--- parser3/src/classes/xdoc.C 2004/02/17 15:08:14 1.136 +++ parser3/src/classes/xdoc.C 2020/12/15 16:37:40 1.199 @@ -1,7 +1,7 @@ /** @file Parser: @b xdoc parser class. - Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ @@ -9,9 +9,6 @@ #ifdef XML -static const char * const IDENT_XDOC_C="$Date: 2004/02/17 15:08:14 $"; - -#include "gdome.h" #include "libxml/tree.h" #include "libxml/HTMLtree.h" #include "libxslt/xsltInternals.h" @@ -29,21 +26,19 @@ static const char * const IDENT_XDOC_C=" #include "pa_vfile.h" #include "pa_xml_exception.h" #include "xnode.h" +#include "pa_charsets.h" + +volatile const char * IDENT_XDOC_C="$Id: xdoc.C,v 1.199 2020/12/15 16:37:40 moko Exp $"; // defines #define XDOC_CLASS_NAME "xdoc" -#define XDOC_OUTPUT_METHOD_OPTION_NAME "method" -#define XDOC_OUTPUT_METHOD_OPTION_VALUE_XML "xml" -#define XDOC_OUTPUT_METHOD_OPTION_VALUE_HTML "html" -#define XDOC_OUTPUT_METHOD_OPTION_VALUE_TEXT "text" - // class class MXdoc: public MXnode { public: // VStateless_class - Value* create_new_value(Pool&) { return new VXdoc(0, 0); } + Value* create_new_value(Pool&) { return new VXdoc(); } public: MXdoc(); @@ -52,7 +47,8 @@ public: // global variable -DECLARE_CLASS_VAR(xdoc, new MXdoc, 0); +DECLARE_CLASS_VAR(xnode, new MXnode); // must be here as Xdoc is inherited from Xnode and should be inited before +DECLARE_CLASS_VAR(xdoc, new MXdoc); // helper classes @@ -122,361 +118,279 @@ private: xsltTransformContext *_Ptr; }; -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; -}; - // methods -static void writeNode(Request& r, GdomeNode *node, - GdomeException exc) { - if(!node || exc) - throw XmlException(0, exc); +static void writeNode(Request& r, VXdoc& xdoc, xmlNode* node) { + if(!node) + throw Exception(PARSER_RUNTIME, + 0, + "error creating node"); // OOM, bad name, things like that // write out result - r.write_no_lang(*new VXnode(&r.charsets, node)); + r.write(xdoc.wrap(*node)); +} + +struct IdsIteratorInfo { + xmlChar *elementId; + xmlNode *element; +}; + +/* switching to calling convetion of libxml */ +extern "C" void idsHashScanner (void *payload, void *data, const xmlChar *name) { + IdsIteratorInfo *priv = (IdsIteratorInfo *)data; + + if (priv->element == NULL && xmlStrEqual (name, priv->elementId)) + { + xmlNode* parent=((xmlID *)payload)->attr->parent; + assert(parent); + priv->element=parent; + } +} + +static xmlNode* pa_getElementById(xmlDoc& xmldoc, xmlChar* elementId) { + xmlHashTable *ids = (xmlHashTable *)xmldoc.ids; + IdsIteratorInfo iter={elementId, NULL}; + xmlHashScan(ids, (xmlHashScanner)idsHashScanner, &iter); + return iter.element; +} + +/* +static xmlNode * +pa_importNode (xmlDoc& xmldoc, xmlNode& importedNode, bool deep) { + xmlNode *result = NULL; + + switch (importedNode.type) { + case XML_ATTRIBUTE_NODE: + result = (xmlNode *)xmlCopyProp(xmldoc, (xmlAttr *)importedNode); + result.parent=0; // no idea + break; + case XML_DOCUMENT_FRAG_NODE: + case XML_ELEMENT_NODE: + case XML_ENTITY_REF_NODE: + case XML_PI_NODE: + case XML_TEXT_NODE: + case XML_CDATA_SECTION_NODE: + case XML_COMMENT_NODE: + result = xmlCopyNode (importedNode->n, deep); + xmlSetTreeDoc (result, priv->n); + break; + default: + *exc = GDOME_NOT_SUPPORTED_ERR; + } + + return result; } +*/ // Element createElement(in DOMString tagName) raises(DOMException); static void _createElement(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* tagName=as_xmlname(r, params, 0, "tagName must be string"); - const String& tagName=params.as_string(0, "tagName must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node= - (GdomeNode *)gdome_doc_createElement(vdoc.get_document(), - r.transcode(tagName).