XDocument* XParser::Parse(const char* szPath, bool keep_blank/*=false*/)
{
if(szPath==NULL)
return NULL;
xmlKeepBlanksDefault(keep_blank?1:0);
xmlDoValidityCheckingDefaultValue =0;
xmlParserCtxtPtr pxParseCtxt = NULL;
pxParseCtxt = xmlCreateFileParserCtxt(szPath);
if(pxParseCtxt==NULL)
{
return NULL;
}
if(pxParseCtxt->directory == NULL)
{
char* directory = xmlParserGetDirectory(szPath);
pxParseCtxt->directory = (char*)xmlStrdup((xmlChar*)directory);
}
if(!ParseContext(pxParseCtxt))
{
xmlFreeParserCtxt(pxParseCtxt);
return NULL;
}
XDocument* pagXmlDoc = NULL;
pagXmlDoc = new XDocument(pxParseCtxt->myDoc);
xmlFreeParserCtxt(pxParseCtxt);
return pagXmlDoc;
}
/* Taken from libxml2, xmlSAXParseMemoryWithData */
xmlDocPtr mxslt_doc_xml_load_entity(mxslt_doc_t * document, char * localfile) {
xmlParserCtxtPtr ctx;
xmlParserInputPtr input;
xmlDocPtr retval;
xmlChar * filename;
mxslt_doc_debug_print(document, MXSLT_DBG_LIBXML | MXSLT_DBG_DEBUG | MXSLT_DBG_VERBOSE0,
"load_entity/xmlCreateMemoryParserCtxt -- replacing entities: %08x\n", xmlSubstituteEntitiesDefaultValue);
/* SNIPPET: This is a good mix&shake of
* xmlCreateMemoryParserCtxt, xmlCreateFileParserCtxt */
ctx=xmlNewParserCtxt();
if(ctx == NULL)
return NULL;
#if LIBXML_VERSION >= 20600
xmlCtxtUseOptions(ctx, MXSLT_XSLT_OPTIONS);
#endif
/* Remember which document we are parsing
* in this context */
/* ctx->_private=document; */
filename=xmlCanonicPath((xmlChar *)localfile);
if(filename == NULL) {
xmlFreeParserCtxt(ctx);
return NULL;
}
input=xmlLoadExternalEntity((char *)filename, NULL, ctx);
xmlFree(filename);
if(input == NULL) {
xmlFreeParserCtxt(ctx);
return NULL;
}
inputPush(ctx, input);
if(ctx->directory == NULL)
ctx->directory=xmlParserGetDirectory(localfile);
/* END SNIPPET */
/* MXSLT_DUMP_CTX(ctx); */
/* Parse document */
xmlParseDocument(ctx);
if(ctx->wellFormed)
retval=ctx->myDoc;
else {
retval=NULL;
xmlFreeDoc(ctx->myDoc);
ctx->myDoc=NULL;
}
xmlFreeParserCtxt(ctx);
return retval;
}
std::auto_ptr<document> dom_parser::parse_file(const std::string& file_name)
{
set_global_variables global_variables;
// Create a libxml2 parser context for parsing the xml file.
dom_parser_context_wrapper context( xmlCreateFileParserCtxt(file_name.c_str()) );
if (context.get() == 0)
{
std::string what = "fail to parse xml file " + file_name + ": "
+ "unable to create libxml2 parser context";
throw dom_error(what);
}
if (context.get()->directory == 0)
{
const xmlChar* dir = detail::to_xml_chars(xmlParserGetDirectory(file_name.c_str()));
context.get()->directory = const_cast<char*>(detail::to_chars(xmlStrdup(dir)));
}
// Parse xml file under the constructed parser context.
