intrusive_ptr<DocumentSource> DocumentSourceGroup::createFromBson(
BSONElement elem, const intrusive_ptr<ExpressionContext>& pExpCtx) {
uassert(15947, "a group's fields must be specified in an object", elem.type() == Object);
intrusive_ptr<DocumentSourceGroup> pGroup(DocumentSourceGroup::create(pExpCtx));
BSONObj groupObj(elem.Obj());
BSONObjIterator groupIterator(groupObj);
VariablesIdGenerator idGenerator;
VariablesParseState vps(&idGenerator);
while (groupIterator.more()) {
BSONElement groupField(groupIterator.next());
const char* pFieldName = groupField.fieldName();
if (str::equals(pFieldName, "_id")) {
uassert(
15948, "a group's _id may only be specified once", pGroup->_idExpressions.empty());
pGroup->parseIdExpression(groupField, vps);
invariant(!pGroup->_idExpressions.empty());
} else if (str::equals(pFieldName, "$doingMerge")) {
massert(17030, "$doingMerge should be true if present", groupField.Bool());
pGroup->setDoingMerge(true);
} else {
/*
Treat as a projection field with the additional ability to
add aggregation operators.
*/
auto parsedAccumulator = Accumulator::parseAccumulator(groupField, vps);
auto fieldName = parsedAccumulator.first.toString();
auto accExpression = parsedAccumulator.second;
auto factory =
Accumulator::getFactory(groupField.embeddedObject().firstElementFieldName());
pGroup->addAccumulator(fieldName, factory, accExpression);
}
}
uassert(15955, "a group specification must include an _id", !pGroup->_idExpressions.empty());
pGroup->_variables.reset(new Variables(idGenerator.getIdCount()));
return pGroup;
}
intrusive_ptr<DocumentSource> DocumentSourceGroup::createFromBson(
BSONElement *pBsonElement,
const intrusive_ptr<ExpressionContext> &pCtx) {
uassert(15947, "a group's fields must be specified in an object",
pBsonElement->type() == Object);
intrusive_ptr<DocumentSourceGroup> pGroup(
DocumentSourceGroup::create(pCtx));
bool idSet = false;
BSONObj groupObj(pBsonElement->Obj());
BSONObjIterator groupIterator(groupObj);
while(groupIterator.more()) {
BSONElement groupField(groupIterator.next());
const char *pFieldName = groupField.fieldName();
if (strcmp(pFieldName, Document::idName.c_str()) == 0) {
uassert(15948, "a group's _id may only be specified once",
!idSet);
BSONType groupType = groupField.type();
if (groupType == Object) {
/*
Use the projection-like set of field paths to create the
group-by key.
*/
Expression::ObjectCtx oCtx(
Expression::ObjectCtx::DOCUMENT_OK);
intrusive_ptr<Expression> pId(
Expression::parseObject(&groupField, &oCtx));
pGroup->setIdExpression(pId);
idSet = true;
}
else if (groupType == String) {
string groupString(groupField.String());
const char *pGroupString = groupString.c_str();
if ((groupString.length() == 0) ||
(pGroupString[0] != '$'))
goto StringConstantId;
string pathString(
Expression::removeFieldPrefix(groupString));
intrusive_ptr<ExpressionFieldPath> pFieldPath(
ExpressionFieldPath::create(pathString));
pGroup->setIdExpression(pFieldPath);
idSet = true;
}
else {
/* pick out the constant types that are allowed */
switch(groupType) {
case NumberDouble:
case String:
case Object:
case Array:
case jstOID:
case Bool:
case Date:
case NumberInt:
case Timestamp:
case NumberLong:
case jstNULL:
StringConstantId: // from string case above
{
intrusive_ptr<const Value> pValue(
Value::createFromBsonElement(&groupField));
intrusive_ptr<ExpressionConstant> pConstant(
ExpressionConstant::create(pValue));
pGroup->setIdExpression(pConstant);
idSet = true;
break;
}
default:
uassert(15949, str::stream() <<
"a group's _id may not include fields of BSON type " << groupType,
false);
}
}
}
else {
/*
Treat as a projection field with the additional ability to
add aggregation operators.
