--- parser3/src/classes/xdoc.C 2004/01/30 14:35:11 1.129 +++ parser3/src/classes/xdoc.C 2004/02/20 10:41:16 1.141 @@ -1,7 +1,7 @@ /** @file Parser: @b xdoc parser class. - Copyright (c) 2001-2003 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ @@ -9,15 +9,16 @@ #ifdef XML -static const char * const IDENT_XDOC_C="$Date: 2004/01/30 14:35:11 $"; +static const char * const IDENT_XDOC_C="$Date: 2004/02/20 10:41:16 $"; #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" @@ -594,6 +595,160 @@ static void prepare_output_options(Reque } } +/// patching piecees from libxslt and libxml not to set meta encoding +static int +pa_xsltSaveResultTo(xmlOutputBufferPtr buf, xmlDocPtr result, + xsltStylesheetPtr style, + const xmlChar *header_encoding) { + const xmlChar *encoding; + int base; + const xmlChar *method; + int indent; + + if ((buf == NULL) || (result == NULL) || (style == NULL)) + return(-1); + if ((result->children == NULL) || + ((result->children->type == XML_DTD_NODE) && + (result->children->next == NULL))) + return(0); + + if ((style->methodURI != NULL) && + ((style->method == NULL) || + (!xmlStrEqual(style->method, (const xmlChar *) "xhtml")))) { + xsltGenericError(xsltGenericErrorContext, + "xsltSaveResultTo : unknown ouput method\n"); + return(-1); + } + + base = buf->written; + + XSLT_GET_IMPORT_PTR(method, style, method) + XSLT_GET_IMPORT_PTR(encoding, style, encoding) + if(header_encoding) + encoding=header_encoding; + XSLT_GET_IMPORT_INT(indent, style, indent); + + if ((method == NULL) && (result->type == XML_HTML_DOCUMENT_NODE)) + method = BAD_CAST "html"; + + int omitXmlDecl; + XSLT_GET_IMPORT_INT(omitXmlDecl, style, omitXmlDeclaration); + + if ((method != NULL) && + (xmlStrEqual(method, (const xmlChar *) "html"))) { + if (indent == -1) + indent = 1; + + int is_xhtml = 0; + xmlDtdPtr dtd = xmlGetIntSubset(result); + if (dtd != NULL) { + is_xhtml = xmlIsXHTML(dtd->SystemID, dtd->ExternalID); + if (is_xhtml < 0) + is_xhtml = 0; + } + if(is_xhtml) { + method = BAD_CAST "xml"; + omitXmlDecl = 0; + } else { + htmlDocContentDumpFormatOutput(buf, result, (const char *) encoding, + indent); + xmlOutputBufferFlush(buf); + goto finish; + } + } + + if ((method != NULL) && + (xmlStrEqual(method, (const xmlChar *) "text"))) { + xmlNodePtr cur; + + 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); + } + xmlOutputBufferFlush(buf); + } else { + int standalone; + + 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 = BAD_CAST 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"); + } + xmlOutputBufferFlush(buf); + } +finish: + return(buf->written - base); +} + + struct Xdoc2buf_result { char* str; size_t length; @@ -601,24 +756,33 @@ struct Xdoc2buf_result { static Xdoc2buf_result xdoc2buf(Request& r, VXdoc& vdoc, MethodParams& params, int index, VXdoc::Output_options& oo, - const String* file_spec) { + const String* file_spec, + bool use_source_charset_to_render_and_client_charset_to_write_to_header=false) { Xdoc2buf_result result; prepare_output_options(r, params, index, oo); - const char* encoding_cstr=oo.encoding->cstr(); - xmlCharEncodingHandler *encoder=xmlFindCharEncodingHandler(encoding_cstr); - if(!encoder) + const char* render_encoding; + const char* header_encoding; + if(use_source_charset_to_render_and_client_charset_to_write_to_header) { + render_encoding=r.charsets.source().NAME_CSTR(); + header_encoding=r.charsets.client().NAME_CSTR(); + } else { + header_encoding=render_encoding=oo.encoding->cstr(); + } + + xmlCharEncodingHandler *renderer=xmlFindCharEncodingHandler(render_encoding); + if(!renderer) throw Exception("parser.runtime", 0, - "encoding '%s' not supported", encoding_cstr); - // UTF-8 encoder contains empty input/output converters, + "encoding '%s' not supported", render_encoding); + // UTF-8 renderer contains empty input/output converters, // which is wrong for xmlOutputBufferCreateIO - // while zero encoder goes perfectly - if(encoder && strcmp(encoder->name, "UTF-8")==0) - encoder=0; + // while zero renderer goes perfectly + if(strcmp(render_encoding, "UTF-8")==0) + renderer=0; - xmlOutputBuffer_auto_ptr outputBuffer(xmlAllocOutputBuffer(encoder)); + xmlOutputBuffer_auto_ptr outputBuffer(xmlAllocOutputBuffer(renderer)); xsltStylesheet_auto_ptr stylesheet(xsltNewStylesheet()); if(!stylesheet.get()) @@ -635,16 +799,16 @@ static Xdoc2buf_result xdoc2buf(Request& OOS2STYLE(encoding); OOS2STYLE(mediaType); // OOS2STYLE(doctypeSystem); - //OOS2STYLE(doctypePublic); +// OOS2STYLE(doctypePublic); OOE2STYLE(indent); OOS2STYLE(version); OOE2STYLE(standalone); OOE2STYLE(omitXmlDeclaration); xmlDoc *document=gdome_xml_doc_get_xmlDoc(vdoc.get_document()); - if(encoder) - document->encoding=BAD_CAST xmlMemStrdup(encoder->name); - if(xsltSaveResultTo(outputBuffer.get(), document, stylesheet.get())<0) { + document->encoding=BAD_CAST xmlMemStrdup(render_encoding); + if(pa_xsltSaveResultTo(outputBuffer.get(), document, stylesheet.get(), BAD_CAST header_encoding)<0 + || xmlHaveGenericErrors()) { GdomeException exc=0; throw XmlException(0, exc); } @@ -712,7 +876,8 @@ static void _string(Request& r, MethodPa VXdoc::Output_options oo(vdoc.output_options); Xdoc2buf_result buf=xdoc2buf(r, vdoc, params, 0, oo, - 0/*not to file, to memory*/); + 0/*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)); } @@ -732,8 +897,10 @@ static void add_xslt_param( *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) { + 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));