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewDocNode(&xmldoc, NULL, tagName, NULL); + writeNode(r, vdoc, node); } // Element createElementNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); static void _createElementNS(Request& r, MethodParams& params) { + xmlChar* namespaceURI=as_xmlnsuri(r, params, 0); + xmlChar* qualifiedName=as_xmlqname(r, params, 1); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - // namespaceURI;localName - const String& namespaceURI=params.as_string(0, "namespaceURI must be string"); - const String& qualifiedName=params.as_string(1, "qualifiedName must be string"); - - GdomeException exc; - GdomeNode *node= - (GdomeNode *)gdome_doc_createElementNS(vdoc.get_document(), - r.transcode(namespaceURI).use(), - r.transcode(qualifiedName).use(), - &exc); - writeNode(r, node, exc); + xmlChar* prefix=0; + xmlChar* localName=xmlSplitQName2(qualifiedName, &prefix); + + xmlNode *node; + if(localName) { + xmlNs& ns=pa_xmlMapNs(xmldoc, namespaceURI, prefix); + node=xmlNewDocNode(&xmldoc, &ns, localName, NULL); + } else + node=xmlNewDocNode(&xmldoc, NULL, qualifiedName/*unqualified, actually*/, NULL); + writeNode(r, vdoc, node); } // DocumentFragment createDocumentFragment() static void _createDocumentFragment(Request& r, MethodParams&) { VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node= - (GdomeNode *)gdome_doc_createDocumentFragment( - vdoc.get_document(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewDocFragment(&xmldoc); + writeNode(r, vdoc, node); } // Text createTextNode(in DOMString data); static void _createTextNode(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* data=as_xmlchar(r, params, 0, XML_DATA_MUST_BE_STRING); - const String& data=params.as_string(0, "data must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node=(GdomeNode *)gdome_doc_createTextNode( - vdoc.get_document(), - r.transcode(data).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewDocText(&xmldoc, data); + writeNode(r, vdoc, node); } // Comment createComment(in DOMString data) static void _createComment(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* data=as_xmlchar(r, params, 0, XML_DATA_MUST_BE_STRING); - const String& data=params.as_string(0, "data must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); - GdomeException exc; - GdomeNode *node=(GdomeNode *)gdome_doc_createComment( - vdoc.get_document(), - r.transcode(data).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewComment(data); + writeNode(r, vdoc, node); } // CDATASection createCDATASection(in DOMString data) raises(DOMException); static void _createCDATASection(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* data=as_xmlchar(r, params, 0, XML_DATA_MUST_BE_STRING); - const String& data=params.as_string(0, "data must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node=(GdomeNode *)gdome_doc_createCDATASection( - vdoc.get_document(), - r.transcode(data).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewCDataBlock(&xmldoc, data, strlen((const char*)data)); + writeNode(r, vdoc, node); } // ProcessingInstruction createProcessingInstruction(in DOMString target,in DOMString data) raises(DOMException); static void _createProcessingInstruction(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* target=as_xmlchar(r, params, 0, XML_DATA_MUST_BE_STRING); + xmlChar* data=as_xmlchar(r, params, 1, XML_DATA_MUST_BE_STRING); - const String& target=params.as_string(0, "data must be string"); - const String& data=params.as_string(1, "data must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node=(GdomeNode *)gdome_doc_createProcessingInstruction( - vdoc.get_document(), - r.transcode(target).use(), - r.transcode(data).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=xmlNewDocPI(&xmldoc, target, data); + writeNode(r, vdoc, node); } // Attr createAttribute(in DOMString name) raises(DOMException); static void _createAttribute(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* name=as_xmlname(r, params, 0); - const String& name=params.as_string(0, "name must be string"); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - GdomeException exc; - GdomeNode *node=(GdomeNode *)gdome_doc_createAttribute( - vdoc.get_document(), - r.transcode(name).use(), - &exc); - writeNode(r, node, exc); + xmlNode *node=(xmlNode*)xmlNewDocProp(&xmldoc, name, 0); + writeNode(r, vdoc, node); } // Attr createAttributeNS(in DOMString namespaceURI, in DOMString qualifiedName) raises(DOMException); static void _createAttributeNS(Request& r, MethodParams& params) { + xmlChar* namespaceURI=as_xmlnsuri(r, params, 0); + xmlChar* qualifiedName=as_xmlqname(r, params, 1); + VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); + + xmlChar* prefix=0; + xmlChar* localName=xmlSplitQName2(qualifiedName, &prefix); - // 