libxml-ruby 2.0.0-x86-mingw32

data/ext/libxml/ruby_xml_document.h

@@ -0,0 +1,17 @@
+/* $Id$ */
+
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RXML_DOCUMENT__
+#define __RXML_DOCUMENT__
+
+extern VALUE cXMLDocument;
+
+void rxml_init_document();
+
+
+#if defined(_WIN32)
+__declspec(dllexport) 
+#endif
+VALUE rxml_document_wrap(xmlDocPtr xnode);
+#endif

data/ext/libxml/ruby_xml_dtd.c

@@ -0,0 +1,257 @@
+#include "ruby_libxml.h"
+#include "ruby_xml_dtd.h"
+
+/*
+ * Document-class: LibXML::XML::Dtd
+ *
+ * The XML::Dtd class is used to prepare DTD's for validation of xml
+ * documents.
+ *
+ * DTDs can be created from a string or a pair of public and system identifiers.
+ * Once a Dtd object is instantiated, an XML document can be validated by the
+ * XML::Document#validate method providing the XML::Dtd object as parameeter.
+ * The method will raise an exception if the document is
+ * not valid.
+ *
+ * Basic usage:
+ *
+ *  # parse DTD
+ *  dtd = XML::Dtd.new(<<EOF)
+ *  <!ELEMENT root (item*) >
+ *  <!ELEMENT item (#PCDATA) >
+ *  EOF
+ *
+ *  # parse xml document to be validated
+ *  instance = XML::Document.file('instance.xml')
+ *
+ *  # validate
+ *  instance.validate(dtd)
+ */
+
+VALUE cXMLDtd;
+
+void rxml_dtd_free(xmlDtdPtr xdtd)
+{
+  /* Set _private to NULL so that we won't reuse the
+   same, freed, Ruby wrapper object later.*/
+  xdtd->_private = NULL;
+
+  if (xdtd->doc == NULL && xdtd->parent == NULL)
+    xmlFreeDtd(xdtd);
+}
+
+void rxml_dtd_mark(xmlDtdPtr xdtd)
+{
+  if (xdtd == NULL)
+    return;
+
+  if (xdtd->_private == NULL)
+  {
+    rb_warning("XmlNode is not bound! (%s:%d)", __FILE__, __LINE__);
+    return;
+  }
+
+  rxml_node_mark((xmlNodePtr) xdtd);
+}
+
+
+static VALUE rxml_dtd_alloc(VALUE klass)
+{
+  return Data_Wrap_Struct(klass, rxml_dtd_mark, rxml_dtd_free, NULL);
+}
+
+VALUE rxml_dtd_wrap(xmlDtdPtr xdtd)
+{
+  VALUE result;
+
+  // This node is already wrapped
+  if (xdtd->_private != NULL)
+    return (VALUE) xdtd->_private;
+
+  result = Data_Wrap_Struct(cXMLDtd, NULL, NULL, xdtd);
+
+  xdtd->_private = (void*) result;
+
+  return result;
+}
+
+/*
+ * call-seq:
+ *    dtd.external_id -> "string"
+ *
+ * Obtain this dtd's external identifer (for a PUBLIC DTD).
+ */
+static VALUE rxml_dtd_external_id_get(VALUE self)
+{
+  xmlDtdPtr xdtd;
+  Data_Get_Struct(self, xmlDtd, xdtd);
+
+
+  if (xdtd->ExternalID == NULL)
+    return (Qnil);
+  else
+    return (rxml_str_new2((const char*) xdtd->ExternalID, xdtd->doc ? xdtd->doc->encoding : NULL));
+}
+
+/*
+ * call-seq:
+ *    dtd.name -> "string"
+ *
+ * Obtain this dtd's name.
