blackwinter-libxslt-ruby 1.0.1

data/ext/libxslt/libxslt.c

@@ -0,0 +1,235 @@
+/* $Id: libxslt.c 42 2007-12-07 06:09:35Z transami $ */
+
+/* Please see the LICENSE file for copyright and distribution information */
+
+#include "libxslt.h"
+#include "libxml/xmlversion.h"
+
+VALUE cLibXSLT;
+VALUE cXSLT;
+VALUE eXSLTError;
+
+/*
+ * Document-class: LibXSLT::XSLT
+ *
+ * The libxslt gem provides Ruby language bindings for GNOME's Libxslt
+ * toolkit. It is free software, released under the MIT License.
+ *
+ * Using the bindings is straightforward:
+ *
+ *  stylesheet_doc = XML::Document.file('stylesheet_file')
+ *  stylesheet = XSLT::Stylesheet.new(stylesheet_doc)
+ *
+ *  xml_doc = XML::Document.file('xml_file')
+ *  result = stylesheet.apply(xml_doc)
+ *
+ */
+
+#ifdef RDOC_NEVER_DEFINED
+  cLibXSLT = rb_define_module("XSLT");
+#endif
+
+
+#if defined(_WIN32)
+__declspec(dllexport)
+#endif
+
+VALUE
+xpathObj2value(xmlXPathObjectPtr obj, xmlDocPtr doc) {
+  VALUE ret = Qnil;
+
+  if (obj == NULL) {
+    return ret;
+  }
+
+  switch (obj->type) {
+    case XPATH_NODESET:
+      ret = rb_ary_new();
+
+      if ((obj->nodesetval != NULL) && (obj->nodesetval->nodeNr != 0)) {
+        xmlBufferPtr buff = xmlBufferCreate();
+        xmlNodePtr node;
+        int i;
+
+        /* this assumes all the nodes are elements, which is a bad idea */
+        for (i = 0; i < obj->nodesetval->nodeNr; i++) {
+          node = obj->nodesetval->nodeTab[i];
+
+          if (node->type == XML_ELEMENT_NODE) {
+            xmlNodeDump(buff, doc, node, 0, 0);
+
+            rb_ary_push(ret, rb_funcall(rb_funcall(cXMLDocument, rb_intern("string"),
+              1, rb_str_new2((char *)buff->content)), rb_intern("root"), 0));
+
+            /* empty the buffer (xmlNodeDump appends rather than replaces) */
+            xmlBufferEmpty(buff);
+          }
+          else if (node->type == XML_TEXT_NODE) {
+            rb_ary_push(ret, rb_str_new2((char *)node->content));
+          }
+          else if (node->type == XML_ATTRIBUTE_NODE) {
+            /* BUG: should ensure children is not null (shouldn't) */
+            rb_ary_push(ret, rb_str_new2((char *)node->children->content));
+          }
+          else {
+            rb_raise(rb_eRuntimeError,
+              "unsupported node type in node set: %d", node->type);
+          }
+        }
+        xmlBufferFree(buff);
+      }
+      break;
+    case XPATH_BOOLEAN:
+      ret = obj->boolval ? Qtrue : Qfalse;
+      break;
+    case XPATH_NUMBER:
+      ret = rb_float_new(obj->floatval);
+      break;
+    case XPATH_STRING:
+      ret = rb_str_new2((char *)obj->stringval);
+      break;
+    /* these cases were in libxslt's python bindings, but i
+       don't know what they are, so i'll leave them alone */
+    case XPATH_XSLT_TREE:
+    case XPATH_POINT:
+    case XPATH_RANGE:
+    case XPATH_LOCATIONSET:
+    default:
+      rb_raise(rb_eTypeError,
+        "can't convert XPath object of type %d to Ruby object", obj->type);
+  }
+  xmlXPathFreeObject(obj);
+  return ret;
+}
+
+xmlXPathObjectPtr
+value2xpathObj(VALUE val) {
+  xmlXPathObjectPtr ret = NULL;
+
+  switch (TYPE(val)) {