*/
uassert(15950, str::stream() <<
"the group aggregate field name " <<
*pFieldName << " cannot be an operator name",
*pFieldName != '$');
uassert(15951, str::stream() <<
"the group aggregate field " << *pFieldName <<
"must be defined as an expression inside an object",
groupField.type() == Object);
BSONObj subField(groupField.Obj());
BSONObjIterator subIterator(subField);
size_t subCount = 0;
for(; subIterator.more(); ++subCount) {
BSONElement subElement(subIterator.next());
/* look for the specified operator */
GroupOpDesc key;
key.pName = subElement.fieldName();
const GroupOpDesc *pOp =
(const GroupOpDesc *)bsearch(
&key, GroupOpTable, NGroupOp, sizeof(GroupOpDesc),
GroupOpDescCmp);
uassert(15952, str::stream() <<
"unknown group operator \"" <<
key.pName << "\"",
pOp);
intrusive_ptr<Expression> pGroupExpr;
BSONType elementType = subElement.type();
if (elementType == Object) {
Expression::ObjectCtx oCtx(
Expression::ObjectCtx::DOCUMENT_OK);
pGroupExpr = Expression::parseObject(
&subElement, &oCtx);
}
else if (elementType == Array) {
uassert(15953, str::stream() <<
"aggregating group operators are unary (" <<
key.pName << ")", false);
}
else { /* assume its an atomic single operand */
pGroupExpr = Expression::parseOperand(&subElement);
}
pGroup->addAccumulator(
pFieldName, pOp->pFactory, pGroupExpr);
}
uassert(15954, str::stream() <<
"the computed aggregate \"" <<
pFieldName << "\" must specify exactly one operator",
subCount == 1);
}
}
uassert(15955, "a group specification must include an _id", idSet);
return pGroup;
}
intrusive_ptr<DocumentSource> DocumentSourceGroup::createFromBson(
BSONElement elem,
const intrusive_ptr<ExpressionContext> &pExpCtx) {
uassert(15947, "a group's fields must be specified in an object",
elem.type() == Object);
intrusive_ptr<DocumentSourceGroup> pGroup(
DocumentSourceGroup::create(pExpCtx));
BSONObj groupObj(elem.Obj());
BSONObjIterator groupIterator(groupObj);
VariablesIdGenerator idGenerator;
VariablesParseState vps(&idGenerator);
while(groupIterator.more()) {
BSONElement groupField(groupIterator.next());
const char *pFieldName = groupField.fieldName();
if (str::equals(pFieldName, "_id")) {
uassert(15948, "a group's _id may only be specified once",
pGroup->_idExpressions.empty());
pGroup->parseIdExpression(groupField, vps);
invariant(!pGroup->_idExpressions.empty());
}
else if (str::equals(pFieldName, "$doingMerge")) {
massert(17030, "$doingMerge should be true if present",
groupField.Bool());
pGroup->setDoingMerge(true);
}
else {
/*
Treat as a projection field with the additional ability to
add aggregation operators.
*/
uassert(16414, str::stream() <<
"the group aggregate field name '" << pFieldName <<
"' cannot be used because $group's field names cannot contain '.'",
!str::contains(pFieldName, '.') );
uassert(15950, str::stream() <<
"the group aggregate field name '" <<
pFieldName << "' cannot be an operator name",
pFieldName[0] != '$');
uassert(15951, str::stream() <<
"the group aggregate field '" << pFieldName <<
"' must be defined as an expression inside an object",
groupField.type() == Object);
BSONObj subField(groupField.Obj());
BSONObjIterator subIterator(subField);
size_t subCount = 0;
for(; subIterator.more(); ++subCount) {
BSONElement subElement(subIterator.next());
/* look for the specified operator */
GroupOpDesc key;
key.name = subElement.fieldName();
const GroupOpDesc *pOp =
(const GroupOpDesc *)bsearch(
&key, GroupOpTable, NGroupOp, sizeof(GroupOpDesc),
GroupOpDescCmp);
uassert(15952, str::stream() << "unknown group operator '" << key.name << "'",
pOp);
intrusive_ptr<Expression> pGroupExpr;
BSONType elementType = subElement.type();
if (elementType == Object) {
Expression::ObjectCtx oCtx(Expression::ObjectCtx::DOCUMENT_OK);
pGroupExpr = Expression::parseObject(subElement.Obj(), &oCtx, vps);
}
else if (elementType == Array) {
uasserted(15953, str::stream()
<< "aggregating group operators are unary (" << key.name << ")");
}
else { /* assume its an atomic single operand */
pGroupExpr = Expression::parseOperand(subElement, vps);
}
pGroup->addAccumulator(pFieldName, pOp->factory, pGroupExpr);
}
uassert(15954, str::stream() <<
"the computed aggregate '" <<
pFieldName << "' must specify exactly one operator",
subCount == 1);
}
}
uassert(15955, "a group specification must include an _id",
!pGroup->_idExpressions.empty());
pGroup->_variables.reset(new Variables(idGenerator.getIdCount()));
return pGroup;
}
//Get max and min chage but also check what type of a,b,c,x,y,z score/ions are present
void TandemReader::getMaxMinCharge(const std::string &fn, bool isDecoy){
int nTot = 0, charge = 0;
namespace xml = xsd::cxx::xml;
ifstream ifs;
ifs.exceptions(ifstream::badbit|ifstream::failbit);
ifs.open(fn.c_str());
if (!ifs)
{
ostringstream temp;
temp << "Error : can not open file " << fn << std::endl;
throw MyException(temp.str());
}
parser p;
try
{
xml_schema::dom::auto_ptr< xercesc::DOMDocument>
doc (p.start (ifs, fn.c_str(),true, schemaDefinition,schema_major, schema_minor, scheme_namespace,true));
assert(doc.get());
for (doc = p.next(); doc.get() != 0; doc = p.next ())
{
//Check that the tag name is group and that its not the inputput parameters
if(XMLString::equals(groupStr,doc->getDocumentElement()->getTagName())
&& XMLString::equals(groupModelStr,doc->getDocumentElement()->getAttribute(groupTypeStr)))
{
tandem_ns::group groupObj(*doc->getDocumentElement()); //Parse to the codesynthesis object model
if(groupObj.z().present()) //We are sure we are not in parameters group so z(the charge) has to be present.