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); + xmlNode *node; + if(localName) { + xmlNs& ns=pa_xmlMapNs(xmldoc, namespaceURI, prefix); + node=(xmlNode*)xmlNewDocProp(&xmldoc, localName, NULL); + xmlSetNs(node, &ns); + } else + node=(xmlNode*)xmlNewDocProp(&xmldoc, qualifiedName/*unqualified, actually*/, NULL); + writeNode(r, vdoc, node); } // EntityReference createEntityReference(in DOMString name) raises(DOMException); static void _createEntityReference(Request& r, MethodParams& params) { - VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlChar* name=as_xmlname(r, params, 0); - 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); -} - -// NodeList getElementsByTagName(in DOMString name); -static void _getElementsByTagName(Request& r, MethodParams& params) { VXdoc& vdoc=GET_SELF(r, VXdoc); + xmlDoc& xmldoc=vdoc.get_xmldoc(); - 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...} - 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); - } + const String& xml=r.process_to_string(param); + String::Body sbody=xml.cstr_to_string_body_untaint(String::L_XML, r.connection(false), &r.charsets); - // must be last action in if, see after if} - } else { // [name] - const String& qualifiedName=param.as_string(); + xmldoc=xmlReadMemory(sbody.cstr(), sbody.length(), NULL, NULL, XML_PARSE_OPTIONS); - 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); + if(!xmldoc || xmlHaveGenericErrors()) + throw XmlException(0, r); - set_encoding=true; // must be last action in if, see after if} + } else { // [localName] + if(const String* value = param.get_string()){ + xmlChar* localName=r.transcode(*value); + if(xmlValidateNCName(localName, 0) != 0) + throw XmlException(0, XML_INVALID_LOCAL_NAME, localName); + + xmldoc=xmlNewDoc(0); + if(!xmldoc || xmlHaveGenericErrors()) + throw XmlException(0, r); + + xmlNode* node=xmlNewChild((xmlNode*)xmldoc, NULL, localName, NULL); + if(!node || xmlHaveGenericErrors()) + throw XmlException(0, r); + + set_encoding=true; + // must be last action in if, see after if} + } else { + VFile* vfile=param.as_vfile(String::L_AS_IS); + xmldoc=xmlReadMemory(vfile->value_ptr(), vfile->value_size(), NULL, NULL, XML_PARSE_OPTIONS); + if(!xmldoc || xmlHaveGenericErrors()) + throw XmlException(0, r); + } } // must be first action after if} // replace any previous parsed source - { - vdoc.set_document(&r.charsets, document); - GdomeException exc; - gdome_doc_unref(document, &exc); - } + vdoc.set_xmldoc(r.charsets, *xmldoc); // 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(); + URI_cstr=r.full_disk_path(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)); + xmldoc->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)); + xmldoc->encoding=source_charset.transcode_buf2xchar(source_charset_name, strlen(source_charset_name)); } } @@ -484,436 +398,192 @@ static void _load(Request& r, MethodPara VXdoc& vdoc=GET_SELF(r, VXdoc); // filespec - const String* uri=¶ms.as_string(0, "uri must be string"); + 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); + uri_cstr=r.full_disk_path(*uri).taint_cstr(String::L_FILE_SPEC); else // xxx:// - uri_cstr=uri->cstr(String::L_AS_IS); // leave as-is for xmlParseFile to handle + uri_cstr=uri->taint_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); - } + /// @todo!! add SAFE MODE!! + xmlDoc* xmldoc=xmlReadFile(uri_cstr, NULL, XML_PARSE_OPTIONS); + if(!xmldoc || xmlHaveGenericErrors()) + throw XmlException(uri, r); + // 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"); - } + vdoc.set_xmldoc(r.charsets, *xmldoc); } -/// 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"); - } - } - - 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); +String::C xdoc2buf(Request& r, VXdoc& vdoc, + XDocOutputOptions& oo, + const String* file_spec, + bool use_source_charset_to_render_and_client_charset_to_write_to_header=false) { + Charset* render=0; + Charset* header=0; + if(use_source_charset_to_render_and_client_charset_to_write_to_header) { + render=&r.charsets.source(); + header=&r.charsets.client(); + } else { + header=render=&pa_charsets.get(*oo.encoding); } -} + const