+    case T_TRUE:
+    case T_FALSE:
+      ret = xmlXPathNewBoolean(RTEST(val));
+      break;
+    case T_FIXNUM:
+    case T_FLOAT:
+      ret = xmlXPathNewFloat(NUM2DBL(val));
+      break;
+    case T_STRING:
+      ret = xmlXPathWrapString(xmlStrdup((const xmlChar *)StringValuePtr(val)));
+      break;
+    case T_NIL:
+      ret = xmlXPathNewNodeSet(NULL);
+      break;
+    case T_ARRAY: {
+      int i, j;
+      ret = xmlXPathNewNodeSet(NULL);
+
+      for (i = RARRAY_LEN(val); i > 0; i--) {
+        xmlXPathObjectPtr obj = value2xpathObj(rb_ary_shift(val));
+        if ((obj->nodesetval != NULL) && (obj->nodesetval->nodeNr != 0)) {
+          for (j = 0; j < obj->nodesetval->nodeNr; j++) {
+            xmlXPathNodeSetAdd(ret->nodesetval, obj->nodesetval->nodeTab[j]);
+          }
+        }
+      }
+      break;
+    }
+    default:
+      rb_raise(rb_eTypeError,
+        "can't convert object of type %s to XPath object", rb_obj_classname(val));
+  }
+
+  return ret;
+}
+
+void
+xmlXPathFuncCallback(xmlXPathParserContextPtr ctxt, int nargs) {
+  VALUE ns_hash, func_hash, args[nargs], block;
+  const xmlChar *namespace, *name;
+  int i;
+
+  if (ctxt == NULL || ctxt->context == NULL) {
+    return;
+  }
+
+  namespace = ctxt->context->functionURI;
+  name = ctxt->context->function;
+
+  if (
+    (ns_hash = rb_ivar_get(cXSLT, rb_intern("@registry"))) == Qnil ||
+    (func_hash = rb_hash_aref(ns_hash, rb_str_new2((char *)namespace))) == Qnil
+  ) {
+    rb_raise(rb_eArgError, "namespace %s not registered!\n", namespace);
+  }
+
+  if ((block = rb_hash_aref(func_hash, rb_str_new2((char *)name))) == Qnil) {
+    rb_raise(rb_eArgError, "name %s not registered!\n", name);
+  }
+
+  for (i = nargs - 1; i >= 0; i--) {
+    args[i] = xpathObj2value(valuePop(ctxt), ctxt->context->doc);
+  }
+
+  valuePush(ctxt, value2xpathObj(
+    rb_funcall2(block, rb_intern("call"), nargs, args)));
+}
+
+static VALUE
+ruby_xslt_register(VALUE class, VALUE namespace, VALUE name) {
+  VALUE ns_hash, func_hash;
+
+  if (!rb_block_given_p()) {
+    rb_raise(rb_eArgError, "no block given");
+  }
+
+  if ((ns_hash = rb_ivar_get(cXSLT, rb_intern("@registry"))) == Qnil) {
+    ns_hash = rb_ivar_set(cXSLT, rb_intern("@registry"), rb_hash_new());
+  }
+
+  if ((func_hash = rb_hash_aref(ns_hash, namespace)) == Qnil) {
+    func_hash = rb_hash_aset(ns_hash, namespace, rb_hash_new());
+  }
+
+  rb_hash_aset(func_hash, name, rb_block_proc());
+
+  xsltRegisterExtModuleFunction(
+    BAD_CAST StringValuePtr(name), BAD_CAST StringValuePtr(namespace), xmlXPathFuncCallback);
+
+  return Qnil;
+}
+
+void
+Init_libxslt_ruby(void) {
+  LIBXML_TEST_VERSION;
+
+  cLibXSLT = rb_define_module("LibXSLT");
+  cXSLT = rb_define_module_under(cLibXSLT, "XSLT");
+
+  cXMLDocument = rb_const_get(rb_const_get(rb_const_get(rb_cObject,
+    rb_intern("LibXML")), rb_intern("XML")), rb_intern("Document"));
+
+  rb_define_singleton_method(cXSLT, "register", ruby_xslt_register, 2);
+
+  rb_define_const(cXSLT, "MAX_DEPTH", INT2NUM(xsltMaxDepth));
+  rb_define_const(cXSLT, "MAX_SORT", INT2NUM(XSLT_MAX_SORT));
+  rb_define_const(cXSLT, "ENGINE_VERSION", rb_str_new2(xsltEngineVersion));