xmlDoc* px = parse_in_context(context.get());
assert(px != 0);
return std::auto_ptr<document>(new document(px));
}
/* Taken from libxml2, xmlSAXParseMemoryWithData */
xmlDocPtr mxslt_doc_xml_parse(mxslt_doc_t * document, xmlParserInputBufferPtr buf, char * localfile) {
xmlParserCtxtPtr ctx;
xmlParserInputPtr input;
xmlDocPtr retval;
mxslt_doc_debug_print(document, MXSLT_DBG_LIBXML | MXSLT_DBG_DEBUG | MXSLT_DBG_VERBOSE0,
"xml_parse/xmlCreateMemoryParserCtxt -- replacing entities: %08x\n", xmlSubstituteEntitiesDefaultValue);
if(buf == NULL)
return NULL;
/* SNIPPET: This is a good mix&shake of
* xmlCreateMemoryParserCtxt, xmlCreateFileParserCtxt */
ctx=xmlNewParserCtxt();
if(ctx == NULL) {
xmlFreeParserInputBuffer(buf);
return NULL;
}
#if LIBXML_VERSION >= 20600
xmlCtxtUseOptions(ctx, MXSLT_XSLT_OPTIONS);
#endif
/* Default are no longer changed... since 2.6.0 they
* are completely ignored, leading to broken modxslt :|| */
if(ctx->sax) {
ctx->sax->resolveEntity=mxslt_sax_resolve_entity;
ctx->sax->processingInstruction=mxslt_sax_processing_instruction;
}
/* Remember which document we are parsing
* in this context */
/* ctx->_private=document; */
input=xmlNewInputStream(ctx);
if(input == NULL) {
xmlFreeParserInputBuffer(buf);
xmlFreeParserCtxt(ctx);
return NULL;
}
input->filename=(char *)xmlCanonicPath((xmlChar *)localfile);
if(input->filename == NULL) {
xmlFreeParserCtxt(ctx);
xmlFreeParserInputBuffer(buf);
xmlFreeInputStream(input);
return NULL;
}
input->buf=buf;
#if LIBXML_VERSION < 20900
input->base=input->buf->buffer->content;
input->cur=input->buf->buffer->content;
input->end=&input->buf->buffer->content[input->buf->buffer->use];
#else
/* With libxml2 2.9.0, the buffer struct can only be accessed through
* methods. */
input->base=xmlBufContent(input->buf->buffer);
input->cur=xmlBufContent(input->buf->buffer);
input->end=xmlBufEnd(input->buf->buffer);
#endif
inputPush(ctx, input);
if(ctx->directory == NULL)
ctx->directory=xmlParserGetDirectory(localfile);
/* END SNIPPET */
if(mxslt_doc_debug_level(document, MXSLT_DBG_LIBXML | MXSLT_DBG_DEBUG | MXSLT_DBG_VERBOSE2))
mxslt_doc_dump_ctx(ctx, document);
/* Parse document */
xmlParseDocument(ctx);
if(ctx->wellFormed)
retval=ctx->myDoc;
else {
retval=NULL;
xmlFreeDoc(ctx->myDoc);
ctx->myDoc=NULL;
}
xmlFreeParserCtxt(ctx);
/* xmlFreeParserInputBuffer(buf); */
return retval;
}
/* This function replaces processingInstruction */
void mxslt_sax_processing_instruction(void * ctx, const xmlChar * target, const xmlChar * data) {
xmlParserCtxtPtr ctxt=(xmlParserCtxtPtr)ctx;
mxslt_doc_t * document=(mxslt_get_state()->document);
mxslt_pi_style_t * style_pi;
mxslt_pi_param_t * param_pi;
mxslt_pi_base_t * base_pi;
char * type = NULL, * tmp;
xmlNodePtr pi;
int status;
/* WARNING: don't change this array! Look below! */
static const mxslt_attr_search_t xml_stylesheet[] = {
{ (xmlChar *)"href", mxslt_sizeof_str("href") },
{ (xmlChar *)"media", mxslt_sizeof_str("media") },
{ (xmlChar *)"type", mxslt_sizeof_str("type") }
};
/* WARNING: don't change this array! Look below! */
static const mxslt_attr_search_t mxslt_param[] = {
{ (xmlChar *)"name", mxslt_sizeof_str("name") },
{ (xmlChar *)"value", mxslt_sizeof_str("value") }
};
/* WARNING: don't change this array! Look below! */
static const mxslt_attr_search_t mxslt_base[] = {
{ (xmlChar *)"value", mxslt_sizeof_str("value") }
};
/* This array _must_ have the same number of elements
* as the array xml_stylesheet */
char * xml_stylesheet_values[] = { NULL, NULL, NULL };
enum xml_stylesheet_types {
t_href=0,
t_media,
t_type
};
/* This array _must_ have the same number of elements
* as the array mxslt_param */
char * mxslt_param_values[] = { NULL, NULL };
enum mxslt_param_types {
t_name=0,
t_select
};
/* This array _must_ have the same number of elements
* as the array mxslt_base */
char * mxslt_base_values[] = { NULL };
enum mxslt_base_types {
t_value=0
};
mxslt_doc_debug_print(document, MXSLT_DBG_LIBXML, "processing instruction \"%s\", attributes |%s|\n", target, data);
