--- parser3/src/classes/xdoc.C 2001/09/27 07:26:27 1.3 +++ parser3/src/classes/xdoc.C 2004/02/17 15:08:14 1.136 @@ -1,36 +1,35 @@ /** @file Parser: @b xdoc parser class. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) - - $Id: xdoc.C,v 1.3 2001/09/27 07:26:27 parser Exp $ + Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ -#include "classes.h" + +#include "pa_config_includes.h" + #ifdef XML +static const char * const IDENT_XDOC_C="$Date: 2004/02/17 15:08:14 $"; + +#include "gdome.h" +#include "libxml/tree.h" +#include "libxml/HTMLtree.h" +#include "libxslt/xsltInternals.h" +#include "libxslt/transform.h" +#include "libxslt/xsltutils.h" +#include "libxslt/variables.h" +#include "libxslt/imports.h" + +#include "pa_vmethod_frame.h" + +#include "pa_stylesheet_manager.h" #include "pa_request.h" #include "pa_vxdoc.h" -#include "pa_xslt_stylesheet_manager.h" -#include "pa_stylesheet_connection.h" +#include "pa_charset.h" #include "pa_vfile.h" +#include "pa_xml_exception.h" #include "xnode.h" -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - // defines #define XDOC_CLASS_NAME "xdoc" @@ -40,370 +39,1024 @@ #define XDOC_OUTPUT_METHOD_OPTION_VALUE_HTML "html" #define XDOC_OUTPUT_METHOD_OPTION_VALUE_TEXT "text" -#define XDOC_OUTPUT_ENCODING_OPTION_NAME "encoding" - -#define XDOC_OUTPUT_DEFAULT_INDENT 4 - // class -class MXdoc : public MXnode { +class MXdoc: public MXnode { public: // VStateless_class - Value *create_new_value(Pool& pool) { return new(pool) VXdoc(pool); } + Value* create_new_value(Pool&) { return new VXdoc(0, 0); } public: - MXdoc(Pool& pool); + MXdoc(); -public: // Methoded - bool used_directly() { return true; } }; -// methods - -class ParserStringXalanOutputStream: public XalanOutputStream { -public: - - explicit ParserStringXalanOutputStream(String& astring) : fstring(astring) {} +// global variable -protected: // XalanOutputStream +DECLARE_CLASS_VAR(xdoc, new MXdoc, 0); - virtual void writeData(const char *theBuffer, unsigned long theBufferLength) { - char *copy=(char *)fstring.malloc((size_t)theBufferLength); - memcpy(copy, theBuffer, (size_t)theBufferLength); - fstring.APPEND_CLEAN(copy, (size_t)theBufferLength, "xdoc", 0); - } +// helper classes - virtual void doFlush() {} +class xmlOutputBuffer_auto_ptr { +public: + explicit xmlOutputBuffer_auto_ptr(xmlOutputBuffer *_APtr = 0) + : _Owns(_APtr != 0), _Ptr(_APtr) {} + xmlOutputBuffer_auto_ptr(const xmlOutputBuffer_auto_ptr& _Y) + : _Owns(_Y._Owns), _Ptr(_Y.release()) {} + xmlOutputBuffer_auto_ptr& operator=(const xmlOutputBuffer_auto_ptr& _Y) + {if (this != &_Y) + {if (_Ptr != _Y.get()) + {if (_Owns && _Ptr) + xmlOutputBufferClose(_Ptr); + _Owns = _Y._Owns; } + else if (_Y._Owns) + _Owns = true; + _Ptr = _Y.release(); } + return (*this); } + ~xmlOutputBuffer_auto_ptr() + {if (_Owns && _Ptr) + xmlOutputBufferClose(_Ptr); } + xmlOutputBuffer& operator*() const + {return (*get()); } + xmlOutputBuffer *operator->() const + {return (get()); } + xmlOutputBuffer *get() const + {return (_Ptr); } + xmlOutputBuffer *release() const + {((xmlOutputBuffer_auto_ptr *)this)->_Owns = false; + return (_Ptr); } +private: + bool _Owns; + xmlOutputBuffer *_Ptr; +}; +class xsltTransformContext_auto_ptr { +public: + explicit xsltTransformContext_auto_ptr(xsltTransformContext *_APtr = 0) + : _Owns(_APtr != 0), _Ptr(_APtr) {} + xsltTransformContext_auto_ptr(const xsltTransformContext_auto_ptr& _Y) + : _Owns(_Y._Owns), _Ptr(_Y.release()) {} + xsltTransformContext_auto_ptr& operator=(const xsltTransformContext_auto_ptr& _Y) + {if (this != &_Y) + {if (_Ptr != _Y.get()) + {if (_Owns && _Ptr) + xsltFreeTransformContext(_Ptr); + _Owns = _Y._Owns; } + else if (_Y._Owns) + _Owns = true; + _Ptr = _Y.release(); } + return (*this); } + ~xsltTransformContext_auto_ptr() + {if (_Owns && _Ptr) + xsltFreeTransformContext(_Ptr); } + xsltTransformContext& operator*() const + {return (*get()); } + xsltTransformContext *operator->() const + {return (get()); } + xsltTransformContext *get() const + {return (_Ptr); } + xsltTransformContext *release() const + {((xsltTransformContext_auto_ptr *)this)->_Owns = false; + return (_Ptr); } private: + bool _Owns; + xsltTransformContext *_Ptr; +}; - String& fstring; - +class xsltStylesheet_auto_ptr { +public: + explicit xsltStylesheet_auto_ptr(xsltStylesheet *_APtr = 0) + : _Owns(_APtr != 0), _Ptr(_APtr) {} + xsltStylesheet_auto_ptr(const xsltStylesheet_auto_ptr& _Y) + : _Owns(_Y._Owns), _Ptr(_Y.release()) {} + xsltStylesheet_auto_ptr& operator=(const xsltStylesheet_auto_ptr& _Y) + {if (this != &_Y) + {if (_Ptr != _Y.get()) + {if (_Owns && _Ptr) + xsltFreeStylesheet(_Ptr); + _Owns = _Y._Owns; } + else if (_Y._Owns) + _Owns = true; + _Ptr = _Y.release(); } + return (*this); } + ~xsltStylesheet_auto_ptr() + {if (_Owns && _Ptr) + xsltFreeStylesheet(_Ptr); } + xsltStylesheet& operator*() const + {return (*get()); } + xsltStylesheet *operator->() const + {return (get()); } + xsltStylesheet *get() const + {return (_Ptr); } + xsltStylesheet *release() const + {((xsltStylesheet_auto_ptr *)this)->_Owns = false; + return (_Ptr); } +private: + bool _Owns; + xsltStylesheet *_Ptr; }; -static void create_optioned_listener( - const char *& content_type, const char *& charset, FormatterListener *& listener, - Pool& pool, - const