+ */
+static VALUE rxml_dtd_name_get(VALUE self)
+{
+  xmlDtdPtr xdtd;
+  Data_Get_Struct(self, xmlDtd, xdtd);
+
+
+  if (xdtd->name == NULL)
+    return (Qnil);
+  else
+    return (rxml_str_new2((const char*) xdtd->name, xdtd->doc ? xdtd->doc->encoding : NULL));
+}
+
+
+/*
+ * call-seq:
+ *    dtd.uri -> "string"
+ *
+ * Obtain this dtd's URI (for a SYSTEM or PUBLIC DTD).
+ */
+static VALUE rxml_dtd_uri_get(VALUE self)
+{
+  xmlDtdPtr xdtd;
+  Data_Get_Struct(self, xmlDtd, xdtd);
+
+
+  if (xdtd->SystemID == NULL)
+    return (Qnil);
+  else
+    return (rxml_str_new2((const char*) xdtd->SystemID, xdtd->doc ? xdtd->doc->encoding : NULL));
+}
+
+/*
+ * call-seq:
+ *    XML::Dtd.new("DTD string") -> dtd
+ *    XML::Dtd.new("public", "system") -> dtd
+ *    XML::Dtd.new("name", "public", "system", document) -> external subset dtd
+ *    XML::Dtd.new("name", "public", "system", document, false) -> internal subset dtd
+ *    XML::Dtd.new("name", "public", "system", document, true) -> internal subset dtd
+ *
+ * Create a new Dtd from the specified public and system
+ * identifiers.
+ */
+static VALUE rxml_dtd_initialize(int argc, VALUE *argv, VALUE self)
+{
+  VALUE external, system, dtd_string;
+  xmlParserInputBufferPtr buffer;
+  xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
+  xmlChar *new_string;
+  xmlDtdPtr xdtd;
+
+  // 1 argument -- string                            --> parsujeme jako dtd
+  // 2 arguments -- public, system                   --> bude se hledat
+  // 3 arguments -- public, system, name             --> creates an external subset (any parameter may be nil)
+  // 4 arguments -- public, system, name, doc        --> creates an external subset (any parameter may be nil)
+  // 5 arguments -- public, system, name, doc, true  --> creates an internal subset (all but last parameter may be nil)
+  switch (argc)
+  {
+  case 3:
+  case 4:
+  case 5: {
+      VALUE name, doc, internal;
+      const xmlChar *xname = NULL, *xpublic = NULL, *xsystem = NULL;
+      xmlDocPtr xdoc = NULL;
+
+      rb_scan_args(argc, argv, "32", &external, &system, &name, &doc, &internal);
+
+      if (external != Qnil) {
+        Check_Type(external, T_STRING);
+        xpublic = (const xmlChar*) StringValuePtr(external);
+      }
+      if (system != Qnil) {
+        Check_Type(system, T_STRING);
+        xsystem = (const xmlChar*) StringValuePtr(system);
+      }
+      if (name != Qnil) {
+        Check_Type(name, T_STRING);
+        xname = (const xmlChar*) StringValuePtr(name);
+      }
+      if (doc != Qnil) {
+        if (rb_obj_is_kind_of(doc, cXMLDocument) == Qfalse)
+          rb_raise(rb_eTypeError, "Must pass an XML::Document object");
+        Data_Get_Struct(doc, xmlDoc, xdoc);
+      }
+
+      if (internal == Qnil || internal == Qfalse)
+        xdtd = xmlNewDtd(xdoc, xname, xpublic, xsystem);
+      else
+        xdtd = xmlCreateIntSubset(xdoc, xname, xpublic, xsystem);
+
+      if (xdtd == NULL)
+        rxml_raise(&xmlLastError);
+
+      /* Document will free this dtd now. */
+      RDATA(self)->dfree = NULL;
+      DATA_PTR(self) = xdtd;
+