{
stringstream chargeStream (stringstream::in | stringstream::out);
chargeStream << groupObj.z();
chargeStream >> charge;
if (minCharge > charge) minCharge = charge;
if (maxCharge < charge) maxCharge = charge;
nTot++;
}
else
{
ostringstream temp;
temp << "Missing charge(attribute z in group element) for one or more groups in: " << fn << endl;
throw MyException(temp.str());
}
if(firstPSM)
{
//Check what type of scores/ions are present
BOOST_FOREACH (const tandem_ns::protein &protObj, groupObj.protein()) { //Protein
tandem_ns::protein::peptide_type peptideObj=protObj.peptide(); //Peptide
BOOST_FOREACH(const tandem_ns::domain &domainObj, peptideObj.domain()) { //Domain
//x,y,z
if (domainObj.x_score().present() && domainObj.x_ions().present()) {
x_score=true;
}
if (domainObj.y_score().present() && domainObj.y_ions().present()) {
y_score=true;
}
if (domainObj.z_score().present() && domainObj.z_ions().present()) {
z_score=true;
}
//a,b,c
if (domainObj.a_score().present() && domainObj.a_ions().present()) {
a_score=true;
}
if (domainObj.b_score().present() && domainObj.b_ions().present()) {
b_score=true;
}
if (domainObj.c_score().present() && domainObj.c_ions().present()) {
c_score=true;
}
} //End of for domain
} //End of for prot
firstPSM=false;
}
}
intrusive_ptr<DocumentSource> DocumentSourceGroup::createFromBson(
BSONElement *pBsonElement,
const intrusive_ptr<ExpressionContext> &pCtx) {
assert(pBsonElement->type() == Object); // CW TODO must be an object
intrusive_ptr<DocumentSourceGroup> pGroup(
DocumentSourceGroup::create(pCtx));
bool idSet = false;
BSONObj groupObj(pBsonElement->Obj());
BSONObjIterator groupIterator(groupObj);
while(groupIterator.more()) {
BSONElement groupField(groupIterator.next());
const char *pFieldName = groupField.fieldName();
if (strcmp(pFieldName, Document::idName.c_str()) == 0) {
assert(!idSet); // CW TODO _id specified multiple times
BSONType groupType = groupField.type();
if (groupType == Object) {
/*
Use the projection-like set of field paths to create the
group-by key.
*/
Expression::ObjectCtx oCtx(
Expression::ObjectCtx::DOCUMENT_OK);
intrusive_ptr<Expression> pId(
Expression::parseObject(&groupField, &oCtx));
pGroup->setIdExpression(pId);
idSet = true;
}
else if (groupType == String) {
string groupString(groupField.String());
const char *pGroupString = groupString.c_str();
if ((groupString.length() == 0) ||
(pGroupString[0] != '$'))
goto StringConstantId;
string pathString(
Expression::removeFieldPrefix(groupString));
intrusive_ptr<ExpressionFieldPath> pFieldPath(
ExpressionFieldPath::create(pathString));
pGroup->setIdExpression(pFieldPath);
idSet = true;
}
else {
/* pick out the constant types that are allowed */
switch(groupType) {
case NumberDouble:
case String:
case Object:
case Array:
case jstOID:
case Bool:
case Date:
case NumberInt:
case Timestamp:
case NumberLong:
case jstNULL:
StringConstantId: // from string case above
{
intrusive_ptr<const Value> pValue(
Value::createFromBsonElement(&groupField));
intrusive_ptr<ExpressionConstant> pConstant(
ExpressionConstant::create(pValue));
pGroup->setIdExpression(pConstant);
idSet = true;
break;
}
default:
assert(false);
// CW TODO disallowed constant group key
}
}
}
else {
/*
Treat as a projection field with the additional ability to
add aggregation operators.
