libxml-ruby 0.8.3 → 0.9.0

data/ext/libxml/ruby_xml_relaxng.c

@@ -1,6 +1,35 @@
 #include "ruby_libxml.h"
 #include "ruby_xml_relaxng.h"
 
+/* 
+ * Document-class: LibXML::XML::RelaxNG
+ *
+ * The XML::RelaxNG class is used to prepare RelaxNG schemas for validation 
+ * of xml documents.
+ * 
+ * Schemas can be created from XML documents, strings or URIs using the
+ * corresponding methods (new for URIs).
+ *
+ * Once a schema is prepared, an XML document can be validated by the
+ * XML::Document#validate_relaxng method providing the XML::RelaxNG object
+ * as parameter. The method will raise an exception if the document is
+ * not valid.
+ *
+ * Basic Usage:
+ *
+ *  # parse schema as xml document
+ *  relaxng_document = XML::Document.file('schema.rng')
+ *  
+ *  # prepare schema for validation
+ *  relaxng_schema = XML::RelaxNG.document(relaxng_document)
+ *
+ *  # parse xml document to be validated
+ *  instance = XML::Document.file('instance.xml')
+ *
+ *  # validate
+ *  instance.validate_relaxng(relaxng_schema)
+ */
+
 VALUE cXMLRelaxNG;
 
 // Rdoc needs to know 
@@ -53,13 +82,13 @@ ruby_xml_relaxng_init_from_uri(VALUE class, VALUE uri) {
  */
 VALUE
 ruby_xml_relaxng_init_from_document(VALUE class, VALUE document) {
-  ruby_xml_document_t *rdoc;
+  xmlDocPtr xdoc;
   ruby_xml_relaxng *relaxng;
   xmlRelaxNGParserCtxtPtr parser;
 
-  Data_Get_Struct(document, ruby_xml_document_t, rdoc);
+  Data_Get_Struct(document, xmlDoc, xdoc);
 
-  parser = xmlRelaxNGNewDocParserCtxt(rdoc->doc);
+  parser = xmlRelaxNGNewDocParserCtxt(xdoc);
   relaxng = ALLOC(ruby_xml_relaxng);
   relaxng->relaxng = xmlRelaxNGParse(parser);
   xmlRelaxNGFreeParserCtxt(parser);

data/ext/libxml/ruby_xml_sax_parser.c

@@ -1,13 +1,45 @@
-/* $Id: ruby_xml_sax_parser.c 461 2008-07-15 21:35:56Z cfis $ */
+/* $Id: ruby_xml_sax_parser.c 578 2008-11-18 09:23:26Z cfis $ */
 
 /* Please see the LICENSE file for copyright and distribution information */
 
 #include "ruby_libxml.h"
 #include "ruby_xml_sax_parser.h"
 
+/*
+ * Document-class: LibXML::XML::SaxParser
+ *
+ * XML::SaxParser provides a callback based API for parsing documents,
+ * in contrast to XML::Parser's tree based API and XML::Reader's stream
+ * based API.  
+ *
+ * Note that the XML::SaxParser API is fairly complex, not well standardized, 
+ * and does not directly support validation making entity, namespace and
+ * base processing relatively hard.
+ *
+ * To use the XML::SaxParser, register a callback class via the
+ * XML::SaxParser#callbacks=.  It is easiest to include the 
+ * XML::SaxParser::Callbacks module in your class and override 
+ * the methods as needed.
+ *
+ * Basic example:
+ *
+ *   class MyCallbacks
+ *     include XML::SaxParser::Callbacks
+ *     def on_start_element(element, attributes)
+ *       puts #Element started: #{element}"
+ *     end
+ *   end
+ *
+ *   parser = XML::SaxParser.new
+ *   parser.callbacks = MyCallbacks.new
+ *   parser.parse
+ */
+
 VALUE cXMLSaxParser;
 VALUE mXMLSaxParserCallbacks;
 
+ID INPUT_ATTR;
+
 VALUE cbidOnInternalSubset;
 VALUE cbidOnIsStandalone;
 VALUE cbidOnHasInternalSubset;
@@ -29,50 +61,53 @@ VALUE cbidOnExternalSubset;
 #include "sax_parser_callbacks.inc"
 
 void
-ruby_xml_sax_parser_free(ruby_xml_sax_parser *nodesp) {
-  /* Apparently this isn't needed: time will tell */
-  /* if (nodesp->xsh != NULL) */
-  /* xmlFreeSax_Parser(nodesp->sax_parser); */
+ruby_xml_sax_parser_free(ruby_xml_sax_parser *rxsp) {
+  xfree(rxsp);
 }
 
 void
-ruby_xml_sax_parser_mark(ruby_xml_sax_parser *nodesp) {
-  if (nodesp->callbackHandler && (nodesp->callbackHandler != Qnil)) {
-    rb_gc_mark(nodesp->callbackHandler);
+ruby_xml_sax_parser_mark(ruby_xml_sax_parser *rxsp) {
+  if (rxsp->callbackHandler != Qnil) {
+    rb_gc_mark(rxsp->callbackHandler);
   }
   