+      xmlSetTreeDoc((xmlNodePtr) xdtd, xdoc);
+    }
+    break;
+
+  case 2:
+    rb_scan_args(argc, argv, "20", &external, &system);
+
+    Check_Type(external, T_STRING);
+    Check_Type(system, T_STRING);
+
+    xdtd = xmlParseDTD((xmlChar*) StringValuePtr(external),
+        (xmlChar*) StringValuePtr(system));
+
+    if (xdtd == NULL)
+      rxml_raise(&xmlLastError);
+
+    DATA_PTR(self) = xdtd;
+
+    xmlSetTreeDoc((xmlNodePtr) xdtd, NULL);
+    break;
+
+  case 1:
+    rb_scan_args(argc, argv, "10", &dtd_string);
+    Check_Type(dtd_string, T_STRING);
+
+    /* Note that buffer is freed by xmlParserInputBufferPush*/
+    buffer = xmlAllocParserInputBuffer(enc);
+    new_string = xmlStrdup((xmlChar*) StringValuePtr(dtd_string));
+    xmlParserInputBufferPush(buffer, xmlStrlen(new_string),
+        (const char*) new_string);
+
+    xdtd = xmlIOParseDTD(NULL, buffer, enc);
+
+    if (xdtd == NULL)
+      rxml_raise(&xmlLastError);
+
+    xmlFree(new_string);
+
+    DATA_PTR(self) = xdtd;
+    break;
+
+  default:
+    rb_raise(rb_eArgError, "wrong number of arguments");
+  }
+
+  return self;
+}
+
+void rxml_init_dtd()
+{
+  cXMLDtd = rb_define_class_under(mXML, "Dtd", rb_cObject);
+  rb_define_alloc_func(cXMLDtd, rxml_dtd_alloc);
+  rb_define_method(cXMLDtd, "initialize", rxml_dtd_initialize, -1);
+  rb_define_method(cXMLDtd, "external_id", rxml_dtd_external_id_get, 0);
+  rb_define_method(cXMLDtd, "name", rxml_dtd_name_get, 0);
+  rb_define_method(cXMLDtd, "uri", rxml_dtd_uri_get, 0);
+
+  rb_define_alias(cXMLDtd, "system_id", "uri");
+}
+

data/ext/libxml/ruby_xml_dtd.h

@@ -0,0 +1,9 @@
+#ifndef __RXML_DTD__
+#define __RXML_DTD__
+
+extern VALUE cXMLDtd;
+
+void  rxml_init_dtd(void);
+VALUE rxml_dtd_wrap(xmlDtdPtr xdtd);
+
+#endif

data/ext/libxml/ruby_xml_encoding.c

@@ -0,0 +1,221 @@
+/* $Id: rxml_input.c 528 2008-11-15 23:43:48Z cfis $ */
+
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include <stdarg.h>
+#include "ruby_libxml.h"
+
+/*
+ * Document-class: LibXML::XML::Encoding
+ *
+ * The encoding class exposes the encodings that libxml
+ * supports via constants.  
+ *
+ * LibXML converts all data sources to UTF8
+ * internally before processing them.  By default, 
+ * LibXML determines a data source's encoding 
+ * using the algorithm described on its
+ * website[http://xmlsoft.org/encoding.html].  
+ *
+ * However, you may override a data source's encoding 
+ * by using the encoding constants defined in this
+ * module.
+ *
+ * Example 1:
+ *
+ *   io = File.open('some_file', 'rb')
+ *   parser = XML::Parser.io(io, :encoding => XML::Encoding::ISO_8859_1)
+ *   doc = parser.parse
+ *
+ * Example 2:
+ *
+ *   parser = XML::HTMLParser.file("some_file", :encoding => XML::Encoding::ISO_8859_1)
+ *   doc = parser.parse
+ *
+ * Example 3:
+ *
+ *   document = XML::Document.new
+ *   document.encoding = XML::Encoding::ISO_8859_1
+ *   doc << XML::Node.new
+ */
+
+VALUE mXMLEncoding;
+
+/*
+ * call-seq:
+ *    Input.s_to_encoding("UTF_8") -> XML::Encoding::UTF_8
+ *
+ * Converts an encoding string to an encoding constant
+ * defined on the XML::Encoding class.
+ */
+static VALUE rxml_encoding_from_s(VALUE klass, VALUE encoding)
+{
+  xmlCharEncoding xencoding;
+  
+  if (encoding == Qnil)
+    return Qnil;
+
+  xencoding = xmlParseCharEncoding(StringValuePtr(encoding));
+  return NUM2INT(xencoding);
+}
+
+/*
+ * call-seq:
+ *    Input.encoding_to_s(Input::ENCODING) -> "encoding"
+ *
+ * Converts an encoding constant defined on the XML::Encoding
+ * class to its text representation.
+ */
+static VALUE rxml_encoding_to_s(VALUE klass, VALUE encoding)
+{
+  const char* xencoding = xmlGetCharEncodingName(NUM2INT(encoding));
+
+  if (!xencoding)
+    return Qnil;
+  else
+    return rxml_str_new2(xencoding, xencoding);
+}
+
+#ifdef HAVE_RUBY_ENCODING_H
+VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
+{
+  ID encoding_name;
+
+  switch (xmlEncoding)
+  {
+    case XML_CHAR_ENCODING_UTF8:
+      encoding_name = rb_intern("UTF_8");
+      break;
+    case XML_CHAR_ENCODING_UTF16LE:
+      encoding_name = rb_intern("UTF_16LE");
+      break;
+    case XML_CHAR_ENCODING_UTF16BE:
+      encoding_name = rb_intern("UTF_16BE");
+      break;
+    case XML_CHAR_ENCODING_UCS4LE:
+      encoding_name = rb_intern("UCS_4LE");
+      break;
+    case XML_CHAR_ENCODING_UCS4BE:
+      encoding_name = rb_intern("UCS_4BE");
+      break;
+    case XML_CHAR_ENCODING_UCS2:
+      encoding_name = rb_intern("UCS_2");
+      break;
+    case XML_CHAR_ENCODING_8859_1:
+      encoding_name = rb_intern("ISO8859_1");
+      break;
+    case XML_CHAR_ENCODING_8859_2:
+      encoding_name = rb_intern("ISO8859_2");
+      break;
+    case XML_CHAR_ENCODING_8859_3:
+      encoding_name = rb_intern("ISO8859_3");
+      break;
+    case XML_CHAR_ENCODING_8859_4:
+      encoding_name = rb_intern("ISO8859_4");
+      break;
+    case XML_CHAR_ENCODING_8859_5:
+      encoding_name = rb_intern("ISO8859_5");
+      break;
+    case XML_CHAR_ENCODING_8859_6:
+      encoding_name = rb_intern("ISO8859_6");
+      break;
+    case XML_CHAR_ENCODING_8859_7:
+      encoding_name = rb_intern("ISO8859_7");
+      break;
+    case XML_CHAR_ENCODING_8859_8:
+      encoding_name = rb_intern("ISO8859_8");
+      break;
+    case XML_CHAR_ENCODING_8859_9:
+      encoding_name = rb_intern("ISO8859_9");
+      break;
+    case XML_CHAR_ENCODING_2022_JP:
+      encoding_name = rb_intern("ISO_2022_JP");
+      break;
+    case XML_CHAR_ENCODING_SHIFT_JIS:
+      encoding_name = rb_intern("SHIFT_JIS");
+      break;
+    case XML_CHAR_ENCODING_EUC_JP:
+      encoding_name = rb_intern("EUC_JP");
+      break;
+    case XML_CHAR_ENCODING_ASCII:
+      encoding_name = rb_intern("US-ASCII");
+      break;
+    default:
+      /* Covers XML_CHAR_ENCODING_ERROR, XML_CHAR_ENCODING_NONE, XML_CHAR_ENCODING_EBCDIC */
+      encoding_name = rb_intern("ASCII_8BIT");
+  }
+  return rb_const_get(rb_cEncoding, encoding_name);
+}
+
+/*
+ * call-seq:
+ *    Input.encoding_to_rb_encoding(Input::ENCODING) -> Encoding
+ *
+ * Converts an encoding constant defined on the XML::Encoding
+ * class to a Ruby encoding object (available on Ruby 1.9.* and higher).