/* Ignore pi when we're parsing the stylesheet */
if(document->flags & MXSLT_STYLE)
return;
/* Create a new pi node */
pi=xmlNewPI(target, data);
if(!pi)
return;
/* Check if target is a known pi */
switch(target[0]) {
case 'x':
if(!xmlStrEqual(target+1, (xmlChar *)"xml-stylesheet"+1))
break;
/* Get href, type and media from attribute list */
status=mxslt_get_static_attr((char *)data, xml_stylesheet, xml_stylesheet_values, mxslt_sizeof_array(xml_stylesheet));
if(status != MXSLT_OK) {
if(xml_stylesheet_values[0])
xfree(xml_stylesheet_values[0]);
if(xml_stylesheet_values[1])
xfree(xml_stylesheet_values[1]);
if(xml_stylesheet_values[2])
xfree(xml_stylesheet_values[2]);
goto error;
}
if(xml_stylesheet_values[t_type] && xml_stylesheet_values[t_href]) {
/* Decode type */
type=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_type],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
/* Skip blanks before type */
for(tmp=type; MXSLT_IS_BLANK(*tmp); tmp++)
;
/* compare type string */
if(!strncmp(tmp, "text/xml", mxslt_sizeof_str("text/xml")) ||
!strncmp(tmp, "text/xsl", mxslt_sizeof_str("text/xsl"))) {
/* trailing spaces are allowed */
for(tmp+=mxslt_sizeof_str("text/x?l"); MXSLT_IS_BLANK(*tmp); tmp++)
;
/* Check there's nothing else beside them */
if(*tmp != '\0') {
xmlFree(type);
break;
}
/* Type is ok, add parameter to processing
* instruction */
style_pi=(mxslt_pi_style_t *)xmalloc(sizeof(mxslt_pi_style_t));
style_pi->type=MXSLT_PI_STDST;
style_pi->ctype=type;
style_pi->href=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_href],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
if(xml_stylesheet_values[t_media])
style_pi->media=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_media],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
else
style_pi->media=NULL;
/* tie up xmlNode and style_pi */
XML_LINK_PI(document, pi, style_pi);
} else {
mxslt_error(document, "warning - <xml-stylesheet type=\"... unknown: '%s'!\n", type);
xmlFree(type);
}
} else {
if(!xml_stylesheet_values[t_href])
mxslt_error(document, "warning - <xml-stylesheet href=\"... is missing, skipping PI!\n");
if(!xml_stylesheet_values[t_type])
mxslt_error(document, "warning - <xml-stylesheet type=\"... is missing, skipping PI!\n");
}
if(xml_stylesheet_values[t_media])
xfree(xml_stylesheet_values[t_media]);
if(xml_stylesheet_values[t_href])
xfree(xml_stylesheet_values[t_href]);
if(xml_stylesheet_values[t_type])
xfree(xml_stylesheet_values[t_type]);
break;
case 'm':
if(target[1] != 'o' || target[2] != 'd' ||
target[3] != 'x' || target[4] != 's' ||
target[5] != 'l' || target[6] != 't' ||
target[7] != '-')
break;
switch(target[8]) {
case 'b':
if(!xmlStrEqual(target+8, (xmlChar *)"modxslt-base"+8)) {
mxslt_error(document, "warning - unknown modxslt PI: %s in %s\n", target, document->localfile);
break;
}
/* Read modxslt-base */
status=mxslt_get_static_attr((char *)data, mxslt_base, mxslt_base_values, mxslt_sizeof_array(mxslt_base));
if(status != MXSLT_OK) {
if(mxslt_base_values[0])
xfree(mxslt_base_values[0]);
goto error;
}
/* Verify we got the base */
if(!mxslt_base_values[0]) {
mxslt_error(document, "warning - <modxslt-base ... base=\" is missing, skipping PI\n", target, document->localfile);
break;
}
/* Remember the pi we found */
base_pi=(mxslt_pi_base_t *)(xmalloc(sizeof(mxslt_pi_base_t)));
base_pi->type=MXSLT_PI_UBASE;
base_pi->file=(char *)xmlCanonicPath((xmlChar *)mxslt_base_values[0]);
base_pi->directory=xmlParserGetDirectory(mxslt_base_values[0]);
xfree(mxslt_base_values[0]);
/* Switch base in the current context */
if(ctxt->input->filename)
xmlFree((char *)ctxt->input->filename);
if(ctxt->directory)
xmlFree(ctxt->directory);
/* Remember those information */
ctxt->input->filename=(char *)xmlStrdup((xmlChar *)base_pi->file);
ctxt->directory=(char *)xmlStrdup((xmlChar *)base_pi->directory);
/* tie up xmlNode and style_pi */
XML_LINK_PI(document, pi, base_pi);
break;
case 'p':
if(!xmlStrEqual(target+8, (xmlChar *)"modxslt-param"+8)) {