-  if (nodesp->filename && (nodesp->filename != Qnil)) {
-    rb_gc_mark(nodesp->filename);
+  if (rxsp->filename != Qnil) {
+    rb_gc_mark(rxsp->filename);
   }
 
-  if (nodesp->str && (nodesp->str != Qnil)) {
-    rb_gc_mark(nodesp->str);
+  if (rxsp->str != Qnil) {
+    rb_gc_mark(rxsp->str);
   }
 }
 
+VALUE
+ruby_xml_sax_parser_alloc(VALUE klass) {
+  ruby_xml_sax_parser *rxsp = ALLOC(ruby_xml_sax_parser);
+  rxsp->xsh = &rubySAXHandlerStruct;
+  rxsp->callbackHandler = Qnil;  
+  rxsp->xpc = NULL;
+  rxsp->filename = Qnil;
+  rxsp->str = Qnil;
+
+  return Data_Wrap_Struct(cXMLSaxParser, 
+                          ruby_xml_sax_parser_mark, ruby_xml_sax_parser_free,
+                          rxsp);
+}
+
+
 /*
  * call-seq:
- *    XML::SaxParser.new -> sax_parser
+ *    sax_parser.initialize -> sax_parser
  * 
- * Create a new XML::SaxParser instance.
+ * Initiliazes instance of parser.
  */
 VALUE
-ruby_xml_sax_parser_new(VALUE class) {
-  ruby_xml_sax_parser *nodesp;
-  
-  nodesp = ALLOC(ruby_xml_sax_parser);
-  nodesp->xsh = &rubySAXHandlerStruct;
-
-  nodesp->callbackHandler = Qnil;  
-  nodesp->xpc = NULL;
-  nodesp->filename = Qnil;
-  nodesp->str = Qnil;
-
-  return(Data_Wrap_Struct(class, ruby_xml_sax_parser_mark, 
-                              ruby_xml_sax_parser_free, nodesp));
+ruby_xml_sax_parser_initialize(VALUE self) {
+  VALUE input = rb_class_new_instance(0, NULL, cXMLInput);
+  rb_iv_set(self, "@input", input);
+  return self;
 }
 
-
 /*
  * call-seq:
  *    sax_parser.callbacks -> #<XML::SaxParser::Callbacks subclass>
@@ -81,9 +116,9 @@ ruby_xml_sax_parser_new(VALUE class) {
  */
 VALUE
 ruby_xml_sax_parser_callbacks_get(VALUE self) {
-  ruby_xml_sax_parser *nodesp;
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  return(nodesp->callbackHandler);
+  ruby_xml_sax_parser *rxsp;
+  Data_Get_Struct(self, ruby_xml_sax_parser, rxsp);
+  return(rxsp->callbackHandler);
 }
 
 
@@ -98,43 +133,12 @@ ruby_xml_sax_parser_callbacks_get(VALUE self) {
  */
 VALUE
 ruby_xml_sax_parser_callbacks_set(VALUE self, VALUE callbacks) {
-  ruby_xml_sax_parser *nodesp;
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  nodesp->callbackHandler = callbacks;
-  return(nodesp->callbackHandler);
+  ruby_xml_sax_parser *rxsp;
+  Data_Get_Struct(self, ruby_xml_sax_parser, rxsp);
+  rxsp->callbackHandler = callbacks;
+  return(rxsp->callbackHandler);
 }
 
-
-/*
- * call-seq:
- *    sax_parser.filename -> "filename"
- * 
- * Obtain the filename this parser reads from.
- */
-VALUE
-ruby_xml_sax_parser_filename_get(VALUE self) {
-  ruby_xml_sax_parser *nodesp;
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  return(nodesp->filename);
-}
-
-
-/*
- * call-seq:
- *    sax_parser.filename = "filename"
- * 
- * Set the filename this parser reads from.
- */
-VALUE
-ruby_xml_sax_parser_filename_set(VALUE self, VALUE filename) {
-  ruby_xml_sax_parser *nodesp;
-  Check_Type(filename, T_STRING);
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  nodesp->filename = filename;
-  return(nodesp->filename);
-}
-
- 
 /*
  * call-seq:
  *    parser.parse -> (true|false)
@@ -146,264 +150,39 @@ VALUE
 ruby_xml_sax_parser_parse(VALUE self) {
   char *str;
   int status = 1;
-  ruby_xml_sax_parser *nodesp;
+  ruby_xml_sax_parser *rxsp;
+  VALUE source;
+  VALUE input = rb_ivar_get(self, INPUT_ATTR);
 
