nokogiri 1.1.1-java

data/ext/nokogiri/xml_reader.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_XML_READER
+#define NOKOGIRI_XML_READER
+
+#include <native.h>
+
+void init_xml_reader();
+
+extern VALUE cNokogiriXmlReader;
+
+#endif

data/ext/nokogiri/xml_sax_parser.c

@@ -0,0 +1,201 @@
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <xml_sax_parser.h>
+
+/*
+ * call-seq:
+ *  parse_memory(data)
+ *
+ * Parse the document stored in +data+
+ */
+static VALUE parse_memory(VALUE self, VALUE data)
+{
+  xmlSAXHandlerPtr handler;
+  Data_Get_Struct(self, xmlSAXHandler, handler);
+  xmlSAXUserParseMemory(  handler,
+                          (void *)self,
+                          StringValuePtr(data),
+                          NUM2INT(rb_funcall(data, rb_intern("length"), 0))
+  );
+  return data;
+}
+
+/*
+ * call-seq:
+ *  native_parse_io(data, encoding)
+ *
+ * Parse the document accessable via +io+
+ */
+static VALUE native_parse_io(VALUE self, VALUE io, VALUE encoding)
+{
+  xmlSAXHandlerPtr handler;
+  Data_Get_Struct(self, xmlSAXHandler, handler);
+
+  xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding); 
+
+  xmlParserCtxtPtr sax_ctx = xmlCreateIOParserCtxt(
+      handler,
+      (void *)self,
+      (xmlInputReadCallback)io_read_callback,
+      (xmlInputCloseCallback)io_close_callback,
+      (void *)io,
+      enc
+  );
+  xmlParseDocument(sax_ctx);
+  xmlFreeParserCtxt(sax_ctx);
+  return io;
+}
+
+static VALUE native_parse_file(VALUE self, VALUE data)
+{
+  xmlSAXHandlerPtr handler;
+  Data_Get_Struct(self, xmlSAXHandler, handler);
+  xmlSAXUserParseFile(  handler,
+                        (void *)self,
+                        StringValuePtr(data)
+  );
+  return data;
+}
+
+static void start_document(void * ctx)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  rb_funcall(doc, rb_intern("start_document"), 0);
+}
+
+static void end_document(void * ctx)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  rb_funcall(doc, rb_intern("end_document"), 0);
+}
+
+static void start_element(void * ctx, const xmlChar *name, const xmlChar **atts)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE attributes = rb_ary_new();
+  const xmlChar * attr;
+  int i = 0;
+  if(atts) {
+    while((attr = atts[i]) != NULL) {
+      rb_funcall(attributes, rb_intern("<<"), 1, rb_str_new2((const char *)attr));
+      i++;
+    }
+  }
+
+  rb_funcall( doc,
+              rb_intern("start_element"),
+              2,
+              rb_str_new2((const char *)name),
+              attributes
+  );
+}
+
+static void end_element(void * ctx, const xmlChar *name)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  rb_funcall(doc, rb_intern("end_element"), 1, rb_str_new2((const char *)name));
+}
+
+static void characters_func(void * ctx, const xmlChar * ch, int len)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE str = rb_str_new((const char *)ch, (long)len);
+  rb_funcall(doc, rb_intern("characters"), 1, str);
+}
+
+static void comment_func(void * ctx, const xmlChar * value)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE str = rb_str_new2((const char *)value);
+  rb_funcall(doc, rb_intern("comment"), 1, str);
+}
+
+#ifndef XP_WIN
+static void warning_func(void * ctx, const char *msg, ...)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  char * message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  rb_funcall(doc, rb_intern("warning"), 1, rb_str_new2(message));
+  free(message);
+}
+#endif
+
+#ifndef XP_WIN
+static void error_func(void * ctx, const char *msg, ...)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  char * message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  rb_funcall(doc, rb_intern("error"), 1, rb_str_new2(message));
+  free(message);
+}
+#endif
+
+static void cdata_block(void * ctx, const xmlChar * value, int len)
+{
+  VALUE self = (VALUE)ctx;
+  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE string = rb_str_new((const char *)value, (long)len);
+  rb_funcall(doc, rb_intern("cdata_block"), 1, string);
+}
+
+static void deallocate(xmlSAXHandlerPtr handler)
+{
+  NOKOGIRI_DEBUG_START(handler);
+  free(handler);
+  NOKOGIRI_DEBUG_END(handler);
+}
+
+static VALUE allocate(VALUE klass)
+{
+  xmlSAXHandlerPtr handler = calloc(1, sizeof(xmlSAXHandler));
+
+  handler->startDocument = start_document;
+  handler->endDocument = end_document;
+  handler->startElement = start_element;
+  handler->endElement = end_element;
+  handler->characters = characters_func;
+  handler->comment = comment_func;
+#ifndef XP_WIN
+  /*
+   * The va*functions aren't in ming, and I don't want to deal with
+   * it right now.....
+   *
+   */
+  handler->warning = warning_func;
+  handler->error = error_func;
+#endif
+  handler->cdataBlock = cdata_block;
+
+  return Data_Wrap_Struct(klass, NULL, deallocate, handler);
+}
+
+VALUE cNokogiriXmlSaxParser ;
+void init_xml_sax_parser()
+{
+  VALUE klass = cNokogiriXmlSaxParser =
+    rb_const_get(mNokogiriXmlSax, rb_intern("Parser"));
+  rb_define_alloc_func(klass, allocate);
+  rb_define_method(klass, "parse_memory", parse_memory, 1);
+  rb_define_private_method(klass, "native_parse_file", native_parse_file, 1);
+  rb_define_private_method(klass, "native_parse_io", native_parse_io, 2);
+}