String& method_name, MethodParams *params, int index, Writer& writer) { - // default encoding from pool - const String *scharset=&pool.get_charset(); - const String *method=0; - XalanDOMString xalan_encoding; +// methods - if(params->size()>index) { - Value& voptions=params->as_no_junction(index, "options must not be code"); - if(voptions.is_defined()) { - if(Hash *options=voptions.get_hash()) { - // $.method[xml|html|text] - if(Value *vmethod=static_cast(options->get(*new(pool) - String(pool, XDOC_OUTPUT_METHOD_OPTION_NAME)))) - method=&vmethod->as_string(); +static void writeNode(Request& r, GdomeNode *node, + GdomeException exc) { + if(!node || exc) + throw XmlException(0, exc); - // $.encoding[windows-1251|...] - if(Value *vencoding=static_cast(options->get(*new(pool) - String(pool, XDOC_OUTPUT_ENCODING_OPTION_NAME)))) { - scharset=&vencoding->as_string(); - } - } else - PTHROW(0, 0, - &method_name, - "options must be hash"); - } - } + // write out result + r.write_no_lang(*new VXnode(&r.charsets, node)); +} - xalan_encoding.append(charset=scharset->cstr()); - if(!method/*default='xml'*/ || *method == XDOC_OUTPUT_METHOD_OPTION_VALUE_XML) { - content_type="text/xml"; - listener=new FormatterToXML(writer, - XalanDOMString(), // version - true, // doIndent - XDOC_OUTPUT_DEFAULT_INDENT, // indent - xalan_encoding // encoding - ); - } else if(*method == XDOC_OUTPUT_METHOD_OPTION_VALUE_HTML) { - content_type="text/html"; - listener=new FormatterToHTML(writer, - xalan_encoding, // encoding - XalanDOMString(), // mediaType - XalanDOMString(), // doctypeSystem; String to be printed at the top of the document - XalanDOMString(), // doctypePublic - true, // doIndent - XDOC_OUTPUT_DEFAULT_INDENT // indent - ); - } else if(*method == XDOC_OUTPUT_METHOD_OPTION_VALUE_TEXT) { - content_type="text/plain"; - listener=new FormatterToText(writer, - xalan_encoding // encoding - ); - } else - PTHROW(0, 0, - method, - XDOC_OUTPUT_METHOD_OPTION_NAME " option is invalid; valid methods are: " - "'" XDOC_OUTPUT_METHOD_OPTION_VALUE_XML "', " - "'" XDOC_OUTPUT_METHOD_OPTION_VALUE_HTML "', " - "'" XDOC_OUTPUT_METHOD_OPTION_VALUE_TEXT "'"); - - // never reached -} - -static void _save(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXnode& vnode=*static_cast(r.self); +// Element createElement(in DOMString tagName) raises(DOMException); +static void _createElement(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& tagName=params.as_string(0, "tagName must be string"); + + GdomeException exc; + GdomeNode *node= + (GdomeNode *)gdome_doc_createElement(vdoc.get_document(), + r.transcode(tagName).use(), + &exc); + writeNode(r, node, exc); +} - // filespec - const String& file_name=params->as_string(0, "file name must not be code"); - const char *filespec=r.absolute(file_name).cstr(String::UL_FILE_SPEC); - - // node - XalanNode& node=vnode.get_node(pool, &method_name); +// Element createElementNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); +static void _createElementNS(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); - try { - XalanFileOutputStream stream(XalanDOMString(filespec, strlen(filespec))); - XalanOutputStreamPrintWriter writer(stream); - const char *content_type, *charset; - FormatterListener *formatterListener; - create_optioned_listener(content_type, charset, formatterListener, - pool, method_name, params, 1, writer); - FormatterTreeWalker treeWalker(*formatterListener); - treeWalker.traverse(&node); // Walk that node and produce the XML... - } catch(const XSLException& e) { - r._throw(&method_name, e); - } -} + // namespaceURI;localName + const String& namespaceURI=params.as_string(0, "namespaceURI must be string"); + const String& qualifiedName=params.as_string(1, "qualifiedName must be string"); -static void _string(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXnode& vnode=*static_cast(r.self); - - // node - XalanNode& node=vnode.get_node(pool, &method_name); - - try { - String parserString=*new(pool) String(pool); - ParserStringXalanOutputStream stream(parserString); - XalanOutputStreamPrintWriter writer(stream); - const char *content_type, *charset; - FormatterListener *formatterListener; - create_optioned_listener(content_type, charset, formatterListener, - pool, method_name, params, 0, writer); - FormatterTreeWalker treeWalker(*formatterListener); - treeWalker.traverse(&node); // Walk that node and produce the XML... + GdomeException exc; + GdomeNode *node= + (GdomeNode *)gdome_doc_createElementNS(vdoc.get_document(), + r.transcode(namespaceURI).use(), + r.transcode(qualifiedName).use(), + &exc); + writeNode(r, node, exc); +} - // write out result - r.write_no_lang(parserString); - } catch(const XSLException& e) { - r._throw(&method_name, e); - } +// DocumentFragment createDocumentFragment() +static void _createDocumentFragment(Request& r, MethodParams&) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + GdomeException exc; + GdomeNode *node= + (GdomeNode *)gdome_doc_createDocumentFragment( + vdoc.get_document(), + &exc); + writeNode(r, node, exc); } +// Text createTextNode(in DOMString data); +static void _createTextNode(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& data=params.as_string(0, "data