+  rb_define_const(cXSLT, "LIBXSLT_VERSION", INT2NUM(xsltLibxsltVersion));
+  rb_define_const(cXSLT, "LIBXML_VERSION", INT2NUM(xsltLibxmlVersion));
+  rb_define_const(cXSLT, "XSLT_NAMESPACE", rb_str_new2((const char*)XSLT_NAMESPACE));
+  rb_define_const(cXSLT, "DEFAULT_VENDOR", rb_str_new2(XSLT_DEFAULT_VENDOR));
+  rb_define_const(cXSLT, "DEFAULT_VERSION", rb_str_new2(XSLT_DEFAULT_VERSION));
+  rb_define_const(cXSLT, "DEFAULT_URL", rb_str_new2(XSLT_DEFAULT_URL));
+  rb_define_const(cXSLT, "NAMESPACE_LIBXSLT", rb_str_new2((const char*)XSLT_LIBXSLT_NAMESPACE));
+  rb_define_const(cXSLT, "NAMESPACE_NORM_SAXON", rb_str_new2((const char*)XSLT_NORM_SAXON_NAMESPACE));
+  rb_define_const(cXSLT, "NAMESPACE_SAXON", rb_str_new2((const char*)XSLT_SAXON_NAMESPACE));
+  rb_define_const(cXSLT, "NAMESPACE_XT", rb_str_new2((const char*)XSLT_XT_NAMESPACE));
+  rb_define_const(cXSLT, "NAMESPACE_XALAN", rb_str_new2((const char*)XSLT_XALAN_NAMESPACE));
+
+  eXSLTError = rb_define_class_under(cLibXSLT, "XSLTError", rb_eRuntimeError);
+
+  ruby_init_xslt_stylesheet();
+
+  /* Now load exslt. */
+  exsltRegisterAll();
+}

data/ext/libxslt/libxslt.h

@@ -0,0 +1,32 @@
+/* $Id: libxslt.h 43 2007-12-07 12:38:59Z transami $ */
+
+/* Please see the LICENSE file for copyright and distribution information */
+
+#ifndef __RUBY_LIBXSLT_H__
+#define __RUBY_LIBXSLT_H__
+
+#include <ruby.h>
+#include <rubyio.h>
+#include <libxml/parser.h>
+#include <libxml/debugXML.h>
+#include <libxslt/extra.h>
+#include <libxslt/xslt.h>
+#include <libxslt/xsltInternals.h>
+#include <libxslt/transform.h>
+#include <libxslt/xsltutils.h>
+#include <libexslt/exslt.h>
+
+#include "ruby_xslt_stylesheet.h"
+
+#include "version.h"
+
+/*#if ((RUBY_LIBXML_VER_MAJ != RUBY_LIBXSLT_VER_MAJ) || (RUBY_LIBXML_VER_MIN != RUBY_LIBXSLT_VER_MIN))
+#error "Incompatible LibXML-Ruby headers - please install same major/micro version"
+#endif*/
+
+extern VALUE cLibXSLT;
+extern VALUE cXSLT;
+extern VALUE eXSLTError;
+extern VALUE cXMLDocument;
+
+#endif

data/ext/libxslt/ruby_xslt_stylesheet.c

@@ -0,0 +1,271 @@
+/* $Id: ruby_xslt_stylesheet.c 42 2007-12-07 06:09:35Z transami $ */
+
+/* See the LICENSE file for copyright and distribution information. */
+
+#include "libxslt.h"
+#include "ruby_xslt_stylesheet.h"
+
+/*
+ * Document-class: LibXSLT::XSLT::Stylesheet
+ *
+ * The XSLT::Stylesheet represents a XSL stylesheet that
+ * can be used to transform an XML document.  For usage information
+ * refer to XSLT::Stylesheet#apply
+ *
+ */
+
+VALUE cXSLTStylesheet;
+
+void
+ruby_xslt_stylesheet_free(xsltStylesheetPtr xstylesheet) {
+  xsltFreeStylesheet(xstylesheet);
+}
+
+static VALUE
+ruby_xslt_stylesheet_alloc(VALUE klass) {
+  return Data_Wrap_Struct(cXSLTStylesheet,
+                          NULL, ruby_xslt_stylesheet_free,
+                          NULL);
+}
+
+
+/* call-seq:
+ *    XSLT::Stylesheet.new(document) -> XSLT::Stylesheet
+ *
+ * Creates a new XSLT stylesheet based on the specified document.
+ * For memory management reasons, a copy of the specified document
+ * will be made, so its best to create a single copy of a stylesheet
+ * and use it multiple times.