data/ext/nokogiri/xml_sax_parser.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_XML_SAX_PARSER
+#define NOKOGIRI_XML_SAX_PARSER
+
+#include <native.h>
+
+void init_xml_sax_parser();
+
+extern VALUE cNokogiriXmlSaxParser ;
+#endif
+

data/ext/nokogiri/xml_syntax_error.c

@@ -0,0 +1,199 @@
+#include <xml_syntax_error.h>
+
+static void dealloc(xmlErrorPtr ptr)
+{
+  NOKOGIRI_DEBUG_START(ptr);
+  xmlResetError(ptr);
+  xmlFree(ptr);
+  NOKOGIRI_DEBUG_END(ptr);
+}
+
+/*
+ * call-seq:
+ *  column
+ *
+ * Column number or 0 if not available
+ */
+static VALUE column(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM(error->int2);
+}
+
+/*
+ * call-seq:
+ *  int1
+ *
+ * Extra number information
+ */
+static VALUE int1(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM(error->int1);
+}
+
+/*
+ * call-seq:
+ *  str3
+ *
+ * Extra string information
+ */
+static VALUE str3(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  if(error->str3)
+    return rb_str_new2(error->str3);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  str2
+ *
+ * Extra string information
+ */
+static VALUE str2(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  if(error->str2)
+    return rb_str_new2(error->str2);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  str1
+ *
+ * Extra string information
+ */
+static VALUE str1(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  if(error->str1)
+    return rb_str_new2(error->str1);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  line
+ *
+ * Get the line number of the error
+ */
+static VALUE line(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM(error->line);
+}
+
+/*
+ * call-seq:
+ *  file
+ *
+ * Get the filename for the error
+ */
+static VALUE file(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  if(error->file)
+    return rb_str_new2(error->file);
+
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  level
+ *
+ * Get the error level
+ */
+static VALUE level(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM((short)error->level);
+}
+
+/*
+ * call-seq:
+ *  code
+ *
+ * Get the error code
+ */
+static VALUE code(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM(error->code);
+}
+
+/*
+ * call-seq:
+ *  domain
+ *
+ * Get the part of the library that raised this exception
+ */
+static VALUE domain(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return INT2NUM(error->domain);
+}
+
+/*
+ * call-seq:
+ *  message
+ *
+ * Get the human readable message.
+ */
+static VALUE message(VALUE self)
+{
+  xmlErrorPtr error;
+  Data_Get_Struct(self, xmlError, error);
+  return rb_str_new2(error->message);
+}
+
+void Nokogiri_error_handler(void * ctx, xmlErrorPtr error)
+{
+  // FIXME: I'm interneting this.  I *think* the pointer passed in here gets
+  // freed on its own, thats why I copy it.
+  // The files are *in* the computer.
+  xmlErrorPtr ptr = calloc(1, sizeof(xmlError));
+  xmlCopyError(error, ptr);
+
+  VALUE err = Data_Wrap_Struct(cNokogiriXmlSyntaxError, NULL, dealloc, ptr);
+  VALUE block = rb_funcall(mNokogiri, rb_intern("error_handler"), 0);
+  rb_funcall(block, rb_intern("call"), 1, err);
+}
+
+VALUE cNokogiriXmlSyntaxError;
+void init_xml_syntax_error()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+
+  /*
+   * The XML::SyntaxError is raised on parse errors
+   */
+  VALUE klass = rb_define_class_under(xml, "SyntaxError", rb_eSyntaxError);
+  cNokogiriXmlSyntaxError = klass;
+
+  rb_define_method(klass, "message", message, 0);
+  rb_define_method(klass, "domain", domain, 0);
+  rb_define_method(klass, "code", code, 0);
+  rb_define_method(klass, "level", level, 0);
+  rb_define_method(klass, "file", file, 0);
+  rb_define_method(klass, "line", line, 0);
+  rb_define_method(klass, "str1", str1, 0);
+  rb_define_method(klass, "str2", str2, 0);
+  rb_define_method(klass, "str3", str3, 0);
+  rb_define_method(klass, "int1", int1, 0);
+  rb_define_method(klass, "column", column, 0);
+}

data/ext/nokogiri/xml_syntax_error.h

@@ -0,0 +1,11 @@
+#ifndef NOKOGIRI_XML_SYNTAX_ERROR
+#define NOKOGIRI_XML_SYNTAX_ERROR
+
+#include <native.h>
+
+void init_xml_syntax_error();
+void Nokogiri_error_handler(void * ctx, xmlErrorPtr error);
+
+extern VALUE cNokogiriXmlSyntaxError;
+#endif
+

data/ext/nokogiri/xml_text.c

@@ -0,0 +1,40 @@
+#include <xml_text.h>
+
+/*
+ * call-seq:
+ *  new(content, document)
+ *
+ * Create a new Text element on the +document+ with +content+
+ */
+static VALUE new(VALUE klass, VALUE string, VALUE document)
+{
+  xmlDocPtr doc;
+  Data_Get_Struct(document, xmlDoc, doc);
+
+  xmlNodePtr node = xmlNewText((xmlChar *)StringValuePtr(string));
+  node->doc = doc;
+
+  VALUE rb_node = Nokogiri_wrap_xml_node(node) ;
+
+  if(rb_block_given_p()) rb_yield(rb_node);
+
+  return rb_node;
+}
+
+VALUE cNokogiriXmlText ;
+void init_xml_text()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  /* */
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+
+  /*
+   * Wraps Text nodes.
+   */
+  VALUE klass = rb_define_class_under(xml, "Text", node);
+
+  cNokogiriXmlText = klass;
+
+  rb_define_singleton_method(klass, "new", new, 2);
+}