must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createTextNode( + vdoc.get_document(), + r.transcode(data).use(), + &exc); + writeNode(r, node, exc); +} -static void _file(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXnode& vnode=*static_cast(r.self); - - // node - XalanNode& node=vnode.get_node(pool, &method_name); - - try { - String& parserString=*new(pool) String(pool); - ParserStringXalanOutputStream stream(parserString); - XalanOutputStreamPrintWriter writer(stream); - const char *content_type, *charset; - FormatterListener *formatterListener; - create_optioned_listener(content_type, charset, formatterListener, - pool, method_name, params, 0, writer); - FormatterTreeWalker treeWalker(*formatterListener); - treeWalker.traverse(&node); // Walk that node and produce the XML... +// Comment createComment(in DOMString data) +static void _createComment(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& data=params.as_string(0, "data must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createComment( + vdoc.get_document(), + r.transcode(data).use(), + &exc); + writeNode(r, node, exc); +} - // write out result - VFile& vfile=*new(pool) VFile(pool); - const char *cstr=parserString.cstr(); - String *scontent_type=new(pool) String(pool, content_type); - Value *vcontent_type; - if(charset) { - VHash *vhcontent_type=new(pool) VHash(pool); - vhcontent_type->hash().put(*value_name, new(pool) VString(*scontent_type)); - String *scharset=new(pool) String(pool, charset); - vhcontent_type->hash().put(*new(pool) String(pool, "charset"), new(pool) VString(*scharset)); - vcontent_type=vhcontent_type; - } else - vcontent_type=new(pool) VString(*scontent_type); - vfile.set(false/*tainted*/, cstr, strlen(cstr), 0/*file_name*/, vcontent_type); - r.write_no_lang(vfile); - } catch(const XSLException& e) { - r._throw(&method_name, e); - } +// CDATASection createCDATASection(in DOMString data) raises(DOMException); +static void _createCDATASection(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& data=params.as_string(0, "data must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createCDATASection( + vdoc.get_document(), + r.transcode(data).use(), + &exc); + writeNode(r, node, exc); } -static void _set(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXdoc& vdom=*static_cast(r.self); - - Value& vxml=params->as_junction(0, "xml must be code"); - Temp_lang temp_lang(r, String::UL_XML); - const String& xml=r.process(vxml).as_string(); - - std::istrstream stream(xml.cstr()); - const XalanParsedSource* parsedSource; - int error=vdom.transformer().parseSource(&stream, parsedSource); - - if(error) - PTHROW(0, 0, - &method_name, - vdom.transformer().getLastError()); +// ProcessingInstruction createProcessingInstruction(in DOMString target,in DOMString data) raises(DOMException); +static void _createProcessingInstruction(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& target=params.as_string(0, "data must be string"); + const String& data=params.as_string(1, "data must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createProcessingInstruction( + vdoc.get_document(), + r.transcode(target).use(), + r.transcode(data).use(), + &exc); + writeNode(r, node, exc); +} - // replace any previous parsed source - vdom.set_parsed_source(*parsedSource); +// Attr createAttribute(in DOMString name) raises(DOMException); +static void _createAttribute(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& name=params.as_string(0, "name must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createAttribute( + vdoc.get_document(), + r.transcode(name).use(), + &exc); + writeNode(r, node, exc); } -static void _load(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXdoc& vdom=*static_cast(r.self); +// Attr createAttributeNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); +static void _createAttributeNS(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + // namespaceURI;qualifiedName + const String& namespaceURI=params.as_string(0, "namespaceURI must be string"); + const String& qualifiedName=params.as_string(1, "name must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createAttributeNS( + vdoc.get_document(), + r.transcode(namespaceURI).use(), + r.transcode(qualifiedName).use(), + &exc); + writeNode(r, node, exc); +} - // filespec - const String& file_name=params->as_string(0, "file name must not be code"); - const char *filespec=r.absolute(file_name).cstr(String::UL_FILE_SPEC); - - const XalanParsedSource* parsedSource; - int error=vdom.transformer().parseSource(filespec, parsedSource); +// EntityReference createEntityReference(in DOMString name) raises(DOMException); +static void _createEntityReference(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& name=params.as_string(0, "name must be string"); + + GdomeException exc; + GdomeNode *node=(GdomeNode *)gdome_doc_createEntityReference( + vdoc.get_document(), + r.transcode(name).use(), + &exc); + writeNode(r, node, exc); +} - if(error) - PTHROW(0, 0, - &file_name, - vdom.transformer().getLastError()); +// NodeList getElementsByTagName(in DOMString name); +static void _getElementsByTagName(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + const String& name=params.as_string(0, "name must be string"); + + VHash& result=*new VHash; + GdomeException exc; + if(GdomeNodeList *nodes= + gdome_doc_getElementsByTagName( + vdoc.get_document(), + r.transcode(name).use(), + &exc)) { + gulong length=gdome_nl_length(nodes, &exc); + for(gulong i=0; i(info); - const char *attribute_cstr=aattribute.cstr(); - const char *meaning_cstr=static_cast(ameaning)->as_string().cstr(); - - transformer.setStylesheetParam( - XalanDOMString(attribute_cstr), - XalanDOMString(meaning_cstr)); -} -static void _xslt(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXdoc& vdom=*static_cast(r.self); - - // params - if(params->size()>1) { - Value& vparams=params->as_no_junction(1, "transform parameters parameter must not be code"); - if(vparams.is_defined()) - if(Hash *params=vparams.get_hash()) - params->for_each(add_xslt_param, &vdom.transformer()); - else - PTHROW(0, 0, - &method_name, - "transform parameters parameter must be hash"); - } +static void _getElementsByTagNameNS(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); - // source - const XalanParsedSource &parsed_source=vdom.get_parsed_source(pool, &method_name); + // namespaceURI;localName + const String& namespaceURI=params.as_string(0, "namespaceURI must be string"); + const String& localName=params.as_string(1, "localName must be string"); - // stylesheet - const String& stylesheet_file_name=params->as_string(0, "file name must not be code"); - const String& stylesheet_filespec=r.absolute(stylesheet_file_name); - //_asm int 3; - Stylesheet_connection& connection=XSLT_stylesheet_manager->get_connection(stylesheet_filespec); - - // target - XalanDocument* target=vdom.parser_liaison().createDocument(); - XSLTResultTarget domResultTarget(target); - - // transform - int error=vdom.transformer().transform( - parsed_source, - &connection.stylesheet(), - domResultTarget); - connection.close(); - if(error) - PTHROW(0, 0, - &stylesheet_file_name, - vdom.transformer().getLastError()); + GdomeException exc; + VHash& result=*new VHash; + if(GdomeNodeList *nodes= + gdome_doc_getElementsByTagNameNS( + vdoc.get_document(), + r.transcode(namespaceURI).use(), + r.transcode(localName).use(), + &exc)) { + gulong length=gdome_nl_length(nodes, &exc); + for(gulong i=0; i(r.self); +static void _getElementById(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); // elementId - const char *elementId=params->as_string(0, "elementID must not be code").cstr(String::UL_AS_IS); + const String& elementId=params.as_string(0, "elementID must be string"); - if(XalanElement *element= - vdoc.get_document(pool, &method_name).getElementById(XalanDOMString(elementId))) { + GdomeException exc; + if(GdomeNode *node=(GdomeNode *)gdome_doc_getElementById( + vdoc.get_document(), + r.transcode(elementId).use(), + &exc)) { // write out result - VXnode& result=*new(pool) VXnode(pool, element); - r.write_no_lang(result); + r.write_no_lang(*new VXnode(&r.charsets, node)); + } else if(exc || xmlHaveGenericErrors()) + throw XmlException(&elementId, exc); +} + +static void _importNode(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + GdomeNode *importedNode= + as_node(params, 0, "importedNode must be node"); + bool deep= + params.as_bool(1, "deep must be bool", r); + + GdomeException exc; + GdomeNode *outputNode=gdome_doc_importNode(vdoc.get_document(), + importedNode, + deep, &exc); + if(exc) + throw XmlException(0, exc); + + // write out result + r.write_no_lang(*new VXnode(&r.charsets, outputNode)); +} +/* +GdomeElement *gdome_doc_createElementNS (GdomeDocument *self, GdomeDOMString *namespaceURI, GdomeDOMString *qualifiedName, GdomeException *exc); +GdomeAttr *gdome_doc_createAttributeNS (GdomeDocument *self, GdomeDOMString *namespaceURI, GdomeDOMString *qualifiedName, GdomeException *exc); +*/ + +static void _create(Request& r, MethodParams& params) { + Charset& source_charset=r.charsets.source(); + VXdoc& vdoc=GET_SELF(r, VXdoc); + + Value& param=params[params.count()-1]; + GdomeDocument *document; + bool set_encoding=false; + if(param.get_junction()) { // {...} + Temp_lang temp_lang(r, String::L_XML); + const String& xml=r.process_to_string(param); + + const char* cstr=xml.cstr(String::L_UNSPECIFIED, 0, &r.charsets); + document=(GdomeDocument *) + gdome_xml_n_mkref((xmlNode *)xmlParseMemory( + cstr, strlen(cstr) + )); + //printf("document=0x%p\n", document); + if(!document || xmlHaveGenericErrors()) { + GdomeException exc=0; + throw XmlException(0, exc); + } + + // must be last action in if, see after if} + } else { // [name] + const String& qualifiedName=param.as_string(); + + GdomeException exc; +#if 0 + GdomeDocumentType *documentType=gdome_di_createDocumentType ( + docimpl, + r.transcode(qualifiedName), + 0/*publicId*/, + 0/*systemId*/, + &exc); + if(!documentType || exc || xmlHaveGenericErrors()) + throw Exception( + method_name, + exc); + /// +xalan createXMLDecl ? +#endif + document=gdome_di_createDocument(domimpl, + 0/*namespaceURI*/, + r.transcode(qualifiedName).use(), + 0/*doctype*/, + &exc); + if(!document || exc || xmlHaveGenericErrors()) + throw XmlException(0, exc); + + set_encoding=true; + // must be last action in if, see after if} + } + // must be first action after if} + // replace any previous parsed source + { + vdoc.set_document(&r.charsets, document); + GdomeException exc; + gdome_doc_unref(document, &exc); + } + + // URI + const char* URI_cstr; + const