char* render_encoding=render->NAME_CSTR(); + const char* header_encoding=header->NAME_CSTR(); -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, + xmlCharEncodingHandler *renderer=xmlFindCharEncodingHandler(render_encoding); + // UTF-8 renderer 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"); + // while zero renderer goes perfectly + if(render->isUTF8()) + renderer=0; + + xmlOutputBuffer_auto_ptr outputBuffer(xmlAllocOutputBuffer(renderer)); + + xsltStylesheet *stylesheet = xsltNewStylesheet(); + if(!stylesheet) + throw Exception(0, 0, "xsltNewStylesheet failed"); + + #define OOSTRING2STYLE(name) \ + stylesheet->name=oo.name?BAD_CAST xmlMemStrdup((const char*)r.transcode(*oo.name)):0 + #define OOBOOL2STYLE(name) \ + if(oo.name>=0) stylesheet->name=oo.name + + OOSTRING2STYLE(method); + OOSTRING2STYLE(encoding); + OOSTRING2STYLE(mediaType); +// OOSTRING2STYLE(doctypeSystem); +// OOSTRING2STYLE(doctypePublic); + OOBOOL2STYLE(indent); + OOSTRING2STYLE(version); + OOBOOL2STYLE(standalone); + OOBOOL2STYLE(omitXmlDeclaration); + + xmlDoc& xmldoc=vdoc.get_xmldoc(); + xmldoc.encoding=BAD_CAST xmlMemStrdup(render_encoding); + if(header_encoding) + stylesheet->encoding=BAD_CAST xmlMemStrdup(header_encoding); + if(xsltSaveResultTo(outputBuffer.get(), &xmldoc, stylesheet)<0 + || xmlHaveGenericErrors()) + throw XmlException(0, r); - #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); + // write out result + char *gnome_str; + size_t gnome_length; +#ifdef LIBXML2_NEW_BUFFER + if(outputBuffer->conv) { + gnome_length=xmlBufUse(outputBuffer->conv); + gnome_str=(char *)xmlBufContent(outputBuffer->conv); + } else { + gnome_length=xmlOutputBufferGetSize(&(*outputBuffer)); + gnome_str=(char *)xmlOutputBufferGetContent(&(*outputBuffer)); + } +#else + if(outputBuffer->conv) { + gnome_length=outputBuffer->conv->use; + gnome_str=(char *)outputBuffer->conv->content; + } else { + gnome_length=outputBuffer->buffer->use; + gnome_str=(char *)outputBuffer->buffer->content; } +#endif - result.length=doc_txt_len; - result.str=(char*)doc_txt_ptr; - - if(file_spec) - file_write(*file_spec, - result.str, result.length, + if(file_spec){ + file_write(r.charsets, + *file_spec, + gnome_str, + gnome_length, true/*as_text*/); + return String::C(); // actually, we don't need this output at all + } else + return String::C(gnome_length ? pa_strdup(gnome_str, gnome_length) : 0, gnome_length); +} - return result; +inline HashStringValue* get_options(MethodParams& params, size_t index){ + return (params.count()>index) ? params.as_hash(index) : 0; } 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(String(buf.str, buf.length)); - // write out result + XDocOutputOptions oo(vdoc.output_options); + oo.append(r, get_options(params, 0), true/* $.name[filename] could be specified by user */); + String::C buf=xdoc2buf(r, vdoc, oo, 0/*file_name. not to file, to memory*/); + 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); + vhcontent_type.hash().put(value_name, new VString(*oo.mediaType)); + vhcontent_type.hash().put("charset", new VString(*oo.encoding)); + + vfile.set_binary(false/*not tainted*/, buf.str?buf.str:""/*to distinguish from stat-ed file*/, buf.length, oo.filename, &vhcontent_type); + + // write out result + r.write(vfile); } static void _save(Request& r, MethodParams& params) { VXdoc& vdoc=GET_SELF(r, VXdoc); - const String& file_spec=r.absolute(params.as_string(0, "file name must be string")); + const String& file_spec=r.full_disk_path(params.as_string(0, FILE_NAME_MUST_BE_STRING)); - VXdoc::Output_options oo(vdoc.output_options); - xdoc2buf(r, vdoc, params, 1, - oo, - &file_spec); + XDocOutputOptions oo(vdoc.output_options); + oo.append(r, get_options(params, 1)); + xdoc2buf(r, vdoc, oo, &file_spec); } 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*/); + + XDocOutputOptions oo(vdoc.output_options); + oo.append(r, get_options(params, 0)); + String::C buf=xdoc2buf(r, vdoc, oo, + 0/*file_name. not to file, to memory*/, + true/*use source charset to render, client charset to put to header*/); + // write out result - r.write_no_lang(String(buf.str, buf.length)); + r.write(String(buf, String::L_AS_IS)); } #ifndef DOXYGEN struct