void MzMLSpectrumDecoder::domParseString(const std::string& in, std::vector<BinaryData>& data_)
{
//-------------------------------------------------------------
// Create parser from input string using MemBufInputSource
//-------------------------------------------------------------
xercesc::MemBufInputSource myxml_buf(reinterpret_cast<const unsigned char*>(in.c_str()), in.length(), "myxml (in memory)");
xercesc::XercesDOMParser* parser = new xercesc::XercesDOMParser();
parser->setDoNamespaces(false);
parser->setDoSchema(false);
parser->setLoadExternalDTD(false);
parser->parse(myxml_buf);
//-------------------------------------------------------------
// Start parsing
// see http://www.yolinux.com/TUTORIALS/XML-Xerces-C.html
//-------------------------------------------------------------
// no need to free this pointer - owned by the parent parser object
xercesc::DOMDocument* doc = parser->getDocument();
// Get the top-level element (for example <spectrum> or <chromatogram>)
xercesc::DOMElement* elementRoot = doc->getDocumentElement();
if (!elementRoot)
{
delete parser;
}
// Extract the binaryDataArray tag (there may be multiple)
XMLCh* tag = xercesc::XMLString::transcode("binaryDataArray");
xercesc::DOMNodeList* li = elementRoot->getElementsByTagName(tag);
xercesc::XMLString::release(&tag);
/// Do the processing of the DOM
for (Size i = 0; i < li->getLength(); i++)
{
handleBinaryDataArray(li->item(i), data_);
}
delete parser;
}
int IndexedMzMLDecoder::domParseIndexedEnd_(std::string in, OffsetVector& spectra_offsets, OffsetVector& chromatograms_offsets)
{
/*
We parse something like
<indexedmzML>
<indexList count="1">
<index name="chromatogram">
<offset idRef="1">9752</offset>
</index>
</indexList>
<indexListOffset>26795</indexListOffset>
<fileChecksum>0</fileChecksum>
</indexedmzML>
*/
//-------------------------------------------------------------
// Create parser from input string using MemBufInputSource
//-------------------------------------------------------------
xercesc::MemBufInputSource myxml_buf(
reinterpret_cast<const unsigned char*>(in.c_str()), in.length(), "myxml (in memory)");
xercesc::XercesDOMParser parser;
parser.setDoNamespaces(false);
parser.setDoSchema(false);
parser.setLoadExternalDTD(false);
parser.parse(myxml_buf);
//-------------------------------------------------------------
// Start parsing
// see http://www.yolinux.com/TUTORIALS/XML-Xerces-C.html
//-------------------------------------------------------------
// no need to free this pointer - owned by the parent parser object
xercesc::DOMDocument* doc = parser.getDocument();
// Get the top-level element ("indexedmzML")
xercesc::DOMElement* elementRoot = doc->getDocumentElement();
if (!elementRoot)
{
std::cerr << "IndexedMzMLDecoder::domParseIndexedEnd Error: " <<
"No root element found:" << std::endl << std::endl << in << std::endl;
return -1;
}
// Extract the indexList tag (there should only be one)
XMLCh* x_tag = xercesc::XMLString::transcode("indexList");
xercesc::DOMNodeList* li = elementRoot->getElementsByTagName(x_tag);
xercesc::XMLString::release(&x_tag);
if (li->getLength() != 1)
{
std::cerr << "IndexedMzMLDecoder::domParseIndexedEnd Error: "
<< "no indexList element found:" << std::endl << std::endl << in << std::endl;
return -1;
}
xercesc::DOMNode* indexListNode = li->item(0);
XMLCh* x_idref_tag = xercesc::XMLString::transcode("idRef");
XMLCh* x_name_tag = xercesc::XMLString::transcode("name");
xercesc::DOMNodeList* index_elems = indexListNode->getChildNodes();
const XMLSize_t nodeCount_ = index_elems->getLength();
// Iterate through indexList elements (only two elements should be present
// which should be either spectrum or chromatogram offsets)
for (XMLSize_t j = 0; j < nodeCount_; ++j)
{
xercesc::DOMNode* currentNode = index_elems->item(j);
if (currentNode->getNodeType() && // true is not NULL
currentNode->getNodeType() == xercesc::DOMNode::ELEMENT_NODE) // is element
{
std::vector<std::pair<std::string, std::streampos> > result;
xercesc::DOMNode* firstChild = currentNode->getFirstChild();
xercesc::DOMNode* lastChild = currentNode->getLastChild();
xercesc::DOMNode* iter = firstChild;
// Iterate through children
// NOTE: Using xercesc::DOMNodeList and "item" is a very bad idea since
// each "item" call has complexity of O(n), see the
// implementation in DOMNodeListImpl.cpp :
// https://svn.apache.org/repos/asf/xerces/c/trunk/src/xercesc/dom/impl/DOMNodeListImpl.cpp
//
while (iter != lastChild)
{
iter = iter->getNextSibling();
xercesc::DOMNode* currentONode = iter;
if (currentONode->getNodeType() && // true is not NULL
currentONode->getNodeType() == xercesc::DOMNode::ELEMENT_NODE) // is element
{
xercesc::DOMElement* currentElement = dynamic_cast<xercesc::DOMElement*>(currentONode);
char* x_name = xercesc::XMLString::transcode(currentElement->getAttribute(x_idref_tag));
char* x_offset = xercesc::XMLString::transcode(currentONode->getTextContent());
std::streampos thisOffset = OpenMS::IndexedMzMLUtils::stringToStreampos( String(x_offset) );
result.push_back(std::make_pair(String(x_name), thisOffset));
xercesc::XMLString::release(&x_name);
xercesc::XMLString::release(&x_offset);
}
}
// should be either spectrum or chromatogram ...
xercesc::DOMElement* currentElement = dynamic_cast<xercesc::DOMElement*>(currentNode);
char* x_indexName = xercesc::XMLString::transcode(currentElement->getAttribute(x_name_tag));
std::string name(x_indexName);
xercesc::XMLString::release(&x_indexName);
if (name == "spectrum")
{
spectra_offsets = result;
}
else if (name == "chromatogram")
{
chromatograms_offsets = result;
}
else
{
std::cerr << "IndexedMzMLDecoder::domParseIndexedEnd Error: expected only " <<
"'spectrum' or 'chromatogram' below indexList but found instead '" <<
name << "'." << std::endl;
xercesc::XMLString::release(&x_idref_tag);
xercesc::XMLString::release(&x_name_tag);
return -1;
}
}
}
xercesc::XMLString::release(&x_idref_tag);
xercesc::XMLString::release(&x_name_tag);
return 0;
}