char* URI_cstr_ptr; + if(params.count()>1) { // absolute(param) + const String& URI=params.as_string(0, "URI must be string"); + URI_cstr=URI_cstr_ptr=r.absolute(URI).cstr(); + } else // default = disk path to requested document + URI_cstr=r.request_info.path_translated; + xmlDoc *doc=gdome_xml_doc_get_xmlDoc(document); + if(URI_cstr) + doc->URL=source_charset.transcode_buf2xchar(URI_cstr, strlen(URI_cstr)); + + if(set_encoding) { + const char* source_charset_name=source_charset.NAME().cstr(); + doc->encoding=source_charset.transcode_buf2xchar(source_charset_name, strlen(source_charset_name)); } } -static void _getElementsByTagName(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXdoc& vdoc=*static_cast(r.self); - - // tagname - const char *tagname=params->as_string(0, "tagname must not be code").cstr(String::UL_AS_IS); - - VHash& result=*new(pool) VHash(pool); - if(const XalanNodeList *nodes= - vdoc.get_document(pool, &method_name).getElementsByTagName(XalanDOMString(tagname))) { - for(int i=0; igetLength(); i++) { - String& skey=*new(pool) String(pool); - { - char *buf=(char *)pool.malloc(MAX_NUMBER); - snprintf(buf, MAX_NUMBER, "%d", i); - skey << buf; +static void _load(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + // filespec + const String* uri=¶ms.as_string(0, "uri must be string"); + const char* uri_cstr; + if(uri->pos("://")==STRING_NOT_FOUND) // disk path + uri_cstr=r.absolute(*uri).cstr(String::L_FILE_SPEC); + else // xxx:// + uri_cstr=uri->cstr(String::L_AS_IS); // leave as-is for xmlParseFile to handle + + /// todo!! add SAFE MODE!! + GdomeDocument *document=(GdomeDocument *) + gdome_xml_n_mkref((xmlNode *)xmlParseFile(uri_cstr)); + if(!document || xmlHaveGenericErrors()) { + GdomeException exc=0; + throw XmlException(uri, exc); + } + // must be first action after if} + // replace any previous parsed source + { + vdoc.set_document(&r.charsets, document); + GdomeException exc; + gdome_doc_unref(document, &exc); + } +/* xmlParseFile does that itself. old peace for xmlParseMemory + const char* URI_cstr=uri->cstr(); + xmlDoc *doc=gdome_xml_doc_get_xmlDoc(document); + if(URI_cstr) + doc->URL=r.charsets.source().transcode_buf2xchar(URI_cstr, strlen(URI_cstr)); +*/ +} + +static void param_option_over_output_option( + HashStringValue& param_options, const char* option_name, + const String*& output_option) { + if(Value* value=param_options.get(String::Body(option_name))) + output_option=&value->as_string(); +} +static void param_option_over_output_option( + HashStringValue& param_options, const char* option_name, + bool& output_option) { + if(Value* value=param_options.get(String::Body(option_name))) { + const String& s=value->as_string(); + if(s=="yes") + output_option=true; + else if(s=="no") + output_option=false; + else + throw Exception("parser.runtime", + &s, + "%s must be either 'yes' or 'no'", option_name); + } +} + +/// @test valid_options check +static void prepare_output_options(Request& r, + MethodParams& params, size_t index, + VXdoc::Output_options& oo) { +/* + +*/ + + // configuring with options from parameter... + if(params.count()>index) { + Value& voptions=params.as_no_junction(index, "options must be string"); + if(voptions.is_defined()) { + if(HashStringValue *options=voptions.get_hash()) { + // $.method[xml|html|text] + if(Value* vmethod=options->get(String::Body(XDOC_OUTPUT_METHOD_OPTION_NAME))) + oo.method=&vmethod->as_string(); + + // $.version[1.0] + param_option_over_output_option(*options, "version", oo.version); + // $.encoding[windows-1251|...] + param_option_over_output_option(*options, "encoding", oo.encoding); + // $.omit-xml-declaration[yes|no] + param_option_over_output_option(*options, "omit-xml-declaration", oo.omitXmlDeclaration); + // $.standalone[yes|no] + param_option_over_output_option(*options, "standalone", oo.standalone); + // $.indent[yes|no] + param_option_over_output_option(*options, "indent", oo.indent); + // $.media-type[text/{html|xml|plain}] + param_option_over_output_option(*options, "media-type", oo.mediaType); } + } + } + + // default encoding from pool + if(!oo.encoding) + oo.encoding=new String(r.charsets.source().NAME(), String::L_TAINTED); + // default method=xml + if(!oo.method) + oo.method=new String(XDOC_OUTPUT_METHOD_OPTION_VALUE_XML); + // default mediaType = depending on method + if(!oo.mediaType) { + if(*oo.method==XDOC_OUTPUT_METHOD_OPTION_VALUE_XML) + oo.mediaType=new String("text/xml"); + else if(*oo.method==XDOC_OUTPUT_METHOD_OPTION_VALUE_HTML) + oo.mediaType=new String("text/html"); + else // XDOC_OUTPUT_METHOD_OPTION_VALUE_TEXT & all others + oo.mediaType=new String("text/plain"); + } +} + +/// patching piece from libxslt not to set meta encoding +static void +pa_xsltSaveResultToMem( + xmlChar*& doc_txt_ptr, int& doc_txt_len, + xmlDocPtr result, + xsltStylesheetPtr style, + xmlCharEncodingHandler* encoder) +{ + const xmlChar *encoding; + int base; + const xmlChar *method; + int indent; + xmlOutputBufferPtr buf = 0; + + if ((result == NULL) || (style == NULL)) + return; + if ((result->children == NULL) || + ((result->children->type == XML_DTD_NODE) && + (result->children->next == NULL))) + return; + + if ((style->methodURI != NULL) && + ((style->method == NULL) || + (!xmlStrEqual(style->method, (const xmlChar *) "xhtml")))) { + xsltGenericError(xsltGenericErrorContext, + "xsltSaveResultTo : unknown ouput method\n"); + return; + } + + XSLT_GET_IMPORT_PTR(method, style, method) + XSLT_GET_IMPORT_PTR(encoding, style, encoding) + XSLT_GET_IMPORT_INT(indent, style, indent); + + if ((method == NULL) && (result->type == XML_HTML_DOCUMENT_NODE)) + method = (const xmlChar *) "html"; + + if ((method != NULL) && + (xmlStrEqual(method, (const xmlChar *) "html") + ||xmlStrEqual(method, (const xmlChar *) "xhtml"))) { + if (indent == -1) + indent = 1; + // + // * xmlDocDumpFormatMemoryEnc: + // Note it is up to the caller of this function to free the + // allocated memory with xmlFree() + // + // we wont free anything, and wont copy that data anymore [already done inside and zeroterminated] + xmlDocDumpFormatMemoryEnc(result, &doc_txt_ptr, &doc_txt_len, (const char *) encoding, + indent); + } else if ((method != NULL) && + (xmlStrEqual(method, (const xmlChar *) "text"))) { + xmlNodePtr cur; + buf = xmlAllocOutputBuffer(encoder); + + cur = result->children; + while (cur != NULL) { + if (cur->type == XML_TEXT_NODE) + xmlOutputBufferWriteString(buf, (const char *) cur->content); + + /* + * Skip to next node + */ + if (cur->children != NULL) { + if ((cur->children->type != XML_ENTITY_DECL) && + (cur->children->type != XML_ENTITY_REF_NODE) && + (cur->children->type != XML_ENTITY_NODE)) { + cur = cur->children; + continue; + } + } + if (cur->next != NULL) { + cur = cur->next; + continue; + } + + do { + cur = cur->parent; + if (cur == NULL) + break; + if (cur == (xmlNodePtr) style->doc) { + cur = NULL; + break; + } + if (cur->next != NULL) { + cur = cur->next; + break; + } + } while (cur != NULL); + } + } else { + int omitXmlDecl; + int standalone; + buf = xmlAllocOutputBuffer(encoder); + + XSLT_GET_IMPORT_INT(omitXmlDecl, style, omitXmlDeclaration); + XSLT_GET_IMPORT_INT(standalone, style, standalone); + + if (omitXmlDecl != 1) { + xmlOutputBufferWriteString(buf, "version != NULL) + xmlBufferWriteQuotedString(buf->buffer, result->version); + else + xmlOutputBufferWriteString(buf, "\"1.0\""); + if (encoding == NULL) { + if (result->encoding != NULL) + encoding = result->encoding; + else if (result->charset != XML_CHAR_ENCODING_UTF8) + encoding = (const xmlChar *) + xmlGetCharEncodingName((xmlCharEncoding) + result->charset); + } + if (encoding != NULL) { + xmlOutputBufferWriteString(buf, " encoding="); + xmlBufferWriteQuotedString(buf->buffer, (xmlChar *) encoding); + } + switch (standalone) { + case 0: + xmlOutputBufferWriteString(buf, " standalone=\"no\""); + break; + case 1: + xmlOutputBufferWriteString(buf, " standalone=\"yes\""); + break; + default: + break; + } + xmlOutputBufferWriteString(buf, "?>\n"); + } + if (result->children != NULL) { + xmlNodePtr child = result->children; + + while (child != NULL) { + xmlNodeDumpOutput(buf, result, child, 0, (indent == 1), + (const char *) encoding); + if (child->type == XML_DTD_NODE) + xmlOutputBufferWriteString(buf, "\n"); + child = child->next; + } + xmlOutputBufferWriteString(buf, "\n"); + } + } - result.hash().put(skey, new(pool) VXnode(pool, nodes->item(i))); + if(buf) { + xmlOutputBufferFlush(buf); + if(buf->conv) { + doc_txt_len=buf->conv->use; + doc_txt_ptr=buf->conv->content; + } else { + doc_txt_len=buf->buffer->use; + doc_txt_ptr=buf->buffer->content; } + + if(doc_txt_ptr && doc_txt_len) + doc_txt_ptr=BAD_CAST pa_strdup((const char*)doc_txt_ptr, doc_txt_len); + + xmlOutputBufferClose(buf); + } +} + +struct Xdoc2buf_result { + char* str; + size_t length; +}; +static Xdoc2buf_result xdoc2buf(Request& r, VXdoc& vdoc, + MethodParams& params, int index, + VXdoc::Output_options& oo, + const String* file_spec) { + Xdoc2buf_result result; + prepare_output_options(r, params, index, + oo); + + const char* encoding_cstr=oo.encoding->cstr(); + xmlCharEncodingHandler *encoder=xmlFindCharEncodingHandler(encoding_cstr); + if(!encoder) + throw Exception("parser.runtime", + 0, + "encoding '%s' not supported", encoding_cstr); + // UTF-8 encoder contains empty input/output converters, + // which is wrong for xmlOutputBufferCreateIO + // while zero encoder goes perfectly + const char* encoder_name=encoder->name; + if(strcmp(encoder_name, "UTF-8")==0) + encoder=0; + + xsltStylesheet_auto_ptr stylesheet(xsltNewStylesheet()); + if(!stylesheet.get()) + throw Exception(0, + 0, + "xsltNewStylesheet failed"); + + #define OOS2STYLE(name) \ + stylesheet->name=oo.name?BAD_CAST xmlMemStrdup(r.transcode(*oo.name)->str):0 + #define OOE2STYLE(name) \ + stylesheet->name=oo.name + + OOS2STYLE(method); + OOS2STYLE(encoding); + OOS2STYLE(mediaType); +// OOS2STYLE(doctypeSystem); +// OOS2STYLE(doctypePublic); + OOE2STYLE(indent); + OOS2STYLE(version); + OOE2STYLE(standalone); + OOE2STYLE(omitXmlDeclaration); + + xmlDoc *document=gdome_xml_doc_get_xmlDoc(vdoc.get_document()); + document->encoding=BAD_CAST xmlMemStrdup(encoder_name); + + xmlChar* doc_txt_ptr=0; + int doc_txt_len=0; + pa_xsltSaveResultToMem(doc_txt_ptr, doc_txt_len, document, stylesheet.get(), encoder); + if(xmlHaveGenericErrors()) { + GdomeException exc=0; + throw XmlException(0, exc); } + result.length=doc_txt_len; + result.str=(char*)doc_txt_ptr; + + if(file_spec) + file_write(*file_spec, + result.str, result.length, + true/*as_text*/); + + return result; +} + +static