Add_xslt_param_info { Request* r; - Array* strings; - const char** current_transform_param; + Array* strings; + const xmlChar** 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; + xmlChar* s; + *info->current_transform_param++=(s=info->r->transcode(attribute)); *info->strings+=s; + *info->current_transform_param++=(s=info->r->transcode(meaning->as_string())); *info->strings+=s; } -static VXdoc& _transform(Request& r, const String* stylesheet_source, - VXdoc& vdoc, xsltStylesheetPtr stylesheet, const char** transform_params) +static VXdoc& _transform(Request& r, const String* stylesheet_source, VXdoc& vdoc, xsltStylesheetPtr stylesheet, const xmlChar** transform_params) { - xmlDoc *document=gdome_xml_doc_get_xmlDoc(vdoc.get_document()); + xmlDoc& xmldoc=vdoc.get_xmldoc(); + xsltTransformContext_auto_ptr transformContext( - xsltNewTransformContext(stylesheet, document)); + xsltNewTransformContext(stylesheet, &xmldoc)); // make params literal if (transformContext->globalVars == NULL) // strangly not initialized by xsltNewTransformContext transformContext->globalVars = xmlHashCreate(20); - xsltQuoteUserParams(transformContext.get(), transform_params); + xsltQuoteUserParams(transformContext.get(), (const char**)transform_params); // do transform xmlDoc *transformed=xsltApplyStylesheetUser( stylesheet, - document, + &xmldoc, 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); - } + if(!transformed || xmlHaveGenericErrors()) + throw XmlException(stylesheet_source, r); //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); + VXdoc& result=*new VXdoc(r.charsets, *transformed); /* grabbing options */ - VXdoc::Output_options& oo=result.output_options; + XDocOutputOptions& 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.indent=stylesheet->indent; oo.version=stylesheet->version?&r.transcode(stylesheet->version):0; - oo.standalone=stylesheet->standalone!=0; - oo.omitXmlDeclaration=stylesheet->omitXmlDeclaration!=0; + oo.standalone=stylesheet->standalone; + oo.omitXmlDeclaration=stylesheet->omitXmlDeclaration; // return return result; @@ -946,68 +616,54 @@ static void _transform(Request& r, Metho 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"); - } + Array transform_strings; + const xmlChar** transform_params=0; + if(params.count()>1) + if(HashStringValue* hash=params.as_hash(1, "transform options")) { + transform_params=new(PointerGC) const xmlChar*[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; + } 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()); + if(Value *vxdoc=params[0].as(VXDOC_TYPE)) { // stylesheet (xdoc) + VXdoc& vstylesheet=static_cast(*vxdoc); + xmlDoc& stylesheetdoc=vstylesheet.get_xmldoc(); + // 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; + xsltStylesheet *stylesheet=xsltParseStylesheetDoc(&stylesheetdoc); + if(xmlHaveGenericErrors()) + throw XmlException(0, r); + if(!stylesheet) + throw Exception("xml", 0, "stylesheet failed to compile"); // transform! - result=&_transform(r, 0, - vdoc, stylesheet_ptr.get(), - transform_params); + result=&_transform(r, 0, vdoc, stylesheet, 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); + const String& stylesheet_filespec=r.full_disk_path(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); + result=&_transform(r, &stylesheet_filespec, vdoc, connection->stylesheet(), transform_params); } // write out result - r.write_no_lang(*result); + r.write(*result); } // constructor /// @test how to create empty type html? -MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME, xnode_class) { +MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME) { + set_base(xnode_class); + /// DOM1 // Element createElement(in DOMString tagName) raises(DOMException); @@ -1026,8 +682,6 @@ MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME, 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 @@ -1043,9 +697,6 @@ MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME, // 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} @@ -1082,6 +733,7 @@ MXdoc::MXdoc(): MXnode(XDOC_CLASS_NAME, // global variable -DECLARE_CLASS_VAR(xdoc, 0, 0); // fictive +DECLARE_CLASS_VAR(xnode, 0); // fictive +DECLARE_CLASS_VAR(xdoc, 0); // fictive #endif