+ *
+ *  stylesheet_doc = XML::Document.file('stylesheet_file')
+ *  stylesheet = XSLT::Stylesheet.new(stylesheet_doc)
+ *
+ */
+static VALUE
+ruby_xslt_stylesheet_initialize(VALUE self, VALUE document) {
+  xmlDocPtr xdoc;
+  xmlDocPtr xcopy;
+  xsltStylesheetPtr xstylesheet;
+
+  if (!rb_obj_is_kind_of(document, cXMLDocument))
+    rb_raise(rb_eTypeError, "Must pass in an XML::Document instance.");
+
+ /* NOTE!! Since the stylesheet own the specified document, the easiest
+  * thing to do from a memory standpoint is too copy it and not expose
+  * the copy to Ruby.  The other solution is expose a memory management
+  * API on the document object for taking ownership of the document
+  * and specifying when it has been freed.  Then the document class
+  * has to be updated to always check and see if the document is
+  * still valid.  That's all doable, but seems like a pain, so
+  * just copy the document for now. */
+  Data_Get_Struct(document, xmlDoc, xdoc);
+  xcopy = xmlCopyDoc(xdoc, 1);
+  xstylesheet = xsltParseStylesheetDoc(xcopy);
+  xstylesheet->_private = (void *)self;
+  DATA_PTR(self) = xstylesheet;
+
+  /* Save a reference to the document as an attribute accessable to ruby */
+  return self;
+}
+
+/* Helper method to convert Ruby params to C params */
+char **
+ruby_xslt_coerce_params(VALUE params) {
+  char** result;
+  size_t length;
+  size_t i;
+
+  length = RARRAY_LEN(params);
+  result = ALLOC_N(char *, length + 2);
+
+  for (i=0; i<length; i++) {
+    VALUE str = rb_String(RARRAY_PTR(params)[i]);
+    int strLen = RSTRING_LEN(str);
+    result[i] = ALLOC_N(char, strLen + 1);
+    memset(result[i], 0, strLen + 1);
+    strncpy(result[i], RSTRING_PTR(str), strLen);
+  }
+
+  /* Null terminate the array - need to empty elements */
+  result[i] = NULL;
+  result[i+1] = NULL;
+
+  return result;
+}
+
+
+/* call-seq:
+ *   stylesheet.apply(document, {params}) -> XML::Document
+ *
+ * Apply this stylesheet transformation to the provided document.
+ * This method may be invoked multiple times.
+ *
+ * Params:
+ * *  document - An instance of an XML::Document
+ * *  params - An optional hash table that specifies the values for xsl:param values embedded in the stylesheet.
+ *
+ * Example:
+ *
+ *  stylesheet_doc = XML::Document.file('stylesheet_file')
+ *  stylesheet = XSLT::Stylesheet.new(stylesheet_doc)
+ *
+ *  xml_doc = XML::Document.file('xml_file')
+ *  result = stylesheet.apply(xml_doc)
+ *  result = stylesheet.apply(xml_doc, {:foo => 'bar'})
+ */
+static VALUE
+ruby_xslt_stylesheet_apply(int argc, VALUE *argv, VALUE self) {
+  xmlDocPtr xdoc;
+  xsltStylesheetPtr xstylesheet;
+  xmlDocPtr result;
+  VALUE document;
+  VALUE params;
+  int i;
+
+  char** pParams;
+
+  if (argc > 2 || argc < 1)
+    rb_raise(rb_eArgError, "wrong number of arguments (need 1 or 2)");
+
+  document = argv[0];
+
+  if (!rb_obj_is_kind_of(document, cXMLDocument))
+    rb_raise(rb_eTypeError, "Must pass in an XML::Document instance.");
+
+  /* Make sure params is a flat array */
+  params = (argc == 2 ? argv[1]: Qnil);
+  params = rb_Array(params);
+  rb_funcall(params, rb_intern("flatten!"), 0);
+  pParams = ruby_xslt_coerce_params(params);
+
+  Data_Get_Struct(document, xmlDoc, xdoc);
+  Data_Get_Struct(self, xsltStylesheet, xstylesheet);
+
+  result = xsltApplyStylesheet(xstylesheet, xdoc, (const char**)pParams);
+
+  if (!result)
+    rb_raise(eXSLTError, "Transformation failed");
+
+  /* Free allocated array of *chars.  Note we don't have to
+     free the last array item since its set to NULL. */
+  for (i=0; i<RARRAY_LEN(params); i++) {
+    ruby_xfree(pParams[i]);
+  }
+  ruby_xfree(pParams);
+
+  return rxml_document_wrap(result);
+}
+
+
+/* call-seq:
+ *   sheet.debug(to = $stdout) => (true|false)
+ *
+ * Output a debug dump of this stylesheet to the specified output
+ * stream (an instance of IO, defaults to $stdout). Requires
+ * libxml/libxslt be compiled with debugging enabled. If this
+ * is not the case, a warning is triggered and the method returns
+ * false.