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
+  Data_Get_Struct(self, ruby_xml_sax_parser, rxsp);
 
-  if (nodesp->filename != Qnil) {
-    status = xmlSAXUserParseFile(nodesp->xsh, nodesp, StringValuePtr(nodesp->filename));
-  } else if (nodesp->str != Qnil) {
-    str = StringValuePtr(nodesp->str);
-    status = //ruby_xml_document_new(cXMLDocument,
-				   xmlSAXUserParseMemory(nodesp->xsh, nodesp,
-                 str, strlen(str)); //);
+  if (rb_ivar_get(input, FILE_ATTR) != Qnil)
+  {
+    source = rb_ivar_get(input, FILE_ATTR);
+    status = xmlSAXUserParseFile(rxsp->xsh, rxsp, StringValuePtr(source));
+  }
+  else if (rb_ivar_get(input, STRING_ATTR) != Qnil)
+  {
+    source = rb_ivar_get(input, STRING_ATTR);
+    str = StringValueCStr(source);
+    status = xmlSAXUserParseMemory(rxsp->xsh, rxsp, str, strlen(str));
+  }
+  else
+  {
+    rb_raise(rb_eArgError, "You must specify a parser data source");
   }
   
-  /* XXX This should return an exception for the various error codes
-   * that can come back in status, but I'm too lazy to do that right
-   * now. */
   if (status)
-    return(Qfalse);
+  {
+    ruby_xml_raise(&xmlLastError);
+    return Qfalse;
+  }
   else
+  {
     return(Qtrue);
+  }
 }
 
-
-/*
- * call-seq:
- *    parser.string -> "xml"
- * 
- * Obtain the parser's input string.
- */
-VALUE
-ruby_xml_sax_parser_str_get(VALUE self) {
-  ruby_xml_sax_parser *nodesp;
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  return(nodesp->str);
-}
-
-
-/*
- * call-seq:
- *    parser.string = "xml"
- * 
- * Set the parser's input string.
- */
-VALUE
-ruby_xml_sax_parser_str_set(VALUE self, VALUE str) {
-  ruby_xml_sax_parser *nodesp;
-  Check_Type(str, T_STRING);
-  Data_Get_Struct(self, ruby_xml_sax_parser, nodesp);
-  nodesp->str = str;
-  return(nodesp->str);
-}
-
-
-/****** SaxCallbacks ******/
-/* These are all the same. Bloody Rdoc... */
-
-/*
- * call-seq:
- *    callbacks.on_internal_subset(name, external_id, system_id)
- *
- * Called for an internal subset event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_internal_subset(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_is_standalone()
- *
- * Called for 'is standalone' event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_is_standalone(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_has_internal_subset()
- *
- * Called for an internal subset notification event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_has_internal_subset(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_has_external_subset()
- *
- * Called for an external subset notification event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_has_external_subset(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_start_document()
- *
- * Called for a start document event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_start_document(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_end_document()
- *
- * Called for an end document event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_end_document(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_start_element(name, attributes_hash)
- * 
- * Called for an element start event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_start_element(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_end_element(name)
- *
- * Called for an element end event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_end_element(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_reference(name)
- *
- * Called for a reference event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_reference(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_characters(chars)
- *
- * Called for a characters event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_characters(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_processing_instruction(target, data)
- *
- * Called for an processing instruction event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_processing_instruction(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_comment(msg)
- *
- * Called for a comment event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_comment(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
-
-
-/*
- * call-seq:
- *    callbacks.on_parser_warning(msg)
- *
- * Called for parser warnings.
- */
-VALUE
-ruby_xml_sax_callbacks_on_parser_warning(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
- 
-/*
- * call-seq:
- *    callbacks.on_parser_error(msg)
- *
- * Called for parser errors.
- */
-VALUE
-ruby_xml_sax_callbacks_on_parser_error(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
-  
-/*
- * call-seq:
- *    callbacks.on_parser_fatal_error(msg)
- *
- * Called for fatal parser errors.
- */
-VALUE
-ruby_xml_sax_callbacks_on_parser_fatal_error(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
-/*
- * call-seq:
- *    callbacks.on_cdata_block(cdata)
- *
- * Called for a CDATA block event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_cdata_block(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
- 
-/*
- * call-seq:
- *    callbacks.on_external_subset(name, external_id, system_id)
- *
- * Called for an external subset event.
- */
-VALUE
-ruby_xml_sax_callbacks_on_external_subset(int argc, VALUE *argv, VALUE self) {
-  return Qnil;
-}
-
-
 // Rdoc needs to know 
 #ifdef RDOC_NEVER_DEFINED
   mLibXML = rb_define_module("LibXML");
@@ -412,23 +191,21 @@ ruby_xml_sax_callbacks_on_external_subset(int argc, VALUE *argv, VALUE self) {
 