*/
assert(*pFieldName != '$');
// CW TODO error: field name can't be an operator
assert(groupField.type() == Object);
// CW TODO error: must be an operator expression
BSONObj subField(groupField.Obj());
BSONObjIterator subIterator(subField);
size_t subCount = 0;
for(; subIterator.more(); ++subCount) {
BSONElement subElement(subIterator.next());
/* look for the specified operator */
GroupOpDesc key;
key.pName = subElement.fieldName();
const GroupOpDesc *pOp =
(const GroupOpDesc *)bsearch(
&key, GroupOpTable, NGroupOp, sizeof(GroupOpDesc),
GroupOpDescCmp);
assert(pOp); // CW TODO error: operator not found
intrusive_ptr<Expression> pGroupExpr;
BSONType elementType = subElement.type();
if (elementType == Object) {
Expression::ObjectCtx oCtx(
Expression::ObjectCtx::DOCUMENT_OK);
pGroupExpr = Expression::parseObject(
&subElement, &oCtx);
}
else if (elementType == Array) {
assert(false); // CW TODO group operators are unary
}
else { /* assume its an atomic single operand */
pGroupExpr = Expression::parseOperand(&subElement);
}
pGroup->addAccumulator(
pFieldName, pOp->pFactory, pGroupExpr);
}
assert(subCount == 1);
// CW TODO error: only one operator allowed
}
}
assert(idSet); // CW TODO error: missing _id specification
return pGroup;
}
intrusive_ptr<DocumentSource> DocumentSourceGroup::createFromBson(
BSONElement *pBsonElement,
const intrusive_ptr<ExpressionContext> &pExpCtx) {
uassert(15947, "a group's fields must be specified in an object",
pBsonElement->type() == Object);
intrusive_ptr<DocumentSourceGroup> pGroup(
DocumentSourceGroup::create(pExpCtx));
bool idSet = false;
BSONObj groupObj(pBsonElement->Obj());
BSONObjIterator groupIterator(groupObj);
while(groupIterator.more()) {
BSONElement groupField(groupIterator.next());
const char *pFieldName = groupField.fieldName();
if (str::equals(pFieldName, "_id")) {
uassert(15948, "a group's _id may only be specified once",
!idSet);
BSONType groupType = groupField.type();
if (groupType == Object) {
/*
Use the projection-like set of field paths to create the
group-by key.
*/
Expression::ObjectCtx oCtx(Expression::ObjectCtx::DOCUMENT_OK);
intrusive_ptr<Expression> pId(
Expression::parseObject(&groupField, &oCtx));
pGroup->setIdExpression(pId);
idSet = true;
}
else if (groupType == String) {
const string groupString = groupField.str();
if (!groupString.empty() && groupString[0] == '$') {
pGroup->setIdExpression(ExpressionFieldPath::parse(groupString));
idSet = true;
}
}
if (!idSet) {
// constant id - single group
pGroup->setIdExpression(ExpressionConstant::create(Value(groupField)));
idSet = true;
}
}
else {
/*
Treat as a projection field with the additional ability to
add aggregation operators.
*/
uassert(16414, str::stream() <<
"the group aggregate field name '" << pFieldName <<
"' cannot be used because $group's field names cannot contain '.'",
!str::contains(pFieldName, '.') );
uassert(15950, str::stream() <<
"the group aggregate field name '" <<
pFieldName << "' cannot be an operator name",
pFieldName[0] != '$');
uassert(15951, str::stream() <<
"the group aggregate field '" << pFieldName <<
"' must be defined as an expression inside an object",
groupField.type() == Object);
BSONObj subField(groupField.Obj());
BSONObjIterator subIterator(subField);
size_t subCount = 0;
for(; subIterator.more(); ++subCount) {
BSONElement subElement(subIterator.next());
/* look for the specified operator */
GroupOpDesc key;
key.name = subElement.fieldName();
const GroupOpDesc *pOp =
(const GroupOpDesc *)bsearch(
&key, GroupOpTable, NGroupOp, sizeof(GroupOpDesc),
GroupOpDescCmp);
uassert(15952, str::stream() << "unknown group operator '" << key.name << "'",
pOp);
intrusive_ptr<Expression> pGroupExpr;
BSONType elementType = subElement.type();
if (elementType == Object) {
Expression::ObjectCtx oCtx(
Expression::ObjectCtx::DOCUMENT_OK);
pGroupExpr = Expression::parseObject(
&subElement, &oCtx);
}
else if (elementType == Array) {
uasserted(15953, str::stream()
<< "aggregating group operators are unary (" << key.name << ")");
}
else { /* assume its an atomic single operand */
pGroupExpr = Expression::parseOperand(&subElement);
}
pGroup->addAccumulator(pFieldName, pOp->factory, pGroupExpr);
}
uassert(15954, str::stream() <<
"the computed aggregate '" <<
pFieldName << "' must specify exactly one operator",
subCount == 1);
}
}
uassert(15955, "a group specification must include an _id", idSet);
return pGroup;
}