void _file(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + VXdoc::Output_options oo(vdoc.output_options); + Xdoc2buf_result buf=xdoc2buf(r, vdoc, params, 0, + oo, + 0/*not to file, to memory*/); // write out result - r.write_no_lang(result); + r.write_no_lang(String(buf.str, buf.length)); + + // write out result + VFile& vfile=*new VFile; + VHash& vhcontent_type=*new VHash; + vhcontent_type.hash().put( + value_name, + new VString(*oo.mediaType)); + vhcontent_type.hash().put( + String::Body("charset"), + new VString(*oo.encoding)); + + vfile.set(false/*tainted*/, buf.str?buf.str:""/*to distinguish from stat-ed file*/, buf.length, + 0/*file_name*/, &vhcontent_type); + r.write_no_lang(vfile); } -static void _getElementsByTagNameNS(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - VXdoc& vdoc=*static_cast(r.self); +static void _save(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); - // namespaceURI;localName - const char *namespaceURI=params->as_string(0, "namespaceURI must not be code").cstr(String::UL_AS_IS); - const char *localName=params->as_string(0, "localName must not be code").cstr(String::UL_AS_IS); + const String& file_spec=r.absolute(params.as_string(0, "file name must be string")); + + VXdoc::Output_options oo(vdoc.output_options); + xdoc2buf(r, vdoc, params, 1, + oo, + &file_spec); +} - VHash& result=*new(pool) VHash(pool); - if(const XalanNodeList *nodes= - vdoc.get_document(pool, &method_name).getElementsByTagNameNS( - XalanDOMString(namespaceURI), XalanDOMString(localName))) { - for(int i=0; igetLength(); i++) { - String& skey=*new(pool) String(pool); - { - char *buf=(char *)pool.malloc(MAX_NUMBER); - snprintf(buf, MAX_NUMBER, "%d", i); - skey << buf; - } +static void _string(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + VXdoc::Output_options oo(vdoc.output_options); + Xdoc2buf_result buf=xdoc2buf(r, vdoc, params, 0, + oo, + 0/*not to file, to memory*/); + // write out result + r.write_no_lang(String(buf.str, buf.length)); +} - result.hash().put(skey, new(pool) VXnode(pool, nodes->item(i))); +#ifndef DOXYGEN +struct Add_xslt_param_info { + Request* r; + Array* strings; + const char** current_transform_param; +}; +#endif +static void add_xslt_param( + HashStringValue::key_type attribute, + HashStringValue::value_type meaning, + Add_xslt_param_info* info) { + GdomeDOMString_auto_ptr s; + *info->current_transform_param++=(s=info->r->transcode(attribute))->str; *info->strings+=s; + *info->current_transform_param++=(s=info->r->transcode(meaning->as_string()))->str; *info->strings+=s; +} + +static VXdoc& _transform(Request& r, const String* stylesheet_source, + VXdoc& vdoc, xsltStylesheetPtr stylesheet, const char** transform_params) +{ + xmlDoc *document=gdome_xml_doc_get_xmlDoc(vdoc.get_document()); + xsltTransformContext_auto_ptr transformContext( + xsltNewTransformContext(stylesheet, document)); + // make params literal + if (transformContext->globalVars == NULL) // strangly not initialized by xsltNewTransformContext + transformContext->globalVars = xmlHashCreate(20); + xsltQuoteUserParams(transformContext.get(), transform_params); + // do transform + xmlDoc *transformed=xsltApplyStylesheetUser( + stylesheet, + document, + 0/*already quoted-inserted transform_params*/, + 0/*const char* output*/, + 0/*FILE *profile*/, + transformContext.get()); + if(!transformed || xmlHaveGenericErrors()) { + GdomeException exc=0; + throw XmlException(stylesheet_source, exc); + } + + //gdome_xml_doc_mkref dislikes XML_HTML_DOCUMENT_NODE type, fixing + transformed->type=XML_DOCUMENT_NODE; + // constructing result + GdomeDocument *gdomeDocument=gdome_xml_doc_mkref(transformed); + if(!gdomeDocument) + throw Exception(0, + 0, + "gdome_xml_doc_mkref failed"); + VXdoc& result=*new VXdoc(&r.charsets, gdomeDocument); + /* grabbing options + + + */ + VXdoc::Output_options& oo=result.output_options; + + oo.method=stylesheet->method?&r.transcode(stylesheet->method):0; + oo.encoding=stylesheet->encoding?&r.transcode(stylesheet->encoding):0; + oo.mediaType=stylesheet->mediaType?&r.transcode(stylesheet->mediaType):0; + oo.indent=stylesheet->indent!=0; + oo.version=stylesheet->version?&r.transcode(stylesheet->version):0; + oo.standalone=stylesheet->standalone!=0; + oo.omitXmlDeclaration=stylesheet->omitXmlDeclaration!=0; + + // return + return result; +} +static void _transform(Request& r, MethodParams& params) { + VXdoc& vdoc=GET_SELF(r, VXdoc); + + // params + Array transform_strings; + const char** transform_params=0; + if(params.count()>1) { + Value& vparams=params.as_no_junction(1, "transform parameters must be hash"); + if(!vparams.is_string()) + if(HashStringValue* hash=vparams.get_hash()) { + transform_params=new(UseGC) const char*[hash->count()*2+1]; + Add_xslt_param_info info={ + &r, + &transform_strings, + transform_params + }; + hash->for_each(add_xslt_param, &info); + transform_params[hash->count()*2]=0; + } else + throw Exception("parser.runtime", + 0, + "transform parameters parameter must be hash"); + } + + VXdoc* result; + if(Value *vxdoc=params[0].as(VXDOC_TYPE, false)) { // stylesheet (xdoc) + xmlDoc *document=gdome_xml_doc_get_xmlDoc( + static_cast(vxdoc)->get_document()); + // compile xdoc stylesheet + xsltStylesheet_auto_ptr