mxslt_error(document, "warning - unknown modxslt PI: %s in %s\n", target, document->localfile);
break;
}
status=mxslt_get_static_attr((char *)data, mxslt_param, mxslt_param_values, mxslt_sizeof_array(mxslt_param));
if(status != MXSLT_OK) {
if(mxslt_param_values[0])
xfree(mxslt_param_values[0]);
if(mxslt_param_values[1])
xfree(mxslt_param_values[1]);
goto error;
}
if(mxslt_param_values[t_name]) {
/* Allocate and create pi */
param_pi=(mxslt_pi_param_t *)xmalloc(sizeof(mxslt_pi_param_t));
param_pi->type=MXSLT_PI_PARAM;
param_pi->param=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)mxslt_param_values[t_name],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
param_pi->value=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)mxslt_param_values[t_select],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
/* Link pi to list of pis */
XML_LINK_PI(document, pi, param_pi);
} else {
mxslt_error(document, "warning - modxslt-param specified without ``name'' attribute in `%s'\n", document->localfile);
}
/* Free memory up */
if(mxslt_param_values[t_select])
xfree(mxslt_param_values[t_select]);
if(mxslt_param_values[t_name])
xfree(mxslt_param_values[t_name]);
break;
case 's':
if(!xmlStrEqual(target+8, (xmlChar *)"modxslt-stylesheet"+8)) {
mxslt_error(document, "warning - unknown modxslt PI: %s in %s\n", target, document->localfile);
break;
}
status=mxslt_get_static_attr((char *)data, xml_stylesheet, xml_stylesheet_values, mxslt_sizeof_array(xml_stylesheet));
if(status != MXSLT_OK) {
if(xml_stylesheet_values[0])
xfree(xml_stylesheet_values[0]);
if(xml_stylesheet_values[1])
xfree(xml_stylesheet_values[1]);
if(xml_stylesheet_values[2])
xfree(xml_stylesheet_values[2]);
goto error;
}
if(xml_stylesheet_values[t_type]) {
/* Decode type */
type=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_type],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
/* Skip blanks before type */
for(tmp=type; MXSLT_IS_BLANK(*tmp); tmp++)
;
/* compare type string */
if(!strncmp(tmp, "text/xml", mxslt_sizeof_str("text/xml")) ||
!strncmp(tmp, "text/xsl", mxslt_sizeof_str("text/xsl"))) {
/* trailing spaces are allowed */
for(tmp+=mxslt_sizeof_str("text/x?l"); MXSLT_IS_BLANK(*tmp); tmp++)
;
/* Check there's nothing else beside them */
if(*tmp != '\0') {
mxslt_error(document, "warning - <modxslt-stylesheet type=\"... trailing junk: '%s'!\n", type);
xmlFree(type);
break;
}
/* Type is ok, add parameter to processing
* instruction */
style_pi=(mxslt_pi_style_t *)xmalloc(sizeof(mxslt_pi_style_t));
style_pi->type=MXSLT_PI_MODST;
style_pi->ctype=type;
if(xml_stylesheet_values[t_href])
style_pi->href=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_href],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
else
style_pi->href=NULL;
if(xml_stylesheet_values[t_media])
style_pi->media=(char *)xmlStringDecodeEntities(ctxt, (xmlChar *)xml_stylesheet_values[t_media],
XML_SUBSTITUTE_REF|XML_SUBSTITUTE_PEREF, 0, 0, 0);
else
style_pi->media=NULL;
/* Link to nodes chain */
XML_LINK_PI(document, pi, style_pi);
} else {
mxslt_error(document, "warning - <modxslt-stylesheet type=\"... unknown: '%s'!\n", type);
xmlFree(type);
}
}
if(xml_stylesheet_values[t_media])
xfree(xml_stylesheet_values[t_media]);
if(xml_stylesheet_values[t_href])
xfree(xml_stylesheet_values[t_href]);
if(xml_stylesheet_values[t_type])
xfree(xml_stylesheet_values[t_type]);
break;
default:
mxslt_error(document, "warning - unknown modxslt PI: %s in %s\n", target, document->localfile);
break;
}
break;
}
/* Taken from `processingInstruction' */
switch(ctxt->inSubset) {
case 1:
xmlAddChild((xmlNodePtr)ctxt->myDoc->intSubset, pi);
return;
case 2:
xmlAddChild((xmlNodePtr)ctxt->myDoc->extSubset, pi);
return;
}
if(!ctxt->myDoc->children || !ctxt->node) {
xmlAddChild((xmlNodePtr)ctxt->myDoc, pi);
return;
}
if(ctxt->node->type == XML_ELEMENT_NODE)
xmlAddChild(ctxt->node, pi);
else
xmlAddSibling(ctxt->node, pi);
return;
error:
mxslt_error(document, "warning - weird data while processing PI: %s in %s\n", target, document->localfile);
xmlFreeNode(pi);
return;
}
