtmlParser, "Options");
 
+
 #if LIBXML_VERSION >= 20621
-  /* Relaxed parsing */
+  /* 1: Relax parsing. */
   rb_define_const(mXMLHtmlParserOptions, "RECOVER", INT2NUM(HTML_PARSE_RECOVER)); 
 #endif
-  /* suppress error reports */
+  /* 32: Suppress error reports. */
   rb_define_const(mXMLHtmlParserOptions, "NOERROR", INT2NUM(HTML_PARSE_NOERROR)); 
-  /* suppress warning reports */
+  /* 64: Suppress warning reports. */
   rb_define_const(mXMLHtmlParserOptions, "NOWARNING", INT2NUM(HTML_PARSE_NOWARNING));
-  /* pedantic error reporting */
+  /* 128: Enable pedantic error reporting. */
   rb_define_const(mXMLHtmlParserOptions, "PEDANTIC", INT2NUM(HTML_PARSE_PEDANTIC)); 
-  /* remove blank nodes */
+  /* 256: Remove blank nodes. */
   rb_define_const(mXMLHtmlParserOptions, "NOBLANKS", INT2NUM(HTML_PARSE_NOBLANKS)); 
 #if LIBXML_VERSION >= 20621
-  /* Forbid network access */
+  /* 2048: Forbid network access. */
   rb_define_const(mXMLHtmlParserOptions, "NONET", INT2NUM(HTML_PARSE_NONET)); 
-  /* compact small text nodes */
+  /* 65536: Compact small text nodes. */
   rb_define_const(mXMLHtmlParserOptions, "COMPACT", INT2NUM(HTML_PARSE_COMPACT));
 #endif
 }

data/ext/libxml/ruby_xml_html_parser_options.h

@@ -7,6 +7,6 @@
 
 extern VALUE mXMLHtmlParserOptions;
 
-void ruby_init_html_parser_options(void);
+void rxml_init_html_parser_options(void);
 
 #endif

data/ext/libxml/ruby_xml_input_cbg.c

@@ -3,7 +3,7 @@
 #include "ruby_libxml.h"
 #include "ruby_xml_input_cbg.h"
 
-/* Document-class: LibXML::XML::EncodingCallbacks
+/* Document-class: LibXML::XML::InputCallbacks
  *
  * Support for adding custom scheme handlers. */
 
@@ -176,13 +176,7 @@ static VALUE input_callbacks_remove_scheme(VALUE self, VALUE scheme_name)
   return Qfalse;
 }
 
-// Rdoc needs to know
-#ifdef RDOC_NEVER_DEFINED
-mLibXML = rb_define_module("LibXML");
-mXML = rb_define_module_under(mLibXML, "XML");
-#endif
-
-void ruby_init_input_callbacks(void)
+void rxml_init_input_callbacks(void)
 {
   VALUE cInputCallbacks;
   cInputCallbacks = rb_define_class_under(mXML, "InputCallbacks", rb_cObject);