stylesheet_ptr(xsltParseStylesheetDoc(document)); + if(xmlHaveGenericErrors()) { + GdomeException exc=0; + throw XmlException(0, exc); } + if(!stylesheet_ptr.get()) + throw Exception("xml", + 0, + "stylesheet failed to compile"); + // strange thing - xsltParseStylesheetDoc records document and destroys it in stylesheet destructor + // we don't need that + stylesheet_ptr->doc=0; + + // transform! + result=&_transform(r, 0, + vdoc, stylesheet_ptr.get(), + transform_params); + } else { // stylesheet (file name) + // extablish stylesheet connection + const String& stylesheet_filespec= + r.absolute(params.as_string(0, "stylesheet must be file name (string) or DOM document (xdoc)")); + Stylesheet_connection_ptr connection=stylesheet_manager->get_connection(stylesheet_filespec); + + // load and compile file to stylesheet [or get cached if any] + // transform! + result=&_transform(r, &stylesheet_filespec, vdoc, connection->stylesheet(), + transform_params); } // write out result - r.write_no_lang(result); + r.write_no_lang(*result); } // constructor -MXdoc::MXdoc(Pool& apool) : MXnode(apool) { - set_name(*NEW String(pool(), XDOC_CLASS_NAME)); +/// @test how to create empty type html? +MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME, xnode_class) { + /// DOM1 + + // Element createElement(in DOMString tagName) raises(DOMException); + add_native_method("createElement", Method::CT_DYNAMIC, _createElement, 1, 1); + // DocumentFragment createDocumentFragment(); + add_native_method("createDocumentFragment", Method::CT_DYNAMIC, _createDocumentFragment, 0, 0); + // Text createTextNode(in DOMString data); + add_native_method("createTextNode", Method::CT_DYNAMIC, _createTextNode, 1, 1); + // Comment createComment(in DOMString data); + add_native_method("createComment", Method::CT_DYNAMIC, _createComment, 1, 1); + // CDATASection createCDATASection(in DOMString data) raises(DOMException); + add_native_method("createCDATASection", Method::CT_DYNAMIC, _createCDATASection, 1, 1); + // ProcessingInstruction createProcessingInstruction(in DOMString target, in DOMString data) raises(DOMException); + add_native_method("createProcessingInstruction", Method::CT_DYNAMIC, _createProcessingInstruction, 2, 2); + // Attr createAttribute(in DOMString name) raises(DOMException); + add_native_method("createAttribute", Method::CT_DYNAMIC, _createAttribute, 1, 1); + // EntityReference createEntityReference(in DOMString name) raises(DOMException); + add_native_method("createEntityReference", Method::CT_DYNAMIC, _createEntityReference, 1, 1); + // NodeList getElementsByTagName(in DOMString name); + add_native_method("getElementsByTagName", Method::CT_DYNAMIC, _getElementsByTagName, 1, 1); + + /// DOM2 + + // ^xdoc.getElementById[elementId] + add_native_method("getElementById", Method::CT_DYNAMIC, _getElementById, 1, 1); + + // Node (in Node importedNode, in boolean deep) raises(DOMException) + add_native_method("importNode", Method::CT_DYNAMIC, _importNode, 2, 2); + + // Attr createAttributeNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); + add_native_method("createAttributeNS", Method::CT_DYNAMIC, _createAttributeNS, 2, 2); + + // Element createElementNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); + add_native_method("createElementNS", Method::CT_DYNAMIC, _createElementNS, 2, 2); + + // NodeList getElementsByTagNameNS(in DOMString namespaceURI, in DOMString localName); + add_native_method("getElementsByTagNameNS", Method::CT_DYNAMIC, _getElementsByTagNameNS, 2, 2); + + /// parser + + // ^xdoc::create{qualifiedName} + // ^xdoc::create[xml] + // ^xdoc::create[URI][xml] + add_native_method("create", Method::CT_DYNAMIC, _create, 1, 2); + // for backward compatibility with <=v 1.82 2002/01/31 11:51:46 paf + add_native_method("set", Method::CT_DYNAMIC, _create, 1, 1); + + // ^xdoc::load[some.xml] + add_native_method("load", Method::CT_DYNAMIC, _load, 1, 1); // ^xdoc.save[some.xml] // ^xdoc.save[some.xml;options hash] @@ -417,40 +1070,18 @@ MXdoc::MXdoc(Pool& apool) : MXnode(apool // ^xdoc.file[options hash] file with "" add_native_method("file", Method::CT_DYNAMIC, _file, 0, 1); - // ^xdoc::set[xml] - add_native_method("set", Method::CT_DYNAMIC, _set, 1, 1); + // ^xdoc.transform[stylesheet file_name/xdoc] + // ^xdoc.transform[stylesheet file_name/xdoc;params hash] + add_native_method("transform", Method::CT_DYNAMIC, _transform, 1, 2); - // ^xdoc::load[some.xml] - add_native_method("load", Method::CT_DYNAMIC, _load, 1, 1); +} - // ^xdoc.xslt[stylesheet file_name] - // ^xdoc.xslt[stylesheet file_name;params hash] - add_native_method("xslt", Method::CT_DYNAMIC, _xslt, 1, 2); +# else - // ^xdoc.getElementById[elementId] - add_native_method("getElementById", Method::CT_DYNAMIC, _getElementById, 1, 1); - - // ^xdoc.getElementsByTagName[tagname] - add_native_method("getElementsByTagName", Method::CT_DYNAMIC, _getElementsByTagName, 1, 1); - - // ^xdoc.getElementsByTagNameNS[namespaceURI;localName] = array of nodes - add_native_method("getElementsByTagNameNS", Method::CT_DYNAMIC, _getElementsByTagNameNS, 2, 2); +#include "classes.h" -} // global variable -Methoded *Xdoc_class; - -// creator +DECLARE_CLASS_VAR(xdoc, 0, 0); // fictive #endif - -Methoded *MXdoc_create(Pool& pool) { - return -#ifdef XML - Xdoc_class=new(pool) MXdoc(pool); -#else - 0 -#endif - ; -}