libxml-ruby 4.1.2 → 5.0.0

data/ext/libxml/ruby_xml_error.c

@@ -50,7 +50,7 @@ static VALUE rxml_error_get_handler(VALUE self)
   return block;
 }
 
-VALUE rxml_error_wrap(xmlErrorPtr xerror)
+VALUE rxml_error_wrap(const xmlError *xerror)
 {
   VALUE result = Qnil;
 
@@ -97,7 +97,7 @@ VALUE rxml_error_wrap(xmlErrorPtr xerror)
 }
 
 /* Hook that receives xml error message */
-static void structuredErrorFunc(void *userData, xmlErrorPtr xerror)
+static void structuredErrorFunc(void *userData, const xmlError *xerror)
 {
   VALUE error = rxml_error_wrap(xerror);
 
@@ -158,11 +158,14 @@ static VALUE rxml_error_reset_handler(VALUE self)
   return self;
 }
 
-NORETURN(void rxml_raise(xmlErrorPtr xerror))
+void rxml_raise(const xmlError *xerror)
 {
-  /* Wrap error up as Ruby object and send it off to ruby */
-  VALUE error = rxml_error_wrap(xerror);
-  rb_exc_raise(error);
+  if (xerror)
+  {
+    /* Wrap error up as Ruby object and send it off to ruby */
+    VALUE error = rxml_error_wrap(xerror);
+    rb_exc_raise(error);
+  }
 }
 
 void rxml_init_error(void)

data/ext/libxml/ruby_xml_error.h

@@ -8,7 +8,7 @@
 extern VALUE eXMLError;
 
 void rxml_init_error(void);
-VALUE rxml_error_wrap(xmlErrorPtr xerror);
-NORETURN(void rxml_raise(xmlErrorPtr xerror));
+VALUE rxml_error_wrap(const xmlError *xerror);
+void rxml_raise(const xmlError *xerror);
 
 #endif

data/ext/libxml/ruby_xml_html_parser_context.c

@@ -146,19 +146,23 @@ static VALUE rxml_html_parser_context_wrap(htmlParserCtxtPtr ctxt)
  *
  * Parameters:
  *
- *  file - A filename or uri.
+ *  file - A filename or uri
+ *  options - A or'ed together list of LibXML::XML::HTMLParser::Options values
 */
-static VALUE rxml_html_parser_context_file(VALUE klass, VALUE file)
+static VALUE rxml_html_parser_context_file(int argc, VALUE* argv, VALUE klass)
 {
+  VALUE file, options;
+  rb_scan_args(argc, argv, "11", &file, &options);
+
   htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(StringValuePtr(file), NULL);
   if (!ctxt)
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
 
   /* This is annoying, but xmlInitParserCtxt (called indirectly above) and 
      xmlCtxtUseOptionsInternal (called below) initialize slightly different
      context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
      sets to 0 and xmlCtxtUseOptionsInternal sets to 1.  So we have to call both. */
-  htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
+  htmlCtxtUseOptions(ctxt, options == Qnil ? 0 : NUM2INT(options));
 
   return rxml_html_parser_context_wrap(ctxt);
 }
@@ -170,10 +174,14 @@ static VALUE rxml_html_parser_context_file(VALUE klass, VALUE file)
  *
  * Parameters:
  *
- *  io - A ruby IO object.
+ *  io - A ruby IO object
+ *  options - A or'ed together list of LibXML::XML::HTMLParser::Options values
 */
-static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
+static VALUE rxml_html_parser_context_io(int argc, VALUE* argv, VALUE klass)
 {
+  VALUE io, options;
+  rb_scan_args(argc, argv, "11", &io, &options);
+
   VALUE result;
   htmlParserCtxtPtr ctxt;
   xmlParserInputBufferPtr input;
@@ -189,14 +197,14 @@ static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
   if (!ctxt)
   {
     xmlFreeParserInputBuffer(input);
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
   }
 
   /* This is annoying, but xmlInitParserCtxt (called indirectly above) and 
      xmlCtxtUseOptionsInternal (called below) initialize slightly different
      context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
      sets to 0 and xmlCtxtUseOptionsInternal sets to 1.  So we have to call both. */
-  htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
+  htmlCtxtUseOptions(ctxt, options == Qnil ? 0 : NUM2INT(options));
 
   stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
 
@@ -204,7 +212,7 @@ static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
   {
     xmlFreeParserInputBuffer(input);
     xmlFreeParserCtxt(ctxt);
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
   }
   inputPush(ctxt, stream);
   result = rxml_html_parser_context_wrap(ctxt);
@@ -222,29 +230,35 @@ static VALUE rxml_html_parser_context_io(VALUE klass, VALUE io)
  *
  * Parameters:
  *
- *  string - A string that contains the data to parse.
+ *  string - A string that contains the data to parse
+ *  options - A or'ed together list of LibXML::XML::HTMLParser::Options values
 */
-static VALUE rxml_html_parser_context_string(VALUE klass, VALUE string)
+static VALUE rxml_html_parser_context_string(int argc, VALUE* argv, VALUE klass)
 {
-  htmlParserCtxtPtr ctxt;
+  VALUE string, options;
+  rb_scan_args(argc, argv, "11", &string, &options);
+
   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),
+  htmlParserCtxtPtr ctxt = xmlCreateMemoryParserCtxt(StringValuePtr(string),
                                    (int)RSTRING_LEN(string));
   if (!ctxt)
-    rxml_raise(&xmlLastError);
+    rxml_raise(xmlGetLastError());
 
   /* This is annoying, but xmlInitParserCtxt (called indirectly above) and 
      xmlCtxtUseOptionsInternal (called below) initialize slightly different
      context options, in particular XML_PARSE_NODICT which xmlInitParserCtxt
      sets to 0 and xmlCtxtUseOptionsInternal sets to 1.  So we have to call both. */
-  htmlCtxtUseOptions(ctxt, rxml_libxml_default_options());
+  htmlCtxtUseOptions(ctxt, options == Qnil ? 0 : NUM2INT(options));
 
-  if (ctxt->sax != NULL)
-    memcpy(ctxt->sax, &htmlDefaultSAXHandler, sizeof(xmlSAXHandlerV1));
+  // Setup sax handler
+  // TODO - there must be a better way? The sax handler is initialized for XML, but we want
+  // to use HTML
+  memset(ctxt->sax, 0, sizeof(xmlSAXHandler));
+  xmlSAX2InitHtmlDefaultSAXHandler(ctxt->sax);
   
   return rxml_html_parser_context_wrap(ctxt);
 }
@@ -288,7 +302,7 @@ static VALUE rxml_html_parser_context_disable_cdata_set(VALUE self, VALUE value)
   if (value)
     ctxt->sax->cdataBlock = NULL;
   else
-    ctxt->sax->cdataBlock = htmlDefaultSAXHandler.cdataBlock;
+    ctxt->sax->cdataBlock = xmlSAX2CDataBlock;
 
   return value;
 }
@@ -328,9 +342,9 @@ 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_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, "close", rxml_html_parser_context_close, 0);
   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_namespace.c

@@ -58,9 +58,6 @@ static VALUE rxml_namespace_initialize(VALUE self, VALUE node, VALUE prefix,
   xmlPrefix = NIL_P(prefix) ? NULL : (xmlChar *)StringValuePtr(prefix);
   xns = xmlNewNs(xnode, (xmlChar*) StringValuePtr(href), xmlPrefix);
 
-  if (!xns)
-    rxml_raise(&xmlLastError);
-
   DATA_PTR(self) = xns;
   return self;
 }