 void
 ruby_init_xml_sax_parser(void) {
+  /* SaxParser */
   cXMLSaxParser = rb_define_class_under(mXML, "SaxParser", rb_cObject);
-  mXMLSaxParserCallbacks = rb_define_module_under(cXMLSaxParser, "Callbacks");
+  rb_define_alloc_func(cXMLSaxParser, ruby_xml_sax_parser_alloc);
+  rb_define_method(cXMLSaxParser, "callbacks", ruby_xml_sax_parser_callbacks_get, 0);
+  rb_define_method(cXMLSaxParser, "callbacks=", ruby_xml_sax_parser_callbacks_set, 1);
+  rb_define_method(cXMLSaxParser, "parse", ruby_xml_sax_parser_parse, 0);
 
-  /* SaxParser */
-  rb_define_singleton_method(cXMLSaxParser, "new", ruby_xml_sax_parser_new, 0);
+  /* Atributes */
+  rb_define_attr(cXMLSaxParser, "input", 1, 0);
+
+  /* Instance Methods */
+  rb_define_method(cXMLSaxParser, "initialize", ruby_xml_sax_parser_initialize, 0);
 
-  rb_define_method(cXMLSaxParser, "filename",
-       ruby_xml_sax_parser_filename_get, 0);
-  rb_define_method(cXMLSaxParser, "filename=",
-       ruby_xml_sax_parser_filename_set, 1);
-  rb_define_method(cXMLSaxParser, "callbacks",
-       ruby_xml_sax_parser_callbacks_get, 0);
-  rb_define_method(cXMLSaxParser, "callbacks=",
-       ruby_xml_sax_parser_callbacks_set, 1);
-  rb_define_method(cXMLSaxParser, "parse", ruby_xml_sax_parser_parse, 0);
-  rb_define_method(cXMLSaxParser, "string", ruby_xml_sax_parser_str_get, 0);
-  rb_define_method(cXMLSaxParser, "string=", ruby_xml_sax_parser_str_set, 1);
+
+  INPUT_ATTR = rb_intern("@input");
 
   /* SaxCallbacks */
   cbidOnInternalSubset = rb_intern("on_internal_subset");
@@ -448,39 +225,4 @@ ruby_init_xml_sax_parser(void) {
   cbidOnXmlParserFatalError = rb_intern("on_parser_fatal_error");
   cbidOnCdataBlock = rb_intern("on_cdata_block");
   cbidOnExternalSubset = rb_intern("on_external_subset");
-
-  rb_define_method(mXMLSaxParserCallbacks, "on_internal_subset", 
-      ruby_xml_sax_callbacks_on_internal_subset, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_is_standalone", 
-      ruby_xml_sax_callbacks_on_is_standalone, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_has_internal_subset", 
-      ruby_xml_sax_callbacks_on_has_internal_subset, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_has_external_subset", 
-      ruby_xml_sax_callbacks_on_has_external_subset, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_start_document", 
-      ruby_xml_sax_callbacks_on_start_document, -1);  
-  rb_define_method(mXMLSaxParserCallbacks, "on_end_document", 
-      ruby_xml_sax_callbacks_on_end_document, -1);  
-  rb_define_method(mXMLSaxParserCallbacks, "on_start_element", 
-      ruby_xml_sax_callbacks_on_start_element, -1);  
-  rb_define_method(mXMLSaxParserCallbacks, "on_end_element", 
-      ruby_xml_sax_callbacks_on_end_element, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_reference", 
-      ruby_xml_sax_callbacks_on_reference, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_characters", 
-      ruby_xml_sax_callbacks_on_characters, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_processing_instruction", 
-      ruby_xml_sax_callbacks_on_processing_instruction, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_comment", 
-      ruby_xml_sax_callbacks_on_comment, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_parser_warning", 
-      ruby_xml_sax_callbacks_on_parser_warning, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_parser_error", 
-      ruby_xml_sax_callbacks_on_parser_error, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_parser_fatal_error", 
-      ruby_xml_sax_callbacks_on_parser_fatal_error, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_cdata_block", 
-      ruby_xml_sax_callbacks_on_cdata_block, -1);        
-  rb_define_method(mXMLSaxParserCallbacks, "on_external_subset", 
-      ruby_xml_sax_callbacks_on_external_subset, -1);              
 }