+ */
+VALUE rxml_encoding_to_rb_encoding(VALUE klass, VALUE encoding)
+{
+  return rxml_xml_encoding_to_rb_encoding(klass, NUM2INT(encoding));
+}
+#endif
+
+void rxml_init_encoding(void)
+{
+  mXMLEncoding = rb_define_module_under(mXML, "Encoding");
+  rb_define_module_function(mXMLEncoding, "from_s", rxml_encoding_from_s, 1);
+  rb_define_module_function(mXMLEncoding, "to_s", rxml_encoding_to_s, 1);
+
+#ifdef HAVE_RUBY_ENCODING_H
+ // rb_define_module_function(mXMLEncoding, "to_rb_encoding", rxml_encoding_to_rb_encoding, 2);
+#endif
+
+  /* -1: No char encoding detected. */
+  rb_define_const(mXMLEncoding, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR));
+  /* 0: No char encoding detected. */
+  rb_define_const(mXMLEncoding, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE));
+  /* 1: UTF-8 */
+  rb_define_const(mXMLEncoding, "UTF_8", INT2NUM(XML_CHAR_ENCODING_UTF8));
+  /* 2: UTF-16 little endian. */
+  rb_define_const(mXMLEncoding, "UTF_16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE));
+  /* 3: UTF-16 big endian. */
+  rb_define_const(mXMLEncoding, "UTF_16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE)); 
+  /* 4: UCS-4 little endian. */
+  rb_define_const(mXMLEncoding, "UCS_4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE));
+  /* 5: UCS-4 big endian. */
+  rb_define_const(mXMLEncoding, "UCS_4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE));
+  /* 6: EBCDIC uh! */
+  rb_define_const(mXMLEncoding, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC)); 
+  /* 7: UCS-4 unusual ordering. */
+  rb_define_const(mXMLEncoding, "UCS_4_2143", INT2NUM(XML_CHAR_ENCODING_UCS4_2143));
+  /* 8: UCS-4 unusual ordering. */
+  rb_define_const(mXMLEncoding, "UCS_4_3412", INT2NUM(XML_CHAR_ENCODING_UCS4_3412));
+  /* 9: UCS-2. */
+  rb_define_const(mXMLEncoding, "UCS_2", INT2NUM(XML_CHAR_ENCODING_UCS2));
+  /* 10: ISO-8859-1 ISO Latin 1. */
+  rb_define_const(mXMLEncoding, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1));
+  /* 11: ISO-8859-2 ISO Latin 2. */
+  rb_define_const(mXMLEncoding, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2));
+  /* 12: ISO-8859-3. */
+  rb_define_const(mXMLEncoding, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3));
+  /* 13: ISO-8859-4. */
+  rb_define_const(mXMLEncoding, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4));
+  /* 14: ISO-8859-5. */
+  rb_define_const(mXMLEncoding, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5));
+  /* 15: ISO-8859-6. */
+  rb_define_const(mXMLEncoding, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6));
+  /* 16: ISO-8859-7. */
+  rb_define_const(mXMLEncoding, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7));
+  /* 17: ISO-8859-8. */
+  rb_define_const(mXMLEncoding, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8));
+  /* 18: ISO-8859-9. */
+  rb_define_const(mXMLEncoding, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9));
+  /* 19: ISO-2022-JP. */
+  rb_define_const(mXMLEncoding, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP));
+  /* 20: Shift_JIS. */
+  rb_define_const(mXMLEncoding, "SHIFT_JIS", INT2NUM(XML_CHAR_ENCODING_SHIFT_JIS));
+  /* 21: EUC-JP. */
+  rb_define_const(mXMLEncoding, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP));
+  /* 22: pure ASCII. */
+  rb_define_const(mXMLEncoding, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII));
+}