+ */
+/*VALUE
+ruby_xslt_stylesheet_debug(int argc, VALUE *argv, VALUE self) {
+#ifdef LIBXML_DEBUG_ENABLED
+  OpenFile *fptr;
+  VALUE io;
+  FILE *out;
+  rxml_document_t *parsed;
+  ruby_xslt_stylesheet *xss;
+
+  Data_Get_Struct(self, ruby_xslt_stylesheet, xss);
+  if (NIL_P(xss->parsed))
+    rb_raise(eXMLXSLTStylesheetRequireParsedDoc, "must have a parsed XML result");
+
+  switch (argc) {
+  case 0:
+    io = rb_stdout;
+    break;
+  case 1:
+    io = argv[0];
+    if (rb_obj_is_kind_of(io, rb_cIO) == Qfalse)
+      rb_raise(rb_eTypeError, "need an IO object");
+    break;
+  default:
+    rb_raise(rb_eArgError, "wrong number of arguments (0 or 1)");
+  }
+
+  Data_Get_Struct(xss->parsed, rxml_document_t, parsed);
+  if (parsed->doc == NULL)
+    return(Qnil);
+
+  GetOpenFile(io, fptr);
+  rb_io_check_writable(fptr);
+  out = GetWriteFile(fptr);
+  xmlDebugDumpDocument(out, parsed->doc);
+  return(Qtrue);
+#else
+  rb_warn("libxml/libxslt was compiled without debugging support.  Please recompile libxml/libxslt and their Ruby modules");
+  return(Qfalse);
+#endif
+}
+*/
+
+/* TODO should this automatically apply the sheet if not already,
+        given that we're unlikely to do much else with it? */
+
+/* call-seq:
+ *   sheet.print(to = $stdout) => number_of_bytes
+ *
+ * Output the result of the transform to the specified output
+ * stream (an IO instance, defaults to $stdout). You *must* call
+ * +apply+ before this method or an exception will be raised.
+ */
+/*VALUE
+ruby_xslt_stylesheet_print(int argc, VALUE *argv, VALUE self) {
+  OpenFile *fptr;
+  VALUE io;
+  FILE *out;
+  rxml_document_t *parsed;
+  ruby_xslt_stylesheet *xss;
+  int bytes;
+
+  Data_Get_Struct(self, ruby_xslt_stylesheet, xss);
+  if (NIL_P(xss->parsed))
+    rb_raise(eXMLXSLTStylesheetRequireParsedDoc, "must have a parsed XML result");
+
+  switch (argc) {
+  case 0:
+    io = rb_stdout;
+    break;
+  case 1:
+    io = argv[0];
+    if (rb_obj_is_kind_of(io, rb_cIO) == Qfalse)
+      rb_raise(rb_eTypeError, "need an IO object");
+    break;
+  default:
+    rb_raise(rb_eArgError, "wrong number of arguments (0 or 1)");
+  }
+
+  Data_Get_Struct(xss->parsed, rxml_document_t, parsed);
+  if (parsed->doc == NULL)
+    return(Qnil);
+
+  GetOpenFile(io, fptr);
+  rb_io_check_writable(fptr);
+  out = GetWriteFile(fptr);
+  bytes = xsltSaveResultToFile(out, parsed->doc, xss->xsp);
+
+  return(INT2NUM(bytes));
+}*/
+
+
+#ifdef RDOC_NEVER_DEFINED
+  cLibXSLT = rb_define_module("LibXSLT");
+  cXSLT = rb_define_module_under(cLibXSLT, "XSLT");
+#endif
+
+void
+ruby_init_xslt_stylesheet(void) {
+  cXSLTStylesheet = rb_define_class_under(cXSLT, "Stylesheet", rb_cObject);
+  rb_define_alloc_func(cXSLTStylesheet, ruby_xslt_stylesheet_alloc);
+  rb_define_method(cXSLTStylesheet, "initialize", ruby_xslt_stylesheet_initialize, 1);
+  rb_define_method(cXSLTStylesheet, "apply", ruby_xslt_